source: src/molecules.cpp@ 683914

Action_Thermostats Add_AtomRandomPerturbation Add_FitFragmentPartialChargesAction Add_RotateAroundBondAction Add_SelectAtomByNameAction Added_ParseSaveFragmentResults AddingActions_SaveParseParticleParameters Adding_Graph_to_ChangeBondActions Adding_MD_integration_tests Adding_ParticleName_to_Atom Adding_StructOpt_integration_tests AtomFragments Automaking_mpqc_open AutomationFragmentation_failures Candidate_v1.5.4 Candidate_v1.6.0 Candidate_v1.6.1 Candidate_v1.7.0 ChangeBugEmailaddress ChangingTestPorts ChemicalSpaceEvaluator CombiningParticlePotentialParsing Combining_Subpackages Debian_Package_split Debian_package_split_molecuildergui_only Disabling_MemDebug Docu_Python_wait EmpiricalPotential_contain_HomologyGraph EmpiricalPotential_contain_HomologyGraph_documentation Enable_parallel_make_install Enhance_userguide Enhanced_StructuralOptimization Enhanced_StructuralOptimization_continued Example_ManyWaysToTranslateAtom Exclude_Hydrogens_annealWithBondGraph FitPartialCharges_GlobalError Fix_BoundInBox_CenterInBox_MoleculeActions Fix_ChargeSampling_PBC Fix_ChronosMutex Fix_FitPartialCharges Fix_FitPotential_needs_atomicnumbers Fix_ForceAnnealing Fix_IndependentFragmentGrids Fix_ParseParticles Fix_ParseParticles_split_forward_backward_Actions Fix_PopActions Fix_QtFragmentList_sorted_selection Fix_Restrictedkeyset_FragmentMolecule Fix_StatusMsg Fix_StepWorldTime_single_argument Fix_Verbose_Codepatterns Fix_fitting_potentials Fixes ForceAnnealing_goodresults ForceAnnealing_oldresults ForceAnnealing_tocheck ForceAnnealing_with_BondGraph ForceAnnealing_with_BondGraph_continued ForceAnnealing_with_BondGraph_continued_betteresults ForceAnnealing_with_BondGraph_contraction-expansion FragmentAction_writes_AtomFragments FragmentMolecule_checks_bonddegrees GeometryObjects Gui_Fixes Gui_displays_atomic_force_velocity ImplicitCharges IndependentFragmentGrids IndependentFragmentGrids_IndividualZeroInstances IndependentFragmentGrids_IntegrationTest IndependentFragmentGrids_Sole_NN_Calculation JobMarket_RobustOnKillsSegFaults JobMarket_StableWorkerPool JobMarket_unresolvable_hostname_fix MoreRobust_FragmentAutomation ODR_violation_mpqc_open PartialCharges_OrthogonalSummation PdbParser_setsAtomName PythonUI_with_named_parameters QtGui_reactivate_TimeChanged_changes Recreated_GuiChecks Rewrite_FitPartialCharges RotateToPrincipalAxisSystem_UndoRedo SaturateAtoms_findBestMatching SaturateAtoms_singleDegree StoppableMakroAction Subpackage_CodePatterns Subpackage_JobMarket Subpackage_LinearAlgebra Subpackage_levmar Subpackage_mpqc_open Subpackage_vmg Switchable_LogView ThirdParty_MPQC_rebuilt_buildsystem TrajectoryDependenant_MaxOrder TremoloParser_IncreasedPrecision TremoloParser_MultipleTimesteps TremoloParser_setsAtomName Ubuntu_1604_changes stable
Last change on this file since 683914 was 683914, checked in by Frederik Heber <heber@…>, 17 years ago

CyclicStructureAnalysis: BUGFIX - MinimumRingSize of non-loop atoms was wrong

  • new bool atom:IsCyclic states whether atoms is part of a cycle or not, this is set in CyclicStructureAnalysis()
  • MinimumRingSize was set to MinimumRingSize plus ShortestPath thereto. This gives correct results only if this goes to a loop-member. However the if-condition was phrased in such a manner, that it became possible that non-loop members (with MinimumRingSize set below AtomCount) would now also be regarded als loop members. This is sort of a triangle inequality problem :), path of A to B plus path from B to C is probably not the same as path from A to C in terms of length
  • Property mode set to 100644
File size: 191.2 KB
Line 
1/** \file molecules.cpp
2 *
3 * Functions for the class molecule.
4 *
5 */
6
7#include "molecules.hpp"
8
9/************************************* Other Functions *************************************/
10
11/** Determines sum of squared distances of \a X to all \a **vectors.
12 * \param *x reference vector
13 * \param *params
14 * \return sum of square distances
15 */
16double LSQ (const gsl_vector * x, void * params)
17{
18 double sum = 0.;
19 struct LSQ_params *par = (struct LSQ_params *)params;
20 vector **vectors = par->vectors;
21 int num = par->num;
22
23 for (int i=num;i--;) {
24 for(int j=NDIM;j--;)
25 sum += (gsl_vector_get(x,j) - (vectors[i])->x[j])*(gsl_vector_get(x,j) - (vectors[i])->x[j]);
26 }
27
28 return sum;
29};
30
31/************************************* Functions for class molecule *********************************/
32
33/** Constructor of class molecule.
34 * Initialises molecule list with correctly referenced start and end, and sets molecule::last_atom to zero.
35 */
36molecule::molecule(periodentafel *teil)
37{
38 // init atom chain list
39 start = new atom;
40 end = new atom;
41 start->father = NULL;
42 end->father = NULL;
43 link(start,end);
44 // init bond chain list
45 first = new bond(start, end, 1, -1);
46 last = new bond(start, end, 1, -1);
47 link(first,last);
48 // other stuff
49 last_atom = 0;
50 elemente = teil;
51 AtomCount = 0;
52 BondCount = 0;
53 NoNonBonds = 0;
54 NoNonHydrogen = 0;
55 NoCyclicBonds = 0;
56 ListOfBondsPerAtom = NULL;
57 NumberOfBondsPerAtom = NULL;
58 ElementCount = 0;
59 for(int i=MAX_ELEMENTS;i--;)
60 ElementsInMolecule[i] = 0;
61 cell_size[0] = cell_size[2] = cell_size[5]= 20.;
62 cell_size[1] = cell_size[3] = cell_size[4]= 0.;
63};
64
65/** Destructor of class molecule.
66 * Initialises molecule list with correctly referenced start and end, and sets molecule::last_atom to zero.
67 */
68molecule::~molecule()
69{
70 if (ListOfBondsPerAtom != NULL)
71 for(int i=AtomCount;i--;)
72 Free((void **)&ListOfBondsPerAtom[i], "molecule::~molecule: ListOfBondsPerAtom[i]");
73 Free((void **)&ListOfBondsPerAtom, "molecule::~molecule: ListOfBondsPerAtom");
74 Free((void **)&NumberOfBondsPerAtom, "molecule::~molecule: NumberOfBondsPerAtom");
75 CleanupMolecule();
76 delete(first);
77 delete(last);
78 delete(end);
79 delete(start);
80};
81
82/** Adds given atom \a *pointer from molecule list.
83 * Increases molecule::last_atom and gives last number to added atom and names it according to its element::abbrev and molecule::AtomCount
84 * \param *pointer allocated and set atom
85 * \return true - succeeded, false - atom not found in list
86 */
87bool molecule::AddAtom(atom *pointer)
88{
89 if (pointer != NULL) {
90 pointer->sort = &pointer->nr;
91 pointer->nr = last_atom++; // increase number within molecule
92 AtomCount++;
93 if (pointer->type != NULL) {
94 if (ElementsInMolecule[pointer->type->Z] == 0)
95 ElementCount++;
96 ElementsInMolecule[pointer->type->Z]++; // increase number of elements
97 if (pointer->type->Z != 1)
98 NoNonHydrogen++;
99 if (pointer->Name == NULL) {
100 Free((void **)&pointer->Name, "molecule::AddAtom: *pointer->Name");
101 pointer->Name = (char *) Malloc(sizeof(char)*6, "molecule::AddAtom: *pointer->Name");
102 sprintf(pointer->Name, "%2s%02d", pointer->type->symbol, pointer->nr+1);
103 }
104 }
105 return add(pointer, end);
106 } else
107 return false;
108};
109
110/** Adds a copy of the given atom \a *pointer from molecule list.
111 * Increases molecule::last_atom and gives last number to added atom.
112 * \param *pointer allocated and set atom
113 * \return true - succeeded, false - atom not found in list
114 */
115atom * molecule::AddCopyAtom(atom *pointer)
116{
117 if (pointer != NULL) {
118 atom *walker = new atom();
119 walker->type = pointer->type; // copy element of atom
120 walker->x.CopyVector(&pointer->x); // copy coordination
121 walker->v.CopyVector(&pointer->v); // copy velocity
122 walker->FixedIon = pointer->FixedIon;
123 walker->sort = &walker->nr;
124 walker->nr = last_atom++; // increase number within molecule
125 walker->father = pointer; //->GetTrueFather();
126 walker->Name = (char *) Malloc(sizeof(char)*strlen(pointer->Name)+1, "molecule::AddCopyAtom: *Name");
127 strcpy (walker->Name, pointer->Name);
128 add(walker, end);
129 if ((pointer->type != NULL) && (pointer->type->Z != 1))
130 NoNonHydrogen++;
131 AtomCount++;
132 return walker;
133 } else
134 return NULL;
135};
136
137/** Adds a Hydrogen atom in replacement for the given atom \a *partner in bond with a *origin.
138 * Here, we have to distinguish between single, double or triple bonds as stated by \a BondDegree, that each demand
139 * a different scheme when adding \a *replacement atom for the given one.
140 * -# Single Bond: Simply add new atom with bond distance rescaled to typical hydrogen one
141 * -# Double Bond: Here, we need the **BondList of the \a *origin atom, by scanning for the other bonds instead of
142 * *Bond, we use the through these connected atoms to determine the plane they lie in, vector::MakeNormalVector().
143 * The orthonormal vector to this plane along with the vector in *Bond direction determines the plane the two
144 * replacing hydrogens shall lie in. Now, all remains to do is take the usual hydrogen double bond angle for the
145 * element of *origin and form the sin/cos admixture of both plane vectors for the new coordinates of the two
146 * hydrogens forming this angle with *origin.
147 * -# Triple Bond: The idea is to set up a tetraoid (C1-H1-H2-H3) (however the lengths \f$b\f$ of the sides of the base
148 * triangle formed by the to be added hydrogens are not equal to the typical bond distance \f$l\f$ but have to be
149 * determined from the typical angle \f$\alpha\f$ for a hydrogen triple connected to the element of *origin):
150 * We have the height \f$d\f$ as the vector in *Bond direction (from triangle C1-H1-H2).
151 * \f[ h = l \cdot \cos{\left (\frac{\alpha}{2} \right )} \qquad b = 2l \cdot \sin{\left (\frac{\alpha}{2} \right)} \quad \rightarrow \quad d = l \cdot \sqrt{\cos^2{\left (\frac{\alpha}{2} \right)}-\frac{1}{3}\cdot\sin^2{\left (\frac{\alpha}{2}\right )}}
152 * \f]
153 * vector::GetNormalVector() creates one orthonormal vector from this *Bond vector and vector::MakeNormalVector creates
154 * the third one from the former two vectors. The latter ones form the plane of the base triangle mentioned above.
155 * The lengths for these are \f$f\f$ and \f$g\f$ (from triangle H1-H2-(center of H1-H2-H3)) with knowledge that
156 * the median lines in an isosceles triangle meet in the center point with a ratio 2:1.
157 * \f[ f = \frac{b}{\sqrt{3}} \qquad g = \frac{b}{2}
158 * \f]
159 * as the coordination of all three atoms in the coordinate system of these three vectors:
160 * \f$\pmatrix{d & f & 0}\f$, \f$\pmatrix{d & -0.5 \cdot f & g}\f$ and \f$\pmatrix{d & -0.5 \cdot f & -g}\f$.
161 *
162 * \param *out output stream for debugging
163 * \param *Bond pointer to bond between \a *origin and \a *replacement
164 * \param *TopOrigin son of \a *origin of upper level molecule (the atom added to this molecule as a copy of \a *origin)
165 * \param *origin pointer to atom which acts as the origin for scaling the added hydrogen to correct bond length
166 * \param *replacement pointer to the atom which shall be copied as a hydrogen atom in this molecule
167 * \param **BondList list of bonds \a *replacement has (necessary to determine plane for double and triple bonds)
168 * \param NumBond number of bonds in \a **BondList
169 * \param isAngstroem whether the coordination of the given atoms is in AtomicLength (false) or Angstrom(true)
170 * \return number of atoms added, if < bond::BondDegree then something went wrong
171 * \todo double and triple bonds splitting (always use the tetraeder angle!)
172 */
173bool molecule::AddHydrogenReplacementAtom(ofstream *out, bond *TopBond, atom *BottomOrigin, atom *TopOrigin, atom *TopReplacement, bond **BondList, int NumBond, bool IsAngstroem)
174{
175 double bondlength; // bond length of the bond to be replaced/cut
176 double bondangle; // bond angle of the bond to be replaced/cut
177 double BondRescale; // rescale value for the hydrogen bond length
178 bool AllWentWell = true; // flag gathering the boolean return value of molecule::AddAtom and other functions, as return value on exit
179 bond *FirstBond = NULL, *SecondBond = NULL; // Other bonds in double bond case to determine "other" plane
180 atom *FirstOtherAtom = NULL, *SecondOtherAtom = NULL, *ThirdOtherAtom = NULL; // pointer to hydrogen atoms to be added
181 double b,l,d,f,g, alpha, factors[NDIM]; // hold temporary values in triple bond case for coordination determination
182 vector OrthoVector1, OrthoVector2; // temporary vectors in coordination construction
183 vector InBondVector; // vector in direction of *Bond
184 bond *Binder = NULL;
185 double *matrix;
186
187// *out << Verbose(3) << "Begin of AddHydrogenReplacementAtom." << endl;
188 // create vector in direction of bond
189 InBondVector.CopyVector(&TopReplacement->x);
190 InBondVector.SubtractVector(&TopOrigin->x);
191 bondlength = InBondVector.Norm();
192
193 // is greater than typical bond distance? Then we have to correct periodically
194 // the problem is not the H being out of the box, but InBondVector have the wrong direction
195 // due to TopReplacement or Origin being on the wrong side!
196 if (bondlength > BondDistance) {
197// *out << Verbose(4) << "InBondVector is: ";
198// InBondVector.Output(out);
199// *out << endl;
200 OrthoVector1.Zero();
201 for (int i=NDIM;i--;) {
202 l = TopReplacement->x.x[i] - TopOrigin->x.x[i];
203 if (fabs(l) > BondDistance) { // is component greater than bond distance
204 OrthoVector1.x[i] = (l < 0) ? -1. : +1.;
205 } // (signs are correct, was tested!)
206 }
207 matrix = ReturnFullMatrixforSymmetric(cell_size);
208 OrthoVector1.MatrixMultiplication(matrix);
209 InBondVector.SubtractVector(&OrthoVector1); // subtract just the additional translation
210 Free((void **)&matrix, "molecule::AddHydrogenReplacementAtom: *matrix");
211 bondlength = InBondVector.Norm();
212// *out << Verbose(4) << "Corrected InBondVector is now: ";
213// InBondVector.Output(out);
214// *out << endl;
215 } // periodic correction finished
216
217 InBondVector.Normalize();
218 // get typical bond length and store as scale factor for later
219 BondRescale = TopOrigin->type->HBondDistance[TopBond->BondDegree-1];
220 if (BondRescale == -1) {
221 cerr << Verbose(3) << "WARNING: There is no typical bond distance for bond (" << TopOrigin->Name << "<->" << TopReplacement->Name << ") of degree " << TopBond->BondDegree << "!" << endl;
222 BondRescale = bondlength;
223 } else {
224 if (!IsAngstroem)
225 BondRescale /= (1.*AtomicLengthToAngstroem);
226 }
227
228 // discern single, double and triple bonds
229 switch(TopBond->BondDegree) {
230 case 1:
231 FirstOtherAtom = new atom(); // new atom
232 FirstOtherAtom->type = elemente->FindElement(1); // element is Hydrogen
233 FirstOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
234 FirstOtherAtom->FixedIon = TopReplacement->FixedIon;
235 if (TopReplacement->type->Z == 1) { // neither rescale nor replace if it's already hydrogen
236 FirstOtherAtom->father = TopReplacement;
237 BondRescale = bondlength;
238 } else {
239 FirstOtherAtom->father = NULL; // if we replace hydrogen, we mark it as our father, otherwise we are just an added hydrogen with no father
240 }
241 InBondVector.Scale(&BondRescale); // rescale the distance vector to Hydrogen bond length
242 FirstOtherAtom->x.CopyVector(&TopOrigin->x); // set coordination to origin ...
243 FirstOtherAtom->x.AddVector(&InBondVector); // ... and add distance vector to replacement atom
244 AllWentWell = AllWentWell && AddAtom(FirstOtherAtom);
245// *out << Verbose(4) << "Added " << *FirstOtherAtom << " at: ";
246// FirstOtherAtom->x.Output(out);
247// *out << endl;
248 Binder = AddBond(BottomOrigin, FirstOtherAtom, 1);
249 Binder->Cyclic = false;
250 Binder->Type = TreeEdge;
251 break;
252 case 2:
253 // determine two other bonds (warning if there are more than two other) plus valence sanity check
254 for (int i=0;i<NumBond;i++) {
255 if (BondList[i] != TopBond) {
256 if (FirstBond == NULL) {
257 FirstBond = BondList[i];
258 FirstOtherAtom = BondList[i]->GetOtherAtom(TopOrigin);
259 } else if (SecondBond == NULL) {
260 SecondBond = BondList[i];
261 SecondOtherAtom = BondList[i]->GetOtherAtom(TopOrigin);
262 } else {
263 *out << Verbose(3) << "WARNING: Detected more than four bonds for atom " << TopOrigin->Name;
264 }
265 }
266 }
267 if (SecondOtherAtom == NULL) { // then we have an atom with valence four, but only 3 bonds: one to replace and one which is TopBond (third is FirstBond)
268 SecondBond = TopBond;
269 SecondOtherAtom = TopReplacement;
270 }
271 if (FirstOtherAtom != NULL) { // then we just have this double bond and the plane does not matter at all
272// *out << Verbose(3) << "Regarding the double bond (" << TopOrigin->Name << "<->" << TopReplacement->Name << ") to be constructed: Taking " << FirstOtherAtom->Name << " and " << SecondOtherAtom->Name << " along with " << TopOrigin->Name << " to determine orthogonal plane." << endl;
273
274 // determine the plane of these two with the *origin
275 AllWentWell = AllWentWell && OrthoVector1.MakeNormalVector(&TopOrigin->x, &FirstOtherAtom->x, &SecondOtherAtom->x);
276 } else {
277 OrthoVector1.GetOneNormalVector(&InBondVector);
278 }
279 //*out << Verbose(3)<< "Orthovector1: ";
280 //OrthoVector1.Output(out);
281 //*out << endl;
282 // orthogonal vector and bond vector between origin and replacement form the new plane
283 OrthoVector1.MakeNormalVector(&InBondVector);
284 OrthoVector1.Normalize();
285 //*out << Verbose(3) << "ReScaleCheck: " << OrthoVector1.Norm() << " and " << InBondVector.Norm() << "." << endl;
286
287 // create the two Hydrogens ...
288 FirstOtherAtom = new atom();
289 SecondOtherAtom = new atom();
290 FirstOtherAtom->type = elemente->FindElement(1);
291 SecondOtherAtom->type = elemente->FindElement(1);
292 FirstOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
293 FirstOtherAtom->FixedIon = TopReplacement->FixedIon;
294 SecondOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
295 SecondOtherAtom->FixedIon = TopReplacement->FixedIon;
296 FirstOtherAtom->father = NULL; // we are just an added hydrogen with no father
297 SecondOtherAtom->father = NULL; // we are just an added hydrogen with no father
298 bondangle = TopOrigin->type->HBondAngle[1];
299 if (bondangle == -1) {
300 *out << Verbose(3) << "WARNING: There is no typical bond angle for bond (" << TopOrigin->Name << "<->" << TopReplacement->Name << ") of degree " << TopBond->BondDegree << "!" << endl;
301 bondangle = 0;
302 }
303 bondangle *= M_PI/180./2.;
304// *out << Verbose(3) << "ReScaleCheck: InBondVector ";
305// InBondVector.Output(out);
306// *out << endl;
307// *out << Verbose(3) << "ReScaleCheck: Orthovector ";
308// OrthoVector1.Output(out);
309// *out << endl;
310// *out << Verbose(3) << "Half the bond angle is " << bondangle << ", sin and cos of it: " << sin(bondangle) << ", " << cos(bondangle) << endl;
311 FirstOtherAtom->x.Zero();
312 SecondOtherAtom->x.Zero();
313 for(int i=NDIM;i--;) { // rotate by half the bond angle in both directions (InBondVector is bondangle = 0 direction)
314 FirstOtherAtom->x.x[i] = InBondVector.x[i] * cos(bondangle) + OrthoVector1.x[i] * (sin(bondangle));
315 SecondOtherAtom->x.x[i] = InBondVector.x[i] * cos(bondangle) + OrthoVector1.x[i] * (-sin(bondangle));
316 }
317 FirstOtherAtom->x.Scale(&BondRescale); // rescale by correct BondDistance
318 SecondOtherAtom->x.Scale(&BondRescale);
319 //*out << Verbose(3) << "ReScaleCheck: " << FirstOtherAtom->x.Norm() << " and " << SecondOtherAtom->x.Norm() << "." << endl;
320 for(int i=NDIM;i--;) { // and make relative to origin atom
321 FirstOtherAtom->x.x[i] += TopOrigin->x.x[i];
322 SecondOtherAtom->x.x[i] += TopOrigin->x.x[i];
323 }
324 // ... and add to molecule
325 AllWentWell = AllWentWell && AddAtom(FirstOtherAtom);
326 AllWentWell = AllWentWell && AddAtom(SecondOtherAtom);
327// *out << Verbose(4) << "Added " << *FirstOtherAtom << " at: ";
328// FirstOtherAtom->x.Output(out);
329// *out << endl;
330// *out << Verbose(4) << "Added " << *SecondOtherAtom << " at: ";
331// SecondOtherAtom->x.Output(out);
332// *out << endl;
333 Binder = AddBond(BottomOrigin, FirstOtherAtom, 1);
334 Binder->Cyclic = false;
335 Binder->Type = TreeEdge;
336 Binder = AddBond(BottomOrigin, SecondOtherAtom, 1);
337 Binder->Cyclic = false;
338 Binder->Type = TreeEdge;
339 break;
340 case 3:
341 // take the "usual" tetraoidal angle and add the three Hydrogen in direction of the bond (height of the tetraoid)
342 FirstOtherAtom = new atom();
343 SecondOtherAtom = new atom();
344 ThirdOtherAtom = new atom();
345 FirstOtherAtom->type = elemente->FindElement(1);
346 SecondOtherAtom->type = elemente->FindElement(1);
347 ThirdOtherAtom->type = elemente->FindElement(1);
348 FirstOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
349 FirstOtherAtom->FixedIon = TopReplacement->FixedIon;
350 SecondOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
351 SecondOtherAtom->FixedIon = TopReplacement->FixedIon;
352 ThirdOtherAtom->v.CopyVector(&TopReplacement->v); // copy velocity
353 ThirdOtherAtom->FixedIon = TopReplacement->FixedIon;
354 FirstOtherAtom->father = NULL; // we are just an added hydrogen with no father
355 SecondOtherAtom->father = NULL; // we are just an added hydrogen with no father
356 ThirdOtherAtom->father = NULL; // we are just an added hydrogen with no father
357
358 // we need to vectors orthonormal the InBondVector
359 AllWentWell = AllWentWell && OrthoVector1.GetOneNormalVector(&InBondVector);
360// *out << Verbose(3) << "Orthovector1: ";
361// OrthoVector1.Output(out);
362// *out << endl;
363 AllWentWell = AllWentWell && OrthoVector2.MakeNormalVector(&InBondVector, &OrthoVector1);
364// *out << Verbose(3) << "Orthovector2: ";
365// OrthoVector2.Output(out);
366// *out << endl;
367
368 // create correct coordination for the three atoms
369 alpha = (TopOrigin->type->HBondAngle[2])/180.*M_PI/2.; // retrieve triple bond angle from database
370 l = BondRescale; // desired bond length
371 b = 2.*l*sin(alpha); // base length of isosceles triangle
372 d = l*sqrt(cos(alpha)*cos(alpha) - sin(alpha)*sin(alpha)/3.); // length for InBondvector
373 f = b/sqrt(3.); // length for OrthVector1
374 g = b/2.; // length for OrthVector2
375// *out << Verbose(3) << "Bond length and half-angle: " << l << ", " << alpha << "\t (b,d,f,g) = " << b << ", " << d << ", " << f << ", " << g << ", " << endl;
376// *out << Verbose(3) << "The three Bond lengths: " << sqrt(d*d+f*f) << ", " << sqrt(d*d+(-0.5*f)*(-0.5*f)+g*g) << ", " << sqrt(d*d+(-0.5*f)*(-0.5*f)+g*g) << endl;
377 factors[0] = d;
378 factors[1] = f;
379 factors[2] = 0.;
380 FirstOtherAtom->x.LinearCombinationOfVectors(&InBondVector, &OrthoVector1, &OrthoVector2, factors);
381 factors[1] = -0.5*f;
382 factors[2] = g;
383 SecondOtherAtom->x.LinearCombinationOfVectors(&InBondVector, &OrthoVector1, &OrthoVector2, factors);
384 factors[2] = -g;
385 ThirdOtherAtom->x.LinearCombinationOfVectors(&InBondVector, &OrthoVector1, &OrthoVector2, factors);
386
387 // rescale each to correct BondDistance
388// FirstOtherAtom->x.Scale(&BondRescale);
389// SecondOtherAtom->x.Scale(&BondRescale);
390// ThirdOtherAtom->x.Scale(&BondRescale);
391
392 // and relative to *origin atom
393 FirstOtherAtom->x.AddVector(&TopOrigin->x);
394 SecondOtherAtom->x.AddVector(&TopOrigin->x);
395 ThirdOtherAtom->x.AddVector(&TopOrigin->x);
396
397 // ... and add to molecule
398 AllWentWell = AllWentWell && AddAtom(FirstOtherAtom);
399 AllWentWell = AllWentWell && AddAtom(SecondOtherAtom);
400 AllWentWell = AllWentWell && AddAtom(ThirdOtherAtom);
401// *out << Verbose(4) << "Added " << *FirstOtherAtom << " at: ";
402// FirstOtherAtom->x.Output(out);
403// *out << endl;
404// *out << Verbose(4) << "Added " << *SecondOtherAtom << " at: ";
405// SecondOtherAtom->x.Output(out);
406// *out << endl;
407// *out << Verbose(4) << "Added " << *ThirdOtherAtom << " at: ";
408// ThirdOtherAtom->x.Output(out);
409// *out << endl;
410 Binder = AddBond(BottomOrigin, FirstOtherAtom, 1);
411 Binder->Cyclic = false;
412 Binder->Type = TreeEdge;
413 Binder = AddBond(BottomOrigin, SecondOtherAtom, 1);
414 Binder->Cyclic = false;
415 Binder->Type = TreeEdge;
416 Binder = AddBond(BottomOrigin, ThirdOtherAtom, 1);
417 Binder->Cyclic = false;
418 Binder->Type = TreeEdge;
419 break;
420 default:
421 cerr << "ERROR: BondDegree does not state single, double or triple bond!" << endl;
422 AllWentWell = false;
423 break;
424 }
425
426// *out << Verbose(3) << "End of AddHydrogenReplacementAtom." << endl;
427 return AllWentWell;
428};
429
430/** Adds given atom \a *pointer from molecule list.
431 * Increases molecule::last_atom and gives last number to added atom.
432 * \param filename name and path of xyz file
433 * \return true - succeeded, false - file not found
434 */
435bool molecule::AddXYZFile(string filename)
436{
437 istringstream *input = NULL;
438 int NumberOfAtoms = 0; // atom number in xyz read
439 int i, j; // loop variables
440 atom *Walker = NULL; // pointer to added atom
441 char shorthand[3]; // shorthand for atom name
442 ifstream xyzfile; // xyz file
443 string line; // currently parsed line
444 double x[3]; // atom coordinates
445
446 xyzfile.open(filename.c_str());
447 if (!xyzfile)
448 return false;
449
450 getline(xyzfile,line,'\n'); // Read numer of atoms in file
451 input = new istringstream(line);
452 *input >> NumberOfAtoms;
453 cout << Verbose(0) << "Parsing " << NumberOfAtoms << " atoms in file." << endl;
454 getline(xyzfile,line,'\n'); // Read comment
455 cout << Verbose(1) << "Comment: " << line << endl;
456
457 for(i=0;i<NumberOfAtoms;i++){
458 Walker = new atom;
459 getline(xyzfile,line,'\n');
460 istringstream *item = new istringstream(line);
461 //istringstream input(line);
462 //cout << Verbose(1) << "Reading: " << line << endl;
463 *item >> shorthand;
464 *item >> x[0];
465 *item >> x[1];
466 *item >> x[2];
467 Walker->type = elemente->FindElement(shorthand);
468 if (Walker->type == NULL) {
469 cerr << "Could not parse the element at line: '" << line << "', setting to H.";
470 Walker->type = elemente->FindElement(1);
471 }
472 for(j=NDIM;j--;)
473 Walker->x.x[j] = x[j];
474 AddAtom(Walker); // add to molecule
475 delete(item);
476 }
477 xyzfile.close();
478 delete(input);
479 return true;
480};
481
482/** Creates a copy of this molecule.
483 * \return copy of molecule
484 */
485molecule *molecule::CopyMolecule()
486{
487 molecule *copy = new molecule(elemente);
488 atom *CurrentAtom = NULL;
489 atom *LeftAtom = NULL, *RightAtom = NULL;
490 atom *Walker = NULL;
491
492 // copy all atoms
493 Walker = start;
494 while(Walker->next != end) {
495 Walker = Walker->next;
496 CurrentAtom = copy->AddCopyAtom(Walker);
497 }
498
499 // copy all bonds
500 bond *Binder = first;
501 bond *NewBond = NULL;
502 while(Binder->next != last) {
503 Binder = Binder->next;
504 // get the pendant atoms of current bond in the copy molecule
505 LeftAtom = copy->start;
506 while (LeftAtom->next != copy->end) {
507 LeftAtom = LeftAtom->next;
508 if (LeftAtom->father == Binder->leftatom)
509 break;
510 }
511 RightAtom = copy->start;
512 while (RightAtom->next != copy->end) {
513 RightAtom = RightAtom->next;
514 if (RightAtom->father == Binder->rightatom)
515 break;
516 }
517 NewBond = copy->AddBond(LeftAtom, RightAtom, Binder->BondDegree);
518 NewBond->Cyclic = Binder->Cyclic;
519 if (Binder->Cyclic)
520 copy->NoCyclicBonds++;
521 NewBond->Type = Binder->Type;
522 }
523 // correct fathers
524 Walker = copy->start;
525 while(Walker->next != copy->end) {
526 Walker = Walker->next;
527 if (Walker->father->father == Walker->father) // same atom in copy's father points to itself
528 Walker->father = Walker; // set father to itself (copy of a whole molecule)
529 else
530 Walker->father = Walker->father->father; // set father to original's father
531 }
532 // copy values
533 copy->CountAtoms((ofstream *)&cout);
534 copy->CountElements();
535 if (first->next != last) { // if adjaceny list is present
536 copy->BondDistance = BondDistance;
537 copy->CreateListOfBondsPerAtom((ofstream *)&cout);
538 }
539
540 return copy;
541};
542
543/** Adds a bond to a the molecule specified by two atoms, \a *first and \a *second.
544 * Also updates molecule::BondCount and molecule::NoNonBonds.
545 * \param *first first atom in bond
546 * \param *second atom in bond
547 * \return pointer to bond or NULL on failure
548 */
549bond * molecule::AddBond(atom *atom1, atom *atom2, int degree=1)
550{
551 bond *Binder = NULL;
552 if ((atom1 != NULL) && (FindAtom(atom1->nr) != NULL) && (atom2 != NULL) && (FindAtom(atom2->nr) != NULL)) {
553 Binder = new bond(atom1, atom2, degree, BondCount++);
554 if ((atom1->type != NULL) && (atom1->type->Z != 1) && (atom2->type != NULL) && (atom2->type->Z != 1))
555 NoNonBonds++;
556 add(Binder, last);
557 } else {
558 cerr << Verbose(1) << "ERROR: Could not add bond between " << atom1->Name << " and " << atom2->Name << " as one or both are not present in the molecule." << endl;
559 }
560 return Binder;
561};
562
563/** Remove bond from bond chain list.
564 * \todo Function not implemented yet
565 * \param *pointer bond pointer
566 * \return true - bound found and removed, false - bond not found/removed
567 */
568bool molecule::RemoveBond(bond *pointer)
569{
570 //cerr << Verbose(1) << "molecule::RemoveBond: Function not implemented yet." << endl;
571 removewithoutcheck(pointer);
572 return true;
573};
574
575/** Remove every bond from bond chain list that atom \a *BondPartner is a constituent of.
576 * \todo Function not implemented yet
577 * \param *BondPartner atom to be removed
578 * \return true - bounds found and removed, false - bonds not found/removed
579 */
580bool molecule::RemoveBonds(atom *BondPartner)
581{
582 cerr << Verbose(1) << "molecule::RemoveBond: Function not implemented yet." << endl;
583 return false;
584};
585
586/** Sets the molecule::cell_size to the components of \a *dim (rectangular box)
587 * \param *dim vector class
588 */
589void molecule::SetBoxDimension(vector *dim)
590{
591 cell_size[0] = dim->x[0];
592 cell_size[1] = 0.;
593 cell_size[2] = dim->x[1];
594 cell_size[3] = 0.;
595 cell_size[4] = 0.;
596 cell_size[5] = dim->x[2];
597};
598
599/** Centers the molecule in the box whose lengths are defined by vector \a *BoxLengths.
600 * \param *out output stream for debugging
601 * \param *BoxLengths box lengths
602 */
603bool molecule::CenterInBox(ofstream *out, vector *BoxLengths)
604{
605 bool status = true;
606 atom *ptr = NULL;
607 vector *min = new vector;
608 vector *max = new vector;
609
610 // gather min and max for each axis
611 ptr = start->next; // start at first in list
612 if (ptr != end) { //list not empty?
613 for (int i=NDIM;i--;) {
614 max->x[i] = ptr->x.x[i];
615 min->x[i] = ptr->x.x[i];
616 }
617 while (ptr->next != end) { // continue with second if present
618 ptr = ptr->next;
619 //ptr->Output(1,1,out);
620 for (int i=NDIM;i--;) {
621 max->x[i] = (max->x[i] < ptr->x.x[i]) ? ptr->x.x[i] : max->x[i];
622 min->x[i] = (min->x[i] > ptr->x.x[i]) ? ptr->x.x[i] : min->x[i];
623 }
624 }
625 }
626 // sanity check
627 for(int i=NDIM;i--;) {
628 if (max->x[i] - min->x[i] > BoxLengths->x[i])
629 status = false;
630 }
631 // warn if check failed
632 if (!status)
633 *out << "WARNING: molecule is bigger than defined box!" << endl;
634 else { // else center in box
635 ptr = start;
636 while (ptr->next != end) {
637 ptr = ptr->next;
638 for (int i=NDIM;i--;)
639 ptr->x.x[i] += -(max->x[i] + min->x[i])/2. + BoxLengths->x[i]/2.; // first term centers molecule at (0,0,0), second shifts to center of new box
640 }
641 }
642
643 // free and exit
644 delete(min);
645 delete(max);
646 return status;
647};
648
649/** Centers the edge of the atoms at (0,0,0).
650 * \param *out output stream for debugging
651 * \param *max coordinates of other edge, specifying box dimensions.
652 */
653void molecule::CenterEdge(ofstream *out, vector *max)
654{
655 vector *min = new vector;
656
657// *out << Verbose(3) << "Begin of CenterEdge." << endl;
658 atom *ptr = start->next; // start at first in list
659 if (ptr != end) { //list not empty?
660 for (int i=NDIM;i--;) {
661 max->x[i] = ptr->x.x[i];
662 min->x[i] = ptr->x.x[i];
663 }
664 while (ptr->next != end) { // continue with second if present
665 ptr = ptr->next;
666 //ptr->Output(1,1,out);
667 for (int i=NDIM;i--;) {
668 max->x[i] = (max->x[i] < ptr->x.x[i]) ? ptr->x.x[i] : max->x[i];
669 min->x[i] = (min->x[i] > ptr->x.x[i]) ? ptr->x.x[i] : min->x[i];
670 }
671 }
672// *out << Verbose(4) << "Maximum is ";
673// max->Output(out);
674// *out << ", Minimum is ";
675// min->Output(out);
676// *out << endl;
677
678 for (int i=NDIM;i--;) {
679 min->x[i] *= -1.;
680 max->x[i] += min->x[i];
681 }
682 Translate(min);
683 }
684 delete(min);
685// *out << Verbose(3) << "End of CenterEdge." << endl;
686};
687
688/** Centers the center of the atoms at (0,0,0).
689 * \param *out output stream for debugging
690 * \param *center return vector for translation vector
691 */
692void molecule::CenterOrigin(ofstream *out, vector *center)
693{
694 int Num = 0;
695 atom *ptr = start->next; // start at first in list
696
697 for(int i=NDIM;i--;) // zero center vector
698 center->x[i] = 0.;
699
700 if (ptr != end) { //list not empty?
701 while (ptr->next != end) { // continue with second if present
702 ptr = ptr->next;
703 Num++;
704 center->AddVector(&ptr->x);
705 }
706 center->Scale(-1./Num); // divide through total number (and sign for direction)
707 Translate(center);
708 }
709};
710
711/** Returns vector pointing to center of gravity.
712 * \param *out output stream for debugging
713 * \return pointer to center of gravity vector
714 */
715vector * molecule::DetermineCenterOfGravity(ofstream *out)
716{
717 atom *ptr = start->next; // start at first in list
718 vector *a = new vector();
719 vector tmp;
720 double Num = 0;
721
722 a->Zero();
723
724 if (ptr != end) { //list not empty?
725 while (ptr->next != end) { // continue with second if present
726 ptr = ptr->next;
727 Num += ptr->type->mass;
728 tmp.CopyVector(&ptr->x);
729 tmp.Scale(ptr->type->mass); // scale by mass
730 a->AddVector(&tmp);
731 }
732 a->Scale(-1./Num); // divide through total mass (and sign for direction)
733 }
734 *out << Verbose(1) << "Resulting center of gravity: ";
735 a->Output(out);
736 *out << endl;
737 return a;
738};
739
740/** Centers the center of gravity of the atoms at (0,0,0).
741 * \param *out output stream for debugging
742 * \param *center return vector for translation vector
743 */
744void molecule::CenterGravity(ofstream *out, vector *center)
745{
746 if (center == NULL) {
747 DetermineCenter(*center);
748 Translate(center);
749 delete(center);
750 } else {
751 Translate(center);
752 }
753};
754
755/** Scales all atoms by \a *factor.
756 * \param *factor pointer to scaling factor
757 */
758void molecule::Scale(double **factor)
759{
760 atom *ptr = start;
761
762 while (ptr->next != end) {
763 ptr = ptr->next;
764 ptr->x.Scale(factor);
765 }
766};
767
768/** Translate all atoms by given vector.
769 * \param trans[] translation vector.
770 */
771void molecule::Translate(const vector *trans)
772{
773 atom *ptr = start;
774
775 while (ptr->next != end) {
776 ptr = ptr->next;
777 ptr->x.Translate(trans);
778 }
779};
780
781/** Mirrors all atoms against a given plane.
782 * \param n[] normal vector of mirror plane.
783 */
784void molecule::Mirror(const vector *n)
785{
786 atom *ptr = start;
787
788 while (ptr->next != end) {
789 ptr = ptr->next;
790 ptr->x.Mirror(n);
791 }
792};
793
794/** Determines center of molecule (yet not considering atom masses).
795 * \param Center reference to return vector
796 */
797void molecule::DetermineCenter(vector &Center)
798{
799 atom *Walker = start;
800 bond *Binder = NULL;
801 double *matrix = ReturnFullMatrixforSymmetric(cell_size);
802 double tmp;
803 bool flag;
804 vector TestVector, TranslationVector;
805
806 do {
807 Center.Zero();
808 flag = true;
809 while (Walker->next != end) {
810 Walker = Walker->next;
811#ifdef ADDHYDROGEN
812 if (Walker->type->Z != 1) {
813#endif
814 TestVector.CopyVector(&Walker->x);
815 TestVector.InverseMatrixMultiplication(matrix);
816 TranslationVector.Zero();
817 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr]; i++) {
818 Binder = ListOfBondsPerAtom[Walker->nr][i];
819 if (Walker->nr < Binder->GetOtherAtom(Walker)->nr) // otherwise we shift one to, the other fro and gain nothing
820 for (int j=0;j<NDIM;j++) {
821 tmp = Walker->x.x[j] - Binder->GetOtherAtom(Walker)->x.x[j];
822 if ((fabs(tmp)) > BondDistance) {
823 flag = false;
824 cout << Verbose(0) << "Hit: atom " << Walker->Name << " in bond " << *Binder << " has to be shifted due to " << tmp << "." << endl;
825 if (tmp > 0)
826 TranslationVector.x[j] -= 1.;
827 else
828 TranslationVector.x[j] += 1.;
829 }
830 }
831 }
832 TestVector.AddVector(&TranslationVector);
833 TestVector.MatrixMultiplication(matrix);
834 Center.AddVector(&TestVector);
835 cout << Verbose(1) << "Vector is: ";
836 TestVector.Output((ofstream *)&cout);
837 cout << endl;
838#ifdef ADDHYDROGEN
839 // now also change all hydrogens
840 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr]; i++) {
841 Binder = ListOfBondsPerAtom[Walker->nr][i];
842 if (Binder->GetOtherAtom(Walker)->type->Z == 1) {
843 TestVector.CopyVector(&Binder->GetOtherAtom(Walker)->x);
844 TestVector.InverseMatrixMultiplication(matrix);
845 TestVector.AddVector(&TranslationVector);
846 TestVector.MatrixMultiplication(matrix);
847 Center.AddVector(&TestVector);
848 cout << Verbose(1) << "Hydrogen Vector is: ";
849 TestVector.Output((ofstream *)&cout);
850 cout << endl;
851 }
852 }
853 }
854#endif
855 }
856 } while (!flag);
857 Free((void **)&matrix, "molecule::DetermineCenter: *matrix");
858 Center.Scale(1./(double)AtomCount);
859};
860
861/** Transforms/Rotates the given molecule into its principal axis system.
862 * \param *out output stream for debugging
863 * \param DoRotate whether to rotate (true) or only to determine the PAS.
864 */
865void molecule::PrincipalAxisSystem(ofstream *out, bool DoRotate)
866{
867 atom *ptr = start; // start at first in list
868 double InertiaTensor[NDIM*NDIM];
869 vector *CenterOfGravity = DetermineCenterOfGravity(out);
870
871 CenterGravity(out, CenterOfGravity);
872
873 // reset inertia tensor
874 for(int i=0;i<NDIM*NDIM;i++)
875 InertiaTensor[i] = 0.;
876
877 // sum up inertia tensor
878 while (ptr->next != end) {
879 ptr = ptr->next;
880 vector x;
881 x.CopyVector(&ptr->x);
882 //x.SubtractVector(CenterOfGravity);
883 InertiaTensor[0] += ptr->type->mass*(x.x[1]*x.x[1] + x.x[2]*x.x[2]);
884 InertiaTensor[1] += ptr->type->mass*(-x.x[0]*x.x[1]);
885 InertiaTensor[2] += ptr->type->mass*(-x.x[0]*x.x[2]);
886 InertiaTensor[3] += ptr->type->mass*(-x.x[1]*x.x[0]);
887 InertiaTensor[4] += ptr->type->mass*(x.x[0]*x.x[0] + x.x[2]*x.x[2]);
888 InertiaTensor[5] += ptr->type->mass*(-x.x[1]*x.x[2]);
889 InertiaTensor[6] += ptr->type->mass*(-x.x[2]*x.x[0]);
890 InertiaTensor[7] += ptr->type->mass*(-x.x[2]*x.x[1]);
891 InertiaTensor[8] += ptr->type->mass*(x.x[0]*x.x[0] + x.x[1]*x.x[1]);
892 }
893 // print InertiaTensor for debugging
894 *out << "The inertia tensor is:" << endl;
895 for(int i=0;i<NDIM;i++) {
896 for(int j=0;j<NDIM;j++)
897 *out << InertiaTensor[i*NDIM+j] << " ";
898 *out << endl;
899 }
900 *out << endl;
901
902 // diagonalize to determine principal axis system
903 gsl_eigen_symmv_workspace *T = gsl_eigen_symmv_alloc(NDIM);
904 gsl_matrix_view m = gsl_matrix_view_array(InertiaTensor, NDIM, NDIM);
905 gsl_vector *eval = gsl_vector_alloc(NDIM);
906 gsl_matrix *evec = gsl_matrix_alloc(NDIM, NDIM);
907 gsl_eigen_symmv(&m.matrix, eval, evec, T);
908 gsl_eigen_symmv_free(T);
909 gsl_eigen_symmv_sort(eval, evec, GSL_EIGEN_SORT_ABS_DESC);
910
911 for(int i=0;i<NDIM;i++) {
912 *out << Verbose(1) << "eigenvalue = " << gsl_vector_get(eval, i);
913 *out << ", eigenvector = (" << evec->data[i * evec->tda + 0] << "," << evec->data[i * evec->tda + 1] << "," << evec->data[i * evec->tda + 2] << ")" << endl;
914 }
915
916 // check whether we rotate or not
917 if (DoRotate) {
918 *out << Verbose(1) << "Transforming molecule into PAS ... ";
919 // the eigenvectors specify the transformation matrix
920 ptr = start;
921 while (ptr->next != end) {
922 ptr = ptr->next;
923 ptr->x.MatrixMultiplication(evec->data);
924 }
925 *out << "done." << endl;
926
927 // summing anew for debugging (resulting matrix has to be diagonal!)
928 // reset inertia tensor
929 for(int i=0;i<NDIM*NDIM;i++)
930 InertiaTensor[i] = 0.;
931
932 // sum up inertia tensor
933 ptr = start;
934 while (ptr->next != end) {
935 ptr = ptr->next;
936 vector x;
937 x.CopyVector(&ptr->x);
938 //x.SubtractVector(CenterOfGravity);
939 InertiaTensor[0] += ptr->type->mass*(x.x[1]*x.x[1] + x.x[2]*x.x[2]);
940 InertiaTensor[1] += ptr->type->mass*(-x.x[0]*x.x[1]);
941 InertiaTensor[2] += ptr->type->mass*(-x.x[0]*x.x[2]);
942 InertiaTensor[3] += ptr->type->mass*(-x.x[1]*x.x[0]);
943 InertiaTensor[4] += ptr->type->mass*(x.x[0]*x.x[0] + x.x[2]*x.x[2]);
944 InertiaTensor[5] += ptr->type->mass*(-x.x[1]*x.x[2]);
945 InertiaTensor[6] += ptr->type->mass*(-x.x[2]*x.x[0]);
946 InertiaTensor[7] += ptr->type->mass*(-x.x[2]*x.x[1]);
947 InertiaTensor[8] += ptr->type->mass*(x.x[0]*x.x[0] + x.x[1]*x.x[1]);
948 }
949 // print InertiaTensor for debugging
950 *out << "The inertia tensor is:" << endl;
951 for(int i=0;i<NDIM;i++) {
952 for(int j=0;j<NDIM;j++)
953 *out << InertiaTensor[i*NDIM+j] << " ";
954 *out << endl;
955 }
956 *out << endl;
957 }
958
959 // free everything
960 delete(CenterOfGravity);
961 gsl_vector_free(eval);
962 gsl_matrix_free(evec);
963};
964
965/** Align all atoms in such a manner that given vector \a *n is along z axis.
966 * \param n[] alignment vector.
967 */
968void molecule::Align(vector *n)
969{
970 atom *ptr = start;
971 double alpha, tmp;
972 vector z_axis;
973 z_axis.x[0] = 0.;
974 z_axis.x[1] = 0.;
975 z_axis.x[2] = 1.;
976
977 // rotate on z-x plane
978 cout << Verbose(0) << "Begin of Aligning all atoms." << endl;
979 alpha = atan(-n->x[0]/n->x[2]);
980 cout << Verbose(1) << "Z-X-angle: " << alpha << " ... ";
981 while (ptr->next != end) {
982 ptr = ptr->next;
983 tmp = ptr->x.x[0];
984 ptr->x.x[0] = cos(alpha) * tmp + sin(alpha) * ptr->x.x[2];
985 ptr->x.x[2] = -sin(alpha) * tmp + cos(alpha) * ptr->x.x[2];
986 }
987 // rotate n vector
988 tmp = n->x[0];
989 n->x[0] = cos(alpha) * tmp + sin(alpha) * n->x[2];
990 n->x[2] = -sin(alpha) * tmp + cos(alpha) * n->x[2];
991 cout << Verbose(1) << "alignment vector after first rotation: ";
992 n->Output((ofstream *)&cout);
993 cout << endl;
994
995 // rotate on z-y plane
996 ptr = start;
997 alpha = atan(-n->x[1]/n->x[2]);
998 cout << Verbose(1) << "Z-Y-angle: " << alpha << " ... ";
999 while (ptr->next != end) {
1000 ptr = ptr->next;
1001 tmp = ptr->x.x[1];
1002 ptr->x.x[1] = cos(alpha) * tmp + sin(alpha) * ptr->x.x[2];
1003 ptr->x.x[2] = -sin(alpha) * tmp + cos(alpha) * ptr->x.x[2];
1004 }
1005 // rotate n vector (for consistency check)
1006 tmp = n->x[1];
1007 n->x[1] = cos(alpha) * tmp + sin(alpha) * n->x[2];
1008 n->x[2] = -sin(alpha) * tmp + cos(alpha) * n->x[2];
1009
1010 cout << Verbose(1) << "alignment vector after second rotation: ";
1011 n->Output((ofstream *)&cout);
1012 cout << Verbose(1) << endl;
1013 cout << Verbose(0) << "End of Aligning all atoms." << endl;
1014};
1015
1016/** Removes atom from molecule list.
1017 * \param *pointer atom to be removed
1018 * \return true - succeeded, false - atom not found in list
1019 */
1020bool molecule::RemoveAtom(atom *pointer)
1021{
1022 if (ElementsInMolecule[pointer->type->Z] != 0) // this would indicate an error
1023 ElementsInMolecule[pointer->type->Z]--; // decrease number of atom of this element
1024 else
1025 cerr << "ERROR: Atom " << pointer->Name << " is of element " << pointer->type->Z << " but the entry in the table of the molecule is 0!" << endl;
1026 if (ElementsInMolecule[pointer->type->Z] == 0) // was last atom of this element?
1027 ElementCount--;
1028 return remove(pointer, start, end);
1029};
1030
1031/** Removes every atom from molecule list.
1032 * \return true - succeeded, false - atom not found in list
1033 */
1034bool molecule::CleanupMolecule()
1035{
1036 return (cleanup(start,end) && cleanup(first,last));
1037};
1038
1039/** Finds an atom specified by its continuous number.
1040 * \param Nr number of atom withim molecule
1041 * \return pointer to atom or NULL
1042 */
1043atom * molecule::FindAtom(int Nr) const{
1044 atom * walker = find(&Nr, start,end);
1045 if (walker != NULL) {
1046 //cout << Verbose(0) << "Found Atom Nr. " << walker->nr << endl;
1047 return walker;
1048 } else {
1049 cout << Verbose(0) << "Atom not found in list." << endl;
1050 return NULL;
1051 }
1052};
1053
1054/** Asks for atom number, and checks whether in list.
1055 * \param *text question before entering
1056 */
1057atom * molecule::AskAtom(string text)
1058{
1059 int No;
1060 atom *ion = NULL;
1061 do {
1062 //cout << Verbose(0) << "============Atom list==========================" << endl;
1063 //mol->Output((ofstream *)&cout);
1064 //cout << Verbose(0) << "===============================================" << endl;
1065 cout << Verbose(0) << text;
1066 cin >> No;
1067 ion = this->FindAtom(No);
1068 } while (ion == NULL);
1069 return ion;
1070};
1071
1072/** Checks if given coordinates are within cell volume.
1073 * \param *x array of coordinates
1074 * \return true - is within, false - out of cell
1075 */
1076bool molecule::CheckBounds(const vector *x) const
1077{
1078 bool result = true;
1079 int j =-1;
1080 for (int i=0;i<NDIM;i++) {
1081 j += i+1;
1082 result = result && ((x->x[i] >= 0) && (x->x[i] < cell_size[j]));
1083 }
1084 //return result;
1085 return true; /// probably not gonna use the check no more
1086};
1087
1088/** Calculates sum over least square distance to line hidden in \a *x.
1089 * \param *x offset and direction vector
1090 * \param *params pointer to lsq_params structure
1091 * \return \f$ sum_i^N | y_i - (a + t_i b)|^2\f$
1092 */
1093double LeastSquareDistance (const gsl_vector * x, void * params)
1094{
1095 double res = 0, t;
1096 vector a,b,c,d;
1097 struct lsq_params *par = (struct lsq_params *)params;
1098 atom *ptr = par->mol->start;
1099
1100 // initialize vectors
1101 a.x[0] = gsl_vector_get(x,0);
1102 a.x[1] = gsl_vector_get(x,1);
1103 a.x[2] = gsl_vector_get(x,2);
1104 b.x[0] = gsl_vector_get(x,3);
1105 b.x[1] = gsl_vector_get(x,4);
1106 b.x[2] = gsl_vector_get(x,5);
1107 // go through all atoms
1108 while (ptr != par->mol->end) {
1109 ptr = ptr->next;
1110 if (ptr->type == ((struct lsq_params *)params)->type) { // for specific type
1111 c.CopyVector(&ptr->x); // copy vector to temporary one
1112 c.SubtractVector(&a); // subtract offset vector
1113 t = c.ScalarProduct(&b); // get direction parameter
1114 d.CopyVector(&b); // and create vector
1115 d.Scale(&t);
1116 c.SubtractVector(&d); // ... yielding distance vector
1117 res += d.ScalarProduct((const vector *)&d); // add squared distance
1118 }
1119 }
1120 return res;
1121};
1122
1123/** By minimizing the least square distance gains alignment vector.
1124 * \bug this is not yet working properly it seems
1125 */
1126void molecule::GetAlignVector(struct lsq_params * par) const
1127{
1128 int np = 6;
1129
1130 const gsl_multimin_fminimizer_type *T =
1131 gsl_multimin_fminimizer_nmsimplex;
1132 gsl_multimin_fminimizer *s = NULL;
1133 gsl_vector *ss;
1134 gsl_multimin_function minex_func;
1135
1136 size_t iter = 0, i;
1137 int status;
1138 double size;
1139
1140 /* Initial vertex size vector */
1141 ss = gsl_vector_alloc (np);
1142
1143 /* Set all step sizes to 1 */
1144 gsl_vector_set_all (ss, 1.0);
1145
1146 /* Starting point */
1147 par->x = gsl_vector_alloc (np);
1148 par->mol = this;
1149
1150 gsl_vector_set (par->x, 0, 0.0); // offset
1151 gsl_vector_set (par->x, 1, 0.0);
1152 gsl_vector_set (par->x, 2, 0.0);
1153 gsl_vector_set (par->x, 3, 0.0); // direction
1154 gsl_vector_set (par->x, 4, 0.0);
1155 gsl_vector_set (par->x, 5, 1.0);
1156
1157 /* Initialize method and iterate */
1158 minex_func.f = &LeastSquareDistance;
1159 minex_func.n = np;
1160 minex_func.params = (void *)par;
1161
1162 s = gsl_multimin_fminimizer_alloc (T, np);
1163 gsl_multimin_fminimizer_set (s, &minex_func, par->x, ss);
1164
1165 do
1166 {
1167 iter++;
1168 status = gsl_multimin_fminimizer_iterate(s);
1169
1170 if (status)
1171 break;
1172
1173 size = gsl_multimin_fminimizer_size (s);
1174 status = gsl_multimin_test_size (size, 1e-2);
1175
1176 if (status == GSL_SUCCESS)
1177 {
1178 printf ("converged to minimum at\n");
1179 }
1180
1181 printf ("%5d ", (int)iter);
1182 for (i = 0; i < (size_t)np; i++)
1183 {
1184 printf ("%10.3e ", gsl_vector_get (s->x, i));
1185 }
1186 printf ("f() = %7.3f size = %.3f\n", s->fval, size);
1187 }
1188 while (status == GSL_CONTINUE && iter < 100);
1189
1190 for (i=0;i<(size_t)np;i++)
1191 gsl_vector_set(par->x, i, gsl_vector_get(s->x, i));
1192 //gsl_vector_free(par->x);
1193 gsl_vector_free(ss);
1194 gsl_multimin_fminimizer_free (s);
1195};
1196
1197/** Prints molecule to *out.
1198 * \param *out output stream
1199 */
1200bool molecule::Output(ofstream *out)
1201{
1202 element *runner = elemente->start;
1203 atom *walker = NULL;
1204 int ElementNo, AtomNo;
1205 CountElements();
1206
1207 if (out == NULL) {
1208 return false;
1209 } else {
1210 *out << "#Ion_TypeNr._Nr.R[0] R[1] R[2] MoveType (0 MoveIon, 1 FixedIon)" << endl;
1211 ElementNo = 0;
1212 while (runner->next != elemente->end) { // go through every element
1213 runner = runner->next;
1214 if (ElementsInMolecule[runner->Z]) { // if this element got atoms
1215 ElementNo++;
1216 AtomNo = 0;
1217 walker = start;
1218 while (walker->next != end) { // go through every atom of this element
1219 walker = walker->next;
1220 if (walker->type == runner) { // if this atom fits to element
1221 AtomNo++;
1222 walker->Output(ElementNo, AtomNo, out);
1223 }
1224 }
1225 }
1226 }
1227 return true;
1228 }
1229};
1230
1231/** Outputs contents of molecule::ListOfBondsPerAtom.
1232 * \param *out output stream
1233 */
1234void molecule::OutputListOfBonds(ofstream *out) const
1235{
1236 *out << Verbose(2) << endl << "From Contents of ListOfBondsPerAtom, all non-hydrogen atoms:" << endl;
1237 atom *Walker = start;
1238 while (Walker->next != end) {
1239 Walker = Walker->next;
1240#ifdef ADDHYDROGEN
1241 if (Walker->type->Z != 1) { // regard only non-hydrogen
1242#endif
1243 *out << Verbose(2) << "Atom " << Walker->Name << " has Bonds: "<<endl;
1244 for(int j=0;j<NumberOfBondsPerAtom[Walker->nr];j++) {
1245 *out << Verbose(3) << *(ListOfBondsPerAtom)[Walker->nr][j] << endl;
1246 }
1247#ifdef ADDHYDROGEN
1248 }
1249#endif
1250 }
1251 *out << endl;
1252};
1253
1254/** Output of element before the actual coordination list.
1255 * \param *out stream pointer
1256 */
1257bool molecule::Checkout(ofstream *out) const
1258{
1259 return elemente->Checkout(out, ElementsInMolecule);
1260};
1261
1262/** Prints molecule to *out as xyz file.
1263 * \param *out output stream
1264 */
1265bool molecule::OutputXYZ(ofstream *out) const
1266{
1267 atom *walker = NULL;
1268 int No = 0;
1269 time_t now;
1270
1271 now = time((time_t *)NULL); // Get the system time and put it into 'now' as 'calender time'
1272 walker = start;
1273 while (walker->next != end) { // go through every atom and count
1274 walker = walker->next;
1275 No++;
1276 }
1277 if (out != NULL) {
1278 *out << No << "\n\tCreated by molecuilder on " << ctime(&now);
1279 walker = start;
1280 while (walker->next != end) { // go through every atom of this element
1281 walker = walker->next;
1282 walker->OutputXYZLine(out);
1283 }
1284 return true;
1285 } else
1286 return false;
1287};
1288
1289/** Brings molecule::AtomCount and atom::*Name up-to-date.
1290 * \param *out output stream for debugging
1291 */
1292void molecule::CountAtoms(ofstream *out)
1293{
1294 int i = 0;
1295 atom *Walker = start;
1296 while (Walker->next != end) {
1297 Walker = Walker->next;
1298 i++;
1299 }
1300 if ((AtomCount == 0) || (i != AtomCount)) {
1301 *out << Verbose(3) << "Mismatch in AtomCount " << AtomCount << " and recounted number " << i << ", renaming all." << endl;
1302 AtomCount = i;
1303
1304 // count NonHydrogen atoms and give each atom a unique name
1305 if (AtomCount != 0) {
1306 i=0;
1307 NoNonHydrogen = 0;
1308 Walker = start;
1309 while (Walker->next != end) {
1310 Walker = Walker->next;
1311 Walker->nr = i; // update number in molecule (for easier referencing in FragmentMolecule lateron)
1312 if (Walker->type->Z != 1) // count non-hydrogen atoms whilst at it
1313 NoNonHydrogen++;
1314 Free((void **)&Walker->Name, "molecule::CountAtoms: *walker->Name");
1315 Walker->Name = (char *) Malloc(sizeof(char)*6, "molecule::CountAtoms: *walker->Name");
1316 sprintf(Walker->Name, "%2s%02d", Walker->type->symbol, Walker->nr+1);
1317 *out << "Naming atom nr. " << Walker->nr << " " << Walker->Name << "." << endl;
1318 i++;
1319 }
1320 } else
1321 *out << Verbose(3) << "AtomCount is still " << AtomCount << ", thus counting nothing." << endl;
1322 }
1323};
1324
1325/** Brings molecule::ElementCount and molecule::ElementsInMolecule up-to-date.
1326 */
1327void molecule::CountElements()
1328{
1329 int i = 0;
1330 for(i=MAX_ELEMENTS;i--;)
1331 ElementsInMolecule[i] = 0;
1332 ElementCount = 0;
1333
1334 atom *walker = start;
1335 while (walker->next != end) {
1336 walker = walker->next;
1337 ElementsInMolecule[walker->type->Z]++;
1338 i++;
1339 }
1340 for(i=MAX_ELEMENTS;i--;)
1341 ElementCount += (ElementsInMolecule[i] != 0 ? 1 : 0);
1342};
1343
1344/** Counts all cyclic bonds and returns their number.
1345 * \note Hydrogen bonds can never by cyclic, thus no check for that
1346 * \param *out output stream for debugging
1347 * \return number opf cyclic bonds
1348 */
1349int molecule::CountCyclicBonds(ofstream *out)
1350{
1351 int No = 0;
1352 int *MinimumRingSize = NULL;
1353 MoleculeLeafClass *Subgraphs = NULL;
1354 bond *Binder = first;
1355 if ((Binder->next != last) && (Binder->next->Type == Undetermined)) {
1356 *out << Verbose(0) << "No Depth-First-Search analysis performed so far, calling ..." << endl;
1357 Subgraphs = DepthFirstSearchAnalysis(out, MinimumRingSize);
1358 while (Subgraphs->next != NULL) {
1359 Subgraphs = Subgraphs->next;
1360 delete(Subgraphs->previous);
1361 }
1362 delete(Subgraphs);
1363 delete[](MinimumRingSize);
1364 }
1365 while(Binder->next != last) {
1366 Binder = Binder->next;
1367 if (Binder->Cyclic)
1368 No++;
1369 }
1370 return No;
1371};
1372/** Returns Shading as a char string.
1373 * \param color the Shading
1374 * \return string of the flag
1375 */
1376string molecule::GetColor(enum Shading color)
1377{
1378 switch(color) {
1379 case white:
1380 return "white";
1381 break;
1382 case lightgray:
1383 return "lightgray";
1384 break;
1385 case darkgray:
1386 return "darkgray";
1387 break;
1388 case black:
1389 return "black";
1390 break;
1391 default:
1392 return "uncolored";
1393 break;
1394 };
1395};
1396
1397
1398/** Counts necessary number of valence electrons and returns number and SpinType.
1399 * \param configuration containing everything
1400 */
1401void molecule::CalculateOrbitals(class config &configuration)
1402{
1403 configuration.MaxPsiDouble = configuration.PsiMaxNoDown = configuration.PsiMaxNoUp = configuration.PsiType = 0;
1404 for(int i=MAX_ELEMENTS;i--;) {
1405 if (ElementsInMolecule[i] != 0) {
1406 //cout << "CalculateOrbitals: " << elemente->FindElement(i)->name << " has a valence of " << (int)elemente->FindElement(i)->Valence << " and there are " << ElementsInMolecule[i] << " of it." << endl;
1407 configuration.MaxPsiDouble += ElementsInMolecule[i]*((int)elemente->FindElement(i)->Valence);
1408 }
1409 }
1410 configuration.PsiMaxNoDown = configuration.MaxPsiDouble/2 + (configuration.MaxPsiDouble % 2);
1411 configuration.PsiMaxNoUp = configuration.MaxPsiDouble/2;
1412 configuration.MaxPsiDouble /= 2;
1413 configuration.PsiType = (configuration.PsiMaxNoDown == configuration.PsiMaxNoUp) ? 0 : 1;
1414 if ((configuration.PsiType == 1) && (configuration.ProcPEPsi < 2)) {
1415 configuration.ProcPEGamma /= 2;
1416 configuration.ProcPEPsi *= 2;
1417 } else {
1418 configuration.ProcPEGamma *= configuration.ProcPEPsi;
1419 configuration.ProcPEPsi = 1;
1420 }
1421 configuration.InitMaxMinStopStep = configuration.MaxMinStopStep = configuration.MaxPsiDouble;
1422};
1423
1424/** Creates an adjacency list of the molecule.
1425 * Generally, we use the CSD approach to bond recognition, that is the the distance
1426 * between two atoms A and B must be within [Rcov(A)+Rcov(B)-t,Rcov(A)+Rcov(B)+t] with
1427 * a threshold t = 0.4 Angstroem.
1428 * To make it O(N log N) the function uses the linked-cell technique as follows:
1429 * The procedure is step-wise:
1430 * -# Remove every bond in list
1431 * -# Count the atoms in the molecule with CountAtoms()
1432 * -# partition cell into smaller linked cells of size \a bonddistance
1433 * -# put each atom into its corresponding cell
1434 * -# go through every cell, check the atoms therein against all possible bond partners in the 27 adjacent cells, add bond if true
1435 * -# create the list of bonds via CreateListOfBondsPerAtom()
1436 * -# correct the bond degree iteratively (single->double->triple bond)
1437 * -# finally print the bond list to \a *out if desired
1438 * \param *out out stream for printing the matrix, NULL if no output
1439 * \param bonddistance length of linked cells (i.e. maximum minimal length checked)
1440 * \param IsAngstroem whether coordinate system is gauged to Angstroem or Bohr radii
1441 */
1442void molecule::CreateAdjacencyList(ofstream *out, double bonddistance, bool IsAngstroem)
1443{
1444 atom *Walker = NULL, *OtherWalker = NULL;
1445 int No, NoBonds;
1446 int NumberCells, divisor[NDIM], n[NDIM], N[NDIM], index, Index, j;
1447 molecule **CellList;
1448 double distance, MinDistance, MaxDistance;
1449 double *matrix = ReturnFullMatrixforSymmetric(cell_size);
1450 vector x;
1451
1452 BondDistance = bonddistance; // * ((IsAngstroem) ? 1. : 1./AtomicLengthToAngstroem);
1453 *out << Verbose(0) << "Begin of CreateAdjacencyList." << endl;
1454 // remove every bond from the list
1455 if ((first->next != last) && (last->previous != first)) { // there are bonds present
1456 cleanup(first,last);
1457 }
1458
1459 // count atoms in molecule = dimension of matrix (also give each unique name and continuous numbering)
1460 CountAtoms(out);
1461 *out << Verbose(1) << "AtomCount " << AtomCount << "." << endl;
1462
1463 if (AtomCount != 0) {
1464 // 1. find divisor for each axis, such that a sphere with radius of at least bonddistance can be placed into each cell
1465 j=-1;
1466 for (int i=0;i<NDIM;i++) {
1467 j += i+1;
1468 divisor[i] = (int)floor(cell_size[j]/bonddistance); // take smaller value such that size of linked cell is at least bonddistance
1469 *out << Verbose(1) << "divisor[" << i << "] = " << divisor[i] << "." << endl;
1470 }
1471 // 2a. allocate memory for the cell list
1472 NumberCells = divisor[0]*divisor[1]*divisor[2];
1473 *out << Verbose(1) << "Allocating " << NumberCells << " cells." << endl;
1474 CellList = (molecule **) Malloc(sizeof(molecule *)*NumberCells, "molecule::CreateAdjacencyList - ** CellList");
1475 for (int i=NumberCells;i--;)
1476 CellList[i] = NULL;
1477
1478 // 2b. put all atoms into its corresponding list
1479 Walker = start;
1480 while(Walker->next != end) {
1481 Walker = Walker->next;
1482 //*out << Verbose(1) << "Current atom is " << *Walker << " with coordinates ";
1483 //Walker->x.Output(out);
1484 //*out << "." << endl;
1485 // compute the cell by the atom's coordinates
1486 j=-1;
1487 for (int i=0;i<NDIM;i++) {
1488 j += i+1;
1489 x.CopyVector(&(Walker->x));
1490 x.KeepPeriodic(out, matrix);
1491 n[i] = (int)floor(x.x[i]/cell_size[j]*(double)divisor[i]);
1492 }
1493 index = n[2] + (n[1] + n[0] * divisor[1]) * divisor[2];
1494 *out << Verbose(1) << "Atom " << *Walker << " goes into cell number [" << n[0] << "," << n[1] << "," << n[2] << "] = " << index << "." << endl;
1495 // add copy atom to this cell
1496 if (CellList[index] == NULL) // allocate molecule if not done
1497 CellList[index] = new molecule(elemente);
1498 OtherWalker = CellList[index]->AddCopyAtom(Walker); // add a copy of walker to this atom, father will be walker for later reference
1499 //*out << Verbose(1) << "Copy Atom is " << *OtherWalker << "." << endl;
1500 }
1501 //for (int i=0;i<NumberCells;i++)
1502 //*out << Verbose(1) << "Cell number " << i << ": " << CellList[i] << "." << endl;
1503
1504 // 3a. go through every cell
1505 for (N[0]=divisor[0];N[0]--;)
1506 for (N[1]=divisor[1];N[1]--;)
1507 for (N[2]=divisor[2];N[2]--;) {
1508 Index = N[2] + (N[1] + N[0] * divisor[1]) * divisor[2];
1509 if (CellList[Index] != NULL) { // if there atoms in this cell
1510 //*out << Verbose(1) << "Current cell is " << Index << "." << endl;
1511 // 3b. for every atom therein
1512 Walker = CellList[Index]->start;
1513 while (Walker->next != CellList[Index]->end) { // go through every atom
1514 Walker = Walker->next;
1515 //*out << Verbose(0) << "Current Atom is " << *Walker << "." << endl;
1516 // 3c. check for possible bond between each atom in this and every one in the 27 cells
1517 for (n[0]=-1;n[0]<=1;n[0]++)
1518 for (n[1]=-1;n[1]<=1;n[1]++)
1519 for (n[2]=-1;n[2]<=1;n[2]++) {
1520 // compute the index of this comparison cell and make it periodic
1521 index = ((N[2]+n[2]+divisor[2])%divisor[2]) + (((N[1]+n[1]+divisor[1])%divisor[1]) + ((N[0]+n[0]+divisor[0])%divisor[0]) * divisor[1]) * divisor[2];
1522 //*out << Verbose(1) << "Number of comparison cell is " << index << "." << endl;
1523 if (CellList[index] != NULL) { // if there are any atoms in this cell
1524 OtherWalker = CellList[index]->start;
1525 while(OtherWalker->next != CellList[index]->end) { // go through every atom in this cell
1526 OtherWalker = OtherWalker->next;
1527 //*out << Verbose(0) << "Current comparison atom is " << *OtherWalker << "." << endl;
1528 /// \todo periodic check is missing here!
1529 //*out << Verbose(1) << "Checking distance " << OtherWalker->x.PeriodicDistance(&(Walker->x), cell_size) << " against typical bond length of " << bonddistance*bonddistance << "." << endl;
1530 MinDistance = OtherWalker->type->CovalentRadius + Walker->type->CovalentRadius;
1531 MinDistance *= (IsAngstroem) ? 1. : 1./AtomicLengthToAngstroem;
1532 MaxDistance = MinDistance + BONDTHRESHOLD;
1533 MinDistance -= BONDTHRESHOLD;
1534 distance = OtherWalker->x.PeriodicDistance(&(Walker->x), cell_size);
1535 if ((OtherWalker->father->nr > Walker->father->nr) && (distance <= MaxDistance*MaxDistance) && (distance >= MinDistance*MinDistance)) { // create bond if distance is smaller
1536 *out << Verbose(0) << "Adding Bond between " << *Walker << " and " << *OtherWalker << "." << endl;
1537 AddBond(Walker->father, OtherWalker->father, 1); // also increases molecule::BondCount
1538 BondCount++;
1539 } else {
1540 //*out << Verbose(1) << "Not Adding: Wrong label order or distance too great." << endl;
1541 }
1542 }
1543 }
1544 }
1545 }
1546 }
1547 }
1548 // 4. free the cell again
1549 for (int i=NumberCells;i--;)
1550 if (CellList[i] != NULL) {
1551 delete(CellList[i]);
1552 }
1553 Free((void **)&CellList, "molecule::CreateAdjacencyList - ** CellList");
1554
1555 // create the adjacency list per atom
1556 CreateListOfBondsPerAtom(out);
1557
1558 // correct Bond degree of each bond by checking of updated(!) sum of bond degree for an atom match its valence count
1559 // bond degrres are correctled iteratively by one, so that 2-2 instead of 1-3 or 3-1 corrections are favoured: We want
1560 // a rather symmetric distribution of higher bond degrees
1561 if (BondCount != 0) {
1562 NoCyclicBonds = 0;
1563 *out << Verbose(1) << "Correcting Bond degree of each bond ... ";
1564 do {
1565 No = 0; // No acts as breakup flag (if 1 we still continue)
1566 Walker = start;
1567 while (Walker->next != end) { // go through every atom
1568 Walker = Walker->next;
1569 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) { // go through each of its bond partners
1570 OtherWalker = ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
1571 // count valence of first partner (updated!), might have changed during last bond partner
1572 NoBonds = 0;
1573 for(j=0;j<NumberOfBondsPerAtom[Walker->nr];j++)
1574 NoBonds += ListOfBondsPerAtom[Walker->nr][j]->BondDegree;
1575 *out << Verbose(3) << "Walker: " << (int)Walker->type->NoValenceOrbitals << " > " << NoBonds << "?" << endl;
1576 if ((int)(Walker->type->NoValenceOrbitals) > NoBonds) { // we have a mismatch, check NoBonds of other atom
1577 // count valence of second partner
1578 NoBonds = 0;
1579 for(j=0;j<NumberOfBondsPerAtom[OtherWalker->nr];j++)
1580 NoBonds += ListOfBondsPerAtom[OtherWalker->nr][j]->BondDegree;
1581 *out << Verbose(3) << "OtherWalker: " << (int)OtherWalker->type->NoValenceOrbitals << " > " << NoBonds << "?" << endl;
1582 if ((int)(OtherWalker->type->NoValenceOrbitals) > NoBonds) // increase bond degree by one
1583 ListOfBondsPerAtom[Walker->nr][i]->BondDegree++;
1584 }
1585 }
1586 }
1587 } while (No);
1588 *out << " done." << endl;
1589 } else
1590 *out << Verbose(1) << "BondCount is " << BondCount << ", no bonds between any of the " << AtomCount << " atoms." << endl;
1591 *out << Verbose(1) << "I detected " << BondCount << " bonds in the molecule with distance " << bonddistance << "." << endl;
1592
1593 // output bonds for debugging (if bond chain list was correctly installed)
1594 *out << Verbose(1) << endl << "From contents of bond chain list:";
1595 bond *Binder = first;
1596 while(Binder->next != last) {
1597 Binder = Binder->next;
1598 *out << *Binder << "\t" << endl;
1599 }
1600 *out << endl;
1601 } else
1602 *out << Verbose(1) << "AtomCount is " << AtomCount << ", thus no bonds, no connections!." << endl;
1603 *out << Verbose(0) << "End of CreateAdjacencyList." << endl;
1604 Free((void **)&matrix, "molecule::CreateAdjacencyList: *matrix");
1605};
1606
1607/** Performs a Depth-First search on this molecule.
1608 * Marks bonds in molecule as cyclic, bridge, ... and atoms as
1609 * articulations points, ...
1610 * We use the algorithm from [Even, Graph Algorithms, p.62].
1611 * \param *out output stream for debugging
1612 * \param *&MinimumRingSize contains smallest ring size in molecular structure on return or -1 if no rings were found
1613 * \return list of each disconnected subgraph as an individual molecule class structure
1614 */
1615MoleculeLeafClass * molecule::DepthFirstSearchAnalysis(ofstream *out, int *&MinimumRingSize)
1616{
1617 class StackClass<atom *> *AtomStack;
1618 AtomStack = new StackClass<atom *>(AtomCount);
1619 class StackClass<bond *> *BackEdgeStack = new StackClass<bond *> (BondCount);
1620 MoleculeLeafClass *SubGraphs = new MoleculeLeafClass(NULL);
1621 MoleculeLeafClass *LeafWalker = SubGraphs;
1622 int CurrentGraphNr = 0, OldGraphNr;
1623 int ComponentNumber = 0;
1624 atom *Walker = NULL, *OtherAtom = NULL, *Root = start->next;
1625 bond *Binder = NULL;
1626 bool BackStepping = false;
1627
1628 *out << Verbose(0) << "Begin of DepthFirstSearchAnalysis" << endl;
1629
1630 ResetAllBondsToUnused();
1631 ResetAllAtomNumbers();
1632 InitComponentNumbers();
1633 BackEdgeStack->ClearStack();
1634 while (Root != end) { // if there any atoms at all
1635 // (1) mark all edges unused, empty stack, set atom->GraphNr = 0 for all
1636 AtomStack->ClearStack();
1637
1638 // put into new subgraph molecule and add this to list of subgraphs
1639 LeafWalker = new MoleculeLeafClass(LeafWalker);
1640 LeafWalker->Leaf = new molecule(elemente);
1641 LeafWalker->Leaf->AddCopyAtom(Root);
1642
1643 OldGraphNr = CurrentGraphNr;
1644 Walker = Root;
1645 do { // (10)
1646 do { // (2) set number and Lowpoint of Atom to i, increase i, push current atom
1647 if (!BackStepping) { // if we don't just return from (8)
1648 Walker->GraphNr = CurrentGraphNr;
1649 Walker->LowpointNr = CurrentGraphNr;
1650 *out << Verbose(1) << "Setting Walker[" << Walker->Name << "]'s number to " << Walker->GraphNr << " with Lowpoint " << Walker->LowpointNr << "." << endl;
1651 AtomStack->Push(Walker);
1652 CurrentGraphNr++;
1653 }
1654 do { // (3) if Walker has no unused egdes, go to (5)
1655 BackStepping = false; // reset backstepping flag for (8)
1656 if (Binder == NULL) // if we don't just return from (11), Binder is already set to next unused
1657 Binder = FindNextUnused(Walker);
1658 if (Binder == NULL)
1659 break;
1660 *out << Verbose(2) << "Current Unused Bond is " << *Binder << "." << endl;
1661 // (4) Mark Binder used, ...
1662 Binder->MarkUsed(black);
1663 OtherAtom = Binder->GetOtherAtom(Walker);
1664 *out << Verbose(2) << "(4) OtherAtom is " << OtherAtom->Name << "." << endl;
1665 if (OtherAtom->GraphNr != -1) {
1666 // (4a) ... if "other" atom has been visited (GraphNr != 0), set lowpoint to minimum of both, go to (3)
1667 Binder->Type = BackEdge;
1668 BackEdgeStack->Push(Binder);
1669 Walker->LowpointNr = ( Walker->LowpointNr < OtherAtom->GraphNr ) ? Walker->LowpointNr : OtherAtom->GraphNr;
1670 *out << Verbose(3) << "(4a) Visited: Setting Lowpoint of Walker[" << Walker->Name << "] to " << Walker->LowpointNr << "." << endl;
1671 } else {
1672 // (4b) ... otherwise set OtherAtom as Ancestor of Walker and Walker as OtherAtom, go to (2)
1673 Binder->Type = TreeEdge;
1674 OtherAtom->Ancestor = Walker;
1675 Walker = OtherAtom;
1676 *out << Verbose(3) << "(4b) Not Visited: OtherAtom[" << OtherAtom->Name << "]'s Ancestor is now " << OtherAtom->Ancestor->Name << ", Walker is OtherAtom " << OtherAtom->Name << "." << endl;
1677 break;
1678 }
1679 Binder = NULL;
1680 } while (1); // (3)
1681 if (Binder == NULL) {
1682 *out << Verbose(2) << "No more Unused Bonds." << endl;
1683 break;
1684 } else
1685 Binder = NULL;
1686 } while (1); // (2)
1687
1688 // if we came from backstepping, yet there were no more unused bonds, we end up here with no Ancestor, because Walker is Root! Then we are finished!
1689 if ((Walker == Root) && (Binder == NULL))
1690 break;
1691
1692 // (5) if Ancestor of Walker is ...
1693 *out << Verbose(1) << "(5) Number of Walker[" << Walker->Name << "]'s Ancestor[" << Walker->Ancestor->Name << "] is " << Walker->Ancestor->GraphNr << "." << endl;
1694 if (Walker->Ancestor->GraphNr != Root->GraphNr) {
1695 // (6) (Ancestor of Walker is not Root)
1696 if (Walker->LowpointNr < Walker->Ancestor->GraphNr) {
1697 // (6a) set Ancestor's Lowpoint number to minimum of of its Ancestor and itself, go to Step(8)
1698 Walker->Ancestor->LowpointNr = (Walker->Ancestor->LowpointNr < Walker->LowpointNr) ? Walker->Ancestor->LowpointNr : Walker->LowpointNr;
1699 *out << Verbose(2) << "(6) Setting Walker[" << Walker->Name << "]'s Ancestor[" << Walker->Ancestor->Name << "]'s Lowpoint to " << Walker->Ancestor->LowpointNr << "." << endl;
1700 } else {
1701 // (7) (Ancestor of Walker is a separating vertex, remove all from stack till Walker (including), these and Ancestor form a component
1702 Walker->Ancestor->SeparationVertex = true;
1703 *out << Verbose(2) << "(7) Walker[" << Walker->Name << "]'s Ancestor[" << Walker->Ancestor->Name << "]'s is a separating vertex, creating component." << endl;
1704 SetNextComponentNumber(Walker->Ancestor, ComponentNumber);
1705 *out << Verbose(3) << "(7) Walker[" << Walker->Name << "]'s Ancestor's Compont is " << ComponentNumber << "." << endl;
1706 SetNextComponentNumber(Walker, ComponentNumber);
1707 *out << Verbose(3) << "(7) Walker[" << Walker->Name << "]'s Compont is " << ComponentNumber << "." << endl;
1708 do {
1709 OtherAtom = AtomStack->PopLast();
1710 LeafWalker->Leaf->AddCopyAtom(OtherAtom);
1711 SetNextComponentNumber(OtherAtom, ComponentNumber);
1712 *out << Verbose(3) << "(7) Other[" << OtherAtom->Name << "]'s Compont is " << ComponentNumber << "." << endl;
1713 } while (OtherAtom != Walker);
1714 ComponentNumber++;
1715 }
1716 // (8) Walker becomes its Ancestor, go to (3)
1717 *out << Verbose(2) << "(8) Walker[" << Walker->Name << "] is now its Ancestor " << Walker->Ancestor->Name << ", backstepping. " << endl;
1718 Walker = Walker->Ancestor;
1719 BackStepping = true;
1720 }
1721 if (!BackStepping) { // coming from (8) want to go to (3)
1722 // (9) remove all from stack till Walker (including), these and Root form a component
1723 AtomStack->Output(out);
1724 SetNextComponentNumber(Root, ComponentNumber);
1725 *out << Verbose(3) << "(9) Root[" << Root->Name << "]'s Component is " << ComponentNumber << "." << endl;
1726 SetNextComponentNumber(Walker, ComponentNumber);
1727 *out << Verbose(3) << "(9) Walker[" << Walker->Name << "]'s Component is " << ComponentNumber << "." << endl;
1728 do {
1729 OtherAtom = AtomStack->PopLast();
1730 LeafWalker->Leaf->AddCopyAtom(OtherAtom);
1731 SetNextComponentNumber(OtherAtom, ComponentNumber);
1732 *out << Verbose(3) << "(7) Other[" << OtherAtom->Name << "]'s Compont is " << ComponentNumber << "." << endl;
1733 } while (OtherAtom != Walker);
1734 ComponentNumber++;
1735
1736 // (11) Root is separation vertex, set Walker to Root and go to (4)
1737 Walker = Root;
1738 Binder = FindNextUnused(Walker);
1739 *out << Verbose(1) << "(10) Walker is Root[" << Root->Name << "], next Unused Bond is " << Binder << "." << endl;
1740 if (Binder != NULL) { // Root is separation vertex
1741 *out << Verbose(1) << "(11) Root is a separation vertex." << endl;
1742 Walker->SeparationVertex = true;
1743 }
1744 }
1745 } while ((BackStepping) || (Binder != NULL)); // (10) halt only if Root has no unused edges
1746
1747 // From OldGraphNr to CurrentGraphNr ranges an disconnected subgraph
1748 *out << Verbose(0) << "Disconnected subgraph ranges from " << OldGraphNr << " to " << CurrentGraphNr << "." << endl;
1749 LeafWalker->Leaf->Output(out);
1750 *out << endl;
1751
1752 // step on to next root
1753 while ((Root != end) && (Root->GraphNr != -1)) {
1754 //*out << Verbose(1) << "Current next subgraph root candidate is " << Root->Name << "." << endl;
1755 if (Root->GraphNr != -1) // if already discovered, step on
1756 Root = Root->next;
1757 }
1758 }
1759 // set cyclic bond criterium on "same LP" basis
1760 Binder = first;
1761 while(Binder->next != last) {
1762 Binder = Binder->next;
1763 if (Binder->rightatom->LowpointNr == Binder->leftatom->LowpointNr) { // cyclic ??
1764 Binder->Cyclic = true;
1765 NoCyclicBonds++;
1766 }
1767 }
1768
1769 // analysis of the cycles (print rings, get minimum cycle length)
1770 CyclicStructureAnalysis(out, BackEdgeStack, MinimumRingSize);
1771
1772 *out << Verbose(1) << "Final graph info for each atom is:" << endl;
1773 Walker = start;
1774 while (Walker->next != end) {
1775 Walker = Walker->next;
1776 *out << Verbose(2) << "Atom " << Walker->Name << " is " << ((Walker->SeparationVertex) ? "a" : "not a") << " separation vertex, components are ";
1777 OutputComponentNumber(out, Walker);
1778 *out << " with Lowpoint " << Walker->LowpointNr << " and Graph Nr. " << Walker->GraphNr << "." << endl;
1779 }
1780
1781 *out << Verbose(1) << "Final graph info for each bond is:" << endl;
1782 Binder = first;
1783 while(Binder->next != last) {
1784 Binder = Binder->next;
1785 *out << Verbose(2) << ((Binder->Type == TreeEdge) ? "TreeEdge " : "BackEdge ") << *Binder << ": <";
1786 *out << ((Binder->leftatom->SeparationVertex) ? "SP," : "") << "L" << Binder->leftatom->LowpointNr << " G" << Binder->leftatom->GraphNr << " Comp.";
1787 OutputComponentNumber(out, Binder->leftatom);
1788 *out << " === ";
1789 *out << ((Binder->rightatom->SeparationVertex) ? "SP," : "") << "L" << Binder->rightatom->LowpointNr << " G" << Binder->rightatom->GraphNr << " Comp.";
1790 OutputComponentNumber(out, Binder->rightatom);
1791 *out << ">." << endl;
1792 if (Binder->Cyclic) // cyclic ??
1793 *out << Verbose(3) << "Lowpoint at each side are equal: CYCLIC!" << endl;
1794 }
1795
1796 // free all and exit
1797 delete(AtomStack);
1798 *out << Verbose(0) << "End of DepthFirstSearchAnalysis" << endl;
1799 return SubGraphs;
1800};
1801
1802/** Analyses the cycles found and returns minimum of all cycle lengths.
1803 * \param *out output stream for debugging
1804 * \param *BackEdgeStack stack with all back edges found during DFS scan
1805 * \param *&MinimumRingSize contains smallest ring size in molecular structure on return or -1 if no rings were found, if set is maximum search distance
1806 * \todo BFS from the not-same-LP to find back to starting point of tributary cycle over more than one bond
1807 */
1808void molecule::CyclicStructureAnalysis(ofstream *out, class StackClass<bond *> * BackEdgeStack, int *&MinimumRingSize)
1809{
1810 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::CyclicStructureAnalysis: **PredecessorList");
1811 int *ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CyclicStructureAnalysis: *ShortestPathList");
1812 enum Shading *ColorList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::CyclicStructureAnalysis: *ColorList");
1813 class StackClass<atom *> *BFSStack = new StackClass<atom *> (AtomCount); // will hold the current ring
1814 class StackClass<atom *> *TouchedStack = new StackClass<atom *> (AtomCount); // contains all "touched" atoms
1815 atom *Walker = NULL, *OtherAtom = NULL, *Root = NULL;
1816 bond *Binder = NULL, *BackEdge = NULL;
1817 int RingSize, NumCycles, MinRingSize = -1;
1818
1819 // initialise each vertex as white with no predecessor, empty queue, color Root lightgray
1820 for (int i=AtomCount;i--;) {
1821 PredecessorList[i] = NULL;
1822 ShortestPathList[i] = -1;
1823 ColorList[i] = white;
1824 }
1825 MinimumRingSize = new int[AtomCount];
1826 for(int i=AtomCount;i--;)
1827 MinimumRingSize[i] = AtomCount;
1828
1829
1830 *out << Verbose(1) << "Back edge list - ";
1831 BackEdgeStack->Output(out);
1832
1833 *out << Verbose(1) << "Analysing cycles ... " << endl;
1834 NumCycles = 0;
1835 while (!BackEdgeStack->IsEmpty()) {
1836 BackEdge = BackEdgeStack->PopFirst();
1837 // this is the target
1838 Root = BackEdge->leftatom;
1839 // this is the source point
1840 Walker = BackEdge->rightatom;
1841 ShortestPathList[Walker->nr] = 0;
1842 BFSStack->ClearStack(); // start with empty BFS stack
1843 BFSStack->Push(Walker);
1844 TouchedStack->Push(Walker);
1845 //*out << Verbose(1) << "---------------------------------------------------------------------------------------------------------" << endl;
1846 OtherAtom = NULL;
1847 while ((Walker != Root) && ((OtherAtom == NULL) || (ShortestPathList[OtherAtom->nr] < MinimumRingSize[Walker->GetTrueFather()->nr]))) { // look for Root
1848 Walker = BFSStack->PopFirst();
1849 //*out << Verbose(2) << "Current Walker is " << *Walker << ", we look for SP to Root " << *Root << "." << endl;
1850 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
1851 Binder = ListOfBondsPerAtom[Walker->nr][i];
1852 if (Binder != BackEdge) { // only walk along DFS spanning tree (otherwise we always find SP of one being backedge Binder)
1853 OtherAtom = Binder->GetOtherAtom(Walker);
1854 //*out << Verbose(2) << "Current OtherAtom is: " << OtherAtom->Name << " for bond " << *Binder << "." << endl;
1855 if (ColorList[OtherAtom->nr] == white) {
1856 TouchedStack->Push(OtherAtom);
1857 ColorList[OtherAtom->nr] = lightgray;
1858 PredecessorList[OtherAtom->nr] = Walker; // Walker is the predecessor
1859 ShortestPathList[OtherAtom->nr] = ShortestPathList[Walker->nr]+1;
1860 //*out << Verbose(2) << "Coloring OtherAtom " << OtherAtom->Name << " lightgray, its predecessor is " << Walker->Name << " and its Shortest Path is " << ShortestPathList[OtherAtom->nr] << " egde(s) long." << endl;
1861 if (ShortestPathList[OtherAtom->nr] < MinimumRingSize[Walker->GetTrueFather()->nr]) { // Check for maximum distance
1862 //*out << Verbose(3) << "Putting OtherAtom into queue." << endl;
1863 BFSStack->Push(OtherAtom);
1864 }
1865 } else {
1866 //*out << Verbose(3) << "Not Adding, has already been visited." << endl;
1867 }
1868 } else {
1869 //*out << Verbose(3) << "Not Visiting, is a back edge." << endl;
1870 }
1871 }
1872 ColorList[Walker->nr] = black;
1873 //*out << Verbose(1) << "Coloring Walker " << Walker->Name << " black." << endl;
1874 }
1875
1876 if (Walker == Root) {
1877 // now climb back the predecessor list and thus find the cycle members
1878 NumCycles++;
1879 RingSize = 1;
1880 Root->GetTrueFather()->IsCyclic = true;
1881 *out << Verbose(1) << "Found ring contains: ";
1882 while (Walker != BackEdge->rightatom) {
1883 *out << Walker->Name << " <-> ";
1884 Walker = PredecessorList[Walker->nr];
1885 Walker->GetTrueFather()->IsCyclic = true;
1886 RingSize++;
1887 }
1888 *out << Walker->Name << " with a length of " << RingSize << "." << endl << endl;
1889 // walk through all and set MinimumRingSize
1890 Walker = Root;
1891 MinimumRingSize[Walker->GetTrueFather()->nr] = RingSize;
1892 while (Walker != BackEdge->rightatom) {
1893 Walker = PredecessorList[Walker->nr];
1894 if (RingSize < MinimumRingSize[Walker->GetTrueFather()->nr])
1895 MinimumRingSize[Walker->GetTrueFather()->nr] = RingSize;
1896 }
1897 if ((RingSize < MinRingSize) || (MinRingSize == -1))
1898 MinRingSize = RingSize;
1899 } else {
1900 *out << Verbose(1) << "No ring containing " << *Root << " with length equal to or smaller than " << MinimumRingSize[Walker->GetTrueFather()->nr] << " found." << endl;
1901 }
1902
1903 // now clean the lists
1904 while (!TouchedStack->IsEmpty()){
1905 Walker = TouchedStack->PopFirst();
1906 PredecessorList[Walker->nr] = NULL;
1907 ShortestPathList[Walker->nr] = -1;
1908 ColorList[Walker->nr] = white;
1909 }
1910 }
1911 if (MinRingSize != -1) {
1912 // go over all atoms
1913 Root = start;
1914 while(Root->next != end) {
1915 Root = Root->next;
1916
1917 if (MinimumRingSize[Root->GetTrueFather()->nr] == AtomCount) { // check whether MinimumRingSize is set, if not BFS to next where it is
1918 Walker = Root;
1919 ShortestPathList[Walker->nr] = 0;
1920 BFSStack->ClearStack(); // start with empty BFS stack
1921 BFSStack->Push(Walker);
1922 TouchedStack->Push(Walker);
1923 //*out << Verbose(1) << "---------------------------------------------------------------------------------------------------------" << endl;
1924 OtherAtom = Walker;
1925 while (OtherAtom != NULL) { // look for Root
1926 Walker = BFSStack->PopFirst();
1927 //*out << Verbose(2) << "Current Walker is " << *Walker << ", we look for SP to Root " << *Root << "." << endl;
1928 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
1929 Binder = ListOfBondsPerAtom[Walker->nr][i];
1930 if ((Binder != BackEdge) || (NumberOfBondsPerAtom[Walker->nr] == 1)) { // only walk along DFS spanning tree (otherwise we always find SP of 1 being backedge Binder), but terminal hydrogens may be connected via backedge, hence extra check
1931 OtherAtom = Binder->GetOtherAtom(Walker);
1932 //*out << Verbose(2) << "Current OtherAtom is: " << OtherAtom->Name << " for bond " << *Binder << "." << endl;
1933 if (ColorList[OtherAtom->nr] == white) {
1934 TouchedStack->Push(OtherAtom);
1935 ColorList[OtherAtom->nr] = lightgray;
1936 PredecessorList[OtherAtom->nr] = Walker; // Walker is the predecessor
1937 ShortestPathList[OtherAtom->nr] = ShortestPathList[Walker->nr]+1;
1938 //*out << Verbose(2) << "Coloring OtherAtom " << OtherAtom->Name << " lightgray, its predecessor is " << Walker->Name << " and its Shortest Path is " << ShortestPathList[OtherAtom->nr] << " egde(s) long." << endl;
1939 if (OtherAtom->GetTrueFather()->IsCyclic) { // if the other atom is connected to a ring
1940 MinimumRingSize[Root->GetTrueFather()->nr] = ShortestPathList[OtherAtom->nr]+MinimumRingSize[OtherAtom->GetTrueFather()->nr];
1941 OtherAtom = NULL; //break;
1942 break;
1943 } else
1944 BFSStack->Push(OtherAtom);
1945 } else {
1946 //*out << Verbose(3) << "Not Adding, has already been visited." << endl;
1947 }
1948 } else {
1949 //*out << Verbose(3) << "Not Visiting, is a back edge." << endl;
1950 }
1951 }
1952 ColorList[Walker->nr] = black;
1953 //*out << Verbose(1) << "Coloring Walker " << Walker->Name << " black." << endl;
1954 }
1955
1956 // now clean the lists
1957 while (!TouchedStack->IsEmpty()){
1958 Walker = TouchedStack->PopFirst();
1959 PredecessorList[Walker->nr] = NULL;
1960 ShortestPathList[Walker->nr] = -1;
1961 ColorList[Walker->nr] = white;
1962 }
1963 }
1964 *out << Verbose(1) << "Minimum ring size of " << *Root << " is " << MinimumRingSize[Root->GetTrueFather()->nr] << "." << endl;
1965 }
1966 *out << Verbose(1) << "Minimum ring size is " << MinRingSize << ", over " << NumCycles << " cycles total." << endl;
1967 } else
1968 *out << Verbose(1) << "No rings were detected in the molecular structure." << endl;
1969
1970 Free((void **)&PredecessorList, "molecule::CyclicStructureAnalysis: **PredecessorList");
1971 Free((void **)&ShortestPathList, "molecule::CyclicStructureAnalysis: **ShortestPathList");
1972 Free((void **)&ColorList, "molecule::CyclicStructureAnalysis: **ColorList");
1973 delete(BFSStack);
1974};
1975
1976/** Sets the next component number.
1977 * This is O(N) as the number of bonds per atom is bound.
1978 * \param *vertex atom whose next atom::*ComponentNr is to be set
1979 * \param nr number to use
1980 */
1981void molecule::SetNextComponentNumber(atom *vertex, int nr)
1982{
1983 int i=0;
1984 if (vertex != NULL) {
1985 for(;i<NumberOfBondsPerAtom[vertex->nr];i++) {
1986 if (vertex->ComponentNr[i] == -1) { // check if not yet used
1987 vertex->ComponentNr[i] = nr;
1988 break;
1989 }
1990 else if (vertex->ComponentNr[i] == nr) // if number is already present, don't add another time
1991 break; // breaking here will not cause error!
1992 }
1993 if (i == NumberOfBondsPerAtom[vertex->nr])
1994 cerr << "Error: All Component entries are already occupied!" << endl;
1995 } else
1996 cerr << "Error: Given vertex is NULL!" << endl;
1997};
1998
1999/** Output a list of flags, stating whether the bond was visited or not.
2000 * \param *out output stream for debugging
2001 */
2002void molecule::OutputComponentNumber(ofstream *out, atom *vertex)
2003{
2004 for(int i=0;i<NumberOfBondsPerAtom[vertex->nr];i++)
2005 *out << vertex->ComponentNr[i] << " ";
2006};
2007
2008/** Allocates memory for all atom::*ComponentNr in this molecule and sets each entry to -1.
2009 */
2010void molecule::InitComponentNumbers()
2011{
2012 atom *Walker = start;
2013 while(Walker->next != end) {
2014 Walker = Walker->next;
2015 if (Walker->ComponentNr != NULL)
2016 Free((void **)&Walker->ComponentNr, "molecule::InitComponentNumbers: **Walker->ComponentNr");
2017 Walker->ComponentNr = (int *) Malloc(sizeof(int)*NumberOfBondsPerAtom[Walker->nr], "molecule::InitComponentNumbers: *Walker->ComponentNr");
2018 for (int i=NumberOfBondsPerAtom[Walker->nr];i--;)
2019 Walker->ComponentNr[i] = -1;
2020 }
2021};
2022
2023/** Returns next unused bond for this atom \a *vertex or NULL of none exists.
2024 * \param *vertex atom to regard
2025 * \return bond class or NULL
2026 */
2027bond * molecule::FindNextUnused(atom *vertex)
2028{
2029 for(int i=0;i<NumberOfBondsPerAtom[vertex->nr];i++)
2030 if (ListOfBondsPerAtom[vertex->nr][i]->IsUsed() == white)
2031 return(ListOfBondsPerAtom[vertex->nr][i]);
2032 return NULL;
2033};
2034
2035/** Resets bond::Used flag of all bonds in this molecule.
2036 * \return true - success, false - -failure
2037 */
2038void molecule::ResetAllBondsToUnused()
2039{
2040 bond *Binder = first;
2041 while (Binder->next != last) {
2042 Binder = Binder->next;
2043 Binder->ResetUsed();
2044 }
2045};
2046
2047/** Resets atom::nr to -1 of all atoms in this molecule.
2048 */
2049void molecule::ResetAllAtomNumbers()
2050{
2051 atom *Walker = start;
2052 while (Walker->next != end) {
2053 Walker = Walker->next;
2054 Walker->GraphNr = -1;
2055 }
2056};
2057
2058/** Output a list of flags, stating whether the bond was visited or not.
2059 * \param *out output stream for debugging
2060 * \param *list
2061 */
2062void OutputAlreadyVisited(ofstream *out, int *list)
2063{
2064 *out << Verbose(4) << "Already Visited Bonds:\t";
2065 for(int i=1;i<=list[0];i++) *out << Verbose(0) << list[i] << " ";
2066 *out << endl;
2067};
2068
2069/** Estimates by educated guessing (using upper limit) the expected number of fragments.
2070 * The upper limit is
2071 * \f[
2072 * n = N \cdot C^k
2073 * \f]
2074 * where \f$C=2^c\f$ and c is the maximum bond degree over N number of atoms.
2075 * \param *out output stream for debugging
2076 * \param order bond order k
2077 * \return number n of fragments
2078 */
2079int molecule::GuesstimateFragmentCount(ofstream *out, int order)
2080{
2081 int c = 0;
2082 int FragmentCount;
2083 // get maximum bond degree
2084 atom *Walker = start;
2085 while (Walker->next != end) {
2086 Walker = Walker->next;
2087 c = (NumberOfBondsPerAtom[Walker->nr] > c) ? NumberOfBondsPerAtom[Walker->nr] : c;
2088 }
2089 FragmentCount = NoNonHydrogen*(1 << (c*order));
2090 *out << Verbose(1) << "Upper limit for this subgraph is " << FragmentCount << " for " << NoNonHydrogen << " non-H atoms with maximum bond degree of " << c << "." << endl;
2091 return FragmentCount;
2092};
2093
2094/** Scans a single line for number and puts them into \a KeySet.
2095 * \param *out output stream for debugging
2096 * \param *buffer buffer to scan
2097 * \param &CurrentSet filled KeySet on return
2098 * \return true - at least one valid atom id parsed, false - CurrentSet is empty
2099 */
2100bool molecule::ScanBufferIntoKeySet(ofstream *out, char *buffer, KeySet &CurrentSet)
2101{
2102 stringstream line;
2103 int AtomNr;
2104 int status = 0;
2105
2106 line.str(buffer);
2107 while (!line.eof()) {
2108 line >> AtomNr;
2109 if ((AtomNr >= 0) && (AtomNr < AtomCount)) {
2110 CurrentSet.insert(AtomNr); // insert at end, hence in same order as in file!
2111 status++;
2112 } // else it's "-1" or else and thus must not be added
2113 }
2114 *out << Verbose(1) << "The scanned KeySet is ";
2115 for(KeySet::iterator runner = CurrentSet.begin(); runner != CurrentSet.end(); runner++) {
2116 *out << (*runner) << "\t";
2117 }
2118 *out << endl;
2119 return (status != 0);
2120};
2121
2122/** Parses the KeySet file and fills \a *FragmentList from the known molecule structure.
2123 * Does two-pass scanning:
2124 * -# Scans the keyset file and initialises a temporary graph
2125 * -# Scans TEFactors file and sets the TEFactor of each key set in the temporary graph accordingly
2126 * Finally, the temporary graph is inserted into the given \a FragmentList for return.
2127 * \param *out output stream for debugging
2128 * \param *path path to file
2129 * \param *FragmentList empty, filled on return
2130 * \return true - parsing successfully, false - failure on parsing (FragmentList will be NULL)
2131 */
2132bool molecule::ParseKeySetFile(ofstream *out, char *path, Graph *&FragmentList)
2133{
2134 bool status = true;
2135 ifstream InputFile;
2136 stringstream line;
2137 GraphTestPair testGraphInsert;
2138 int NumberOfFragments = 0;
2139 double TEFactor;
2140 char *filename = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::ParseKeySetFile - filename");
2141
2142 if (FragmentList == NULL) { // check list pointer
2143 FragmentList = new Graph;
2144 }
2145
2146 // 1st pass: open file and read
2147 *out << Verbose(1) << "Parsing the KeySet file ... " << endl;
2148 sprintf(filename, "%s/%s%s", path, FRAGMENTPREFIX, KEYSETFILE);
2149 InputFile.open(filename);
2150 if (InputFile != NULL) {
2151 // each line represents a new fragment
2152 char *buffer = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::ParseKeySetFile - *buffer");
2153 // 1. parse keysets and insert into temp. graph
2154 while (!InputFile.eof()) {
2155 InputFile.getline(buffer, MAXSTRINGSIZE);
2156 KeySet CurrentSet;
2157 if ((strlen(buffer) > 0) && (ScanBufferIntoKeySet(out, buffer, CurrentSet))) { // if at least one valid atom was added, write config
2158 testGraphInsert = FragmentList->insert(GraphPair (CurrentSet,pair<int,double>(NumberOfFragments++,1))); // store fragment number and current factor
2159 if (!testGraphInsert.second) {
2160 cerr << "KeySet file must be corrupt as there are two equal key sets therein!" << endl;
2161 }
2162 //FragmentList->ListOfMolecules[NumberOfFragments++] = StoreFragmentFromKeySet(out, CurrentSet, IsAngstroem);
2163 }
2164 }
2165 // 2. Free and done
2166 InputFile.close();
2167 InputFile.clear();
2168 Free((void **)&buffer, "molecule::ParseKeySetFile - *buffer");
2169 *out << Verbose(1) << "done." << endl;
2170 } else {
2171 *out << Verbose(1) << "File " << filename << " not found." << endl;
2172 status = false;
2173 }
2174
2175 // 2nd pass: open TEFactors file and read
2176 *out << Verbose(1) << "Parsing the TEFactors file ... " << endl;
2177 sprintf(filename, "%s/%s%s", path, FRAGMENTPREFIX, TEFACTORSFILE);
2178 InputFile.open(filename);
2179 if (InputFile != NULL) {
2180 // 3. add found TEFactors to each keyset
2181 NumberOfFragments = 0;
2182 for(Graph::iterator runner = FragmentList->begin();runner != FragmentList->end(); runner++) {
2183 if (!InputFile.eof()) {
2184 InputFile >> TEFactor;
2185 (*runner).second.second = TEFactor;
2186 *out << Verbose(2) << "Setting " << ++NumberOfFragments << " fragment's TEFactor to " << (*runner).second.second << "." << endl;
2187 } else {
2188 status = false;
2189 break;
2190 }
2191 }
2192 // 4. Free and done
2193 InputFile.close();
2194 *out << Verbose(1) << "done." << endl;
2195 } else {
2196 *out << Verbose(1) << "File " << filename << " not found." << endl;
2197 status = false;
2198 }
2199
2200 // free memory
2201 Free((void **)&filename, "molecule::ParseKeySetFile - filename");
2202
2203 return status;
2204};
2205
2206/** Stores keysets and TEFactors to file.
2207 * \param *out output stream for debugging
2208 * \param KeySetList Graph with Keysets and factors
2209 * \param *path path to file
2210 * \return true - file written successfully, false - writing failed
2211 */
2212bool molecule::StoreKeySetFile(ofstream *out, Graph &KeySetList, char *path)
2213{
2214 ofstream output;
2215 bool status = true;
2216 string line;
2217
2218 // open KeySet file
2219 line = path;
2220 line.append("/");
2221 line += FRAGMENTPREFIX;
2222 line += KEYSETFILE;
2223 output.open(line.c_str(), ios::out);
2224 *out << Verbose(1) << "Saving key sets of the total graph ... ";
2225 if(output != NULL) {
2226 for(Graph::iterator runner = KeySetList.begin(); runner != KeySetList.end(); runner++) {
2227 for (KeySet::iterator sprinter = (*runner).first.begin();sprinter != (*runner).first.end(); sprinter++) {
2228 if (sprinter != (*runner).first.begin())
2229 output << "\t";
2230 output << *sprinter;
2231 }
2232 output << endl;
2233 }
2234 *out << "done." << endl;
2235 } else {
2236 cerr << "Unable to open " << line << " for writing keysets!" << endl;
2237 status = false;
2238 }
2239 output.close();
2240 output.clear();
2241
2242 // open TEFactors file
2243 line = path;
2244 line.append("/");
2245 line += FRAGMENTPREFIX;
2246 line += TEFACTORSFILE;
2247 output.open(line.c_str(), ios::out);
2248 *out << Verbose(1) << "Saving TEFactors of the total graph ... ";
2249 if(output != NULL) {
2250 for(Graph::iterator runner = KeySetList.begin(); runner != KeySetList.end(); runner++)
2251 output << (*runner).second.second << endl;
2252 *out << Verbose(1) << "done." << endl;
2253 } else {
2254 *out << Verbose(1) << "failed to open " << line << "." << endl;
2255 status = false;
2256 }
2257 output.close();
2258
2259 return status;
2260};
2261
2262/** Storing the bond structure of a molecule to file.
2263 * Simply stores Atom::nr and then the Atom::nr of all bond partners per line.
2264 * \param *out output stream for debugging
2265 * \param *path path to file
2266 * \return true - file written successfully, false - writing failed
2267 */
2268bool molecule::StoreAdjacencyToFile(ofstream *out, char *path)
2269{
2270 ofstream AdjacencyFile;
2271 atom *Walker = NULL;
2272 stringstream line;
2273 bool status = true;
2274
2275 line << path << "/" << FRAGMENTPREFIX << ADJACENCYFILE;
2276 AdjacencyFile.open(line.str().c_str(), ios::out);
2277 *out << Verbose(1) << "Saving adjacency list ... ";
2278 if (AdjacencyFile != NULL) {
2279 Walker = start;
2280 while(Walker->next != end) {
2281 Walker = Walker->next;
2282 AdjacencyFile << Walker->nr << "\t";
2283 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++)
2284 AdjacencyFile << ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker)->nr << "\t";
2285 AdjacencyFile << endl;
2286 }
2287 AdjacencyFile.close();
2288 *out << Verbose(1) << "done." << endl;
2289 } else {
2290 *out << Verbose(1) << "failed to open file " << line.str() << "." << endl;
2291 status = false;
2292 }
2293
2294 return status;
2295};
2296
2297/** Checks contents of adjacency file against bond structure in structure molecule.
2298 * \param *out output stream for debugging
2299 * \param *path path to file
2300 * \param **ListOfAtoms allocated (molecule::AtomCount) and filled lookup table for ids (Atom::nr) to *Atom
2301 * \return true - structure is equal, false - not equivalence
2302 */
2303bool molecule::CheckAdjacencyFileAgainstMolecule(ofstream *out, char *path, atom **ListOfAtoms)
2304{
2305 ifstream File;
2306 stringstream filename;
2307 bool status = true;
2308 char *buffer = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::CheckAdjacencyFileAgainstMolecule: *buffer");
2309
2310 filename << path << "/" << FRAGMENTPREFIX << ADJACENCYFILE;
2311 File.open(filename.str().c_str(), ios::out);
2312 *out << Verbose(1) << "Looking at bond structure stored in adjacency file and comparing to present one ... ";
2313 if (File != NULL) {
2314 // allocate storage structure
2315 int NonMatchNumber = 0; // will number of atoms with differing bond structure
2316 int *CurrentBonds = (int *) Malloc(sizeof(int)*8, "molecule::CheckAdjacencyFileAgainstMolecule - CurrentBonds"); // contains parsed bonds of current atom
2317 int CurrentBondsOfAtom;
2318
2319 // Parse the file line by line and count the bonds
2320 while (!File.eof()) {
2321 File.getline(buffer, MAXSTRINGSIZE);
2322 stringstream line;
2323 line.str(buffer);
2324 int AtomNr = -1;
2325 line >> AtomNr;
2326 CurrentBondsOfAtom = -1; // we count one too far due to line end
2327 // parse into structure
2328 if ((AtomNr >= 0) && (AtomNr < AtomCount)) {
2329 while (!line.eof())
2330 line >> CurrentBonds[ ++CurrentBondsOfAtom ];
2331 // compare against present bonds
2332 //cout << Verbose(2) << "Walker is " << *Walker << ", bond partners: ";
2333 if (CurrentBondsOfAtom == NumberOfBondsPerAtom[AtomNr]) {
2334 for(int i=0;i<NumberOfBondsPerAtom[AtomNr];i++) {
2335 int id = ListOfBondsPerAtom[AtomNr][i]->GetOtherAtom(ListOfAtoms[AtomNr])->nr;
2336 int j = 0;
2337 for (;(j<CurrentBondsOfAtom) && (CurrentBonds[j++] != id);); // check against all parsed bonds
2338 if (CurrentBonds[j-1] != id) { // no match ? Then mark in ListOfAtoms
2339 ListOfAtoms[AtomNr] = NULL;
2340 NonMatchNumber++;
2341 status = false;
2342 //out << "[" << id << "]\t";
2343 } else {
2344 //out << id << "\t";
2345 }
2346 }
2347 //out << endl;
2348 } else {
2349 *out << "Number of bonds for Atom " << *ListOfAtoms[AtomNr] << " does not match, parsed " << CurrentBondsOfAtom << " against " << NumberOfBondsPerAtom[AtomNr] << "." << endl;
2350 status = false;
2351 }
2352 }
2353 }
2354 File.close();
2355 File.clear();
2356 if (status) { // if equal we parse the KeySetFile
2357 *out << Verbose(1) << "done: Equal." << endl;
2358 status = true;
2359 } else
2360 *out << Verbose(1) << "done: Not equal by " << NonMatchNumber << " atoms." << endl;
2361 Free((void **)&CurrentBonds, "molecule::CheckAdjacencyFileAgainstMolecule - **CurrentBonds");
2362 } else {
2363 *out << Verbose(1) << "Adjacency file not found." << endl;
2364 status = false;
2365 }
2366 *out << endl;
2367 Free((void **)&buffer, "molecule::CheckAdjacencyFileAgainstMolecule: *buffer");
2368
2369 return status;
2370};
2371
2372/** Checks whether the OrderAtSite is still below \a Order at some site.
2373 * \param *out output stream for debugging
2374 * \param *AtomMask defines true/false per global Atom::nr to mask in/out each nuclear site, used to activate given number of site to increment order adaptively
2375 * \param *GlobalKeySetList list of keysets with global ids (valid in "this" molecule) needed for adaptive increase
2376 * \param Order desired Order if positive, desired exponent in threshold criteria if negative (0 is single-step)
2377 * \param *MinimumRingSize array of max. possible order to avoid loops
2378 * \param *path path to ENERGYPERFRAGMENT file (may be NULL if Order is non-negative)
2379 * \return true - needs further fragmentation, false - does not need fragmentation
2380 */
2381bool molecule::CheckOrderAtSite(ofstream *out, bool *AtomMask, Graph *GlobalKeySetList, int Order, int *MinimumRingSize, char *path)
2382{
2383 atom *Walker = start;
2384 bool status = false;
2385 ifstream InputFile;
2386
2387 // initialize mask list
2388 for(int i=AtomCount;i--;)
2389 AtomMask[i] = false;
2390
2391 if (Order < 0) { // adaptive increase of BondOrder per site
2392 if (AtomMask[AtomCount] == true) // break after one step
2393 return false;
2394 // parse the EnergyPerFragment file
2395 char *buffer = (char *) Malloc(sizeof(char)*MAXSTRINGSIZE, "molecule::CheckOrderAtSite: *buffer");
2396 sprintf(buffer, "%s/%s%s.dat", path, FRAGMENTPREFIX, ENERGYPERFRAGMENT);
2397 InputFile.open(buffer, ios::in);
2398 if ((InputFile != NULL) && (GlobalKeySetList != NULL)) {
2399 // transmorph graph keyset list into indexed KeySetList
2400 map<int,KeySet> IndexKeySetList;
2401 for(Graph::iterator runner = GlobalKeySetList->begin(); runner != GlobalKeySetList->end(); runner++) {
2402 IndexKeySetList.insert( pair<int,KeySet>(runner->second.first,runner->first) );
2403 }
2404 int lines = 0;
2405 // count the number of lines, i.e. the number of fragments
2406 InputFile.getline(buffer, MAXSTRINGSIZE); // skip comment lines
2407 InputFile.getline(buffer, MAXSTRINGSIZE);
2408 while(!InputFile.eof()) {
2409 InputFile.getline(buffer, MAXSTRINGSIZE);
2410 lines++;
2411 }
2412 *out << Verbose(2) << "Scanned " << lines-1 << " lines." << endl; // one endline too much
2413 InputFile.clear();
2414 InputFile.seekg(ios::beg);
2415 map<int, pair<double,int> > AdaptiveCriteriaList; // (Root No., (Value, Order)) !
2416 int No, FragOrder;
2417 double Value;
2418 // each line represents a fragment root (Atom::nr) id and its energy contribution
2419 InputFile.getline(buffer, MAXSTRINGSIZE); // skip comment lines
2420 InputFile.getline(buffer, MAXSTRINGSIZE);
2421 while(!InputFile.eof()) {
2422 InputFile.getline(buffer, MAXSTRINGSIZE);
2423 if (strlen(buffer) > 2) {
2424 //*out << Verbose(2) << "Scanning: " << buffer;
2425 stringstream line(buffer);
2426 line >> FragOrder;
2427 line >> ws >> No;
2428 line >> ws >> Value; // skip time entry
2429 line >> ws >> Value;
2430 No -= 1; // indices start at 1 in file, not 0
2431 //*out << Verbose(2) << " - yields (" << No << "," << Value << ")" << endl;
2432
2433 // clean the list of those entries that have been superceded by higher order terms already
2434 map<int,KeySet>::iterator marker = IndexKeySetList.find(No); // find keyset to Frag No.
2435 if (marker != IndexKeySetList.end()) { // if found
2436 // as the smallest number in each set has always been the root (we use global id to keep the doubles away), seek smallest and insert into AtomMask
2437 pair <map<int, pair<double,int> >::iterator, bool> InsertedElement = AdaptiveCriteriaList.insert( make_pair(*((*marker).second.begin()), pair<double,int>( Value, Order) ));
2438 map<int, pair<double,int> >::iterator PresentItem = InsertedElement.first;
2439 if (!InsertedElement.second) { // this root is already present
2440 if ((*PresentItem).second.second < FragOrder) // if order there is lower, update entry with higher-order term
2441 //if ((*PresentItem).second.first < (*runner).first) // as higher-order terms are not always better, we skip this part (which would always include this site into adaptive increase)
2442 { // if value is smaller, update value and order
2443 (*PresentItem).second.first = Value;
2444 (*PresentItem).second.second = FragOrder;
2445 *out << Verbose(2) << "Updated element (" << (*PresentItem).first << ",[" << (*PresentItem).second.first << "," << (*PresentItem).second.second << "])." << endl;
2446 }
2447 } else {
2448 *out << Verbose(2) << "Inserted element (" << (*PresentItem).first << ",[" << (*PresentItem).second.first << "," << (*PresentItem).second.second << "])." << endl;
2449 }
2450 } else {
2451 *out << Verbose(1) << "No Fragment under No. " << No << "found." << endl;
2452 }
2453 }
2454 }
2455 // then map back onto (Value, (Root Nr., Order)) (i.e. sorted by value to pick the highest ones)
2456 map<double, pair<int,int> > FinalRootCandidates;
2457 *out << Verbose(1) << "Root candidate list is: " << endl;
2458 for(map<int, pair<double,int> >::iterator runner = AdaptiveCriteriaList.begin(); runner != AdaptiveCriteriaList.end(); runner++) {
2459 Walker = FindAtom((*runner).first);
2460 if (Walker != NULL) {
2461 if ((*runner).second.second >= Walker->AdaptiveOrder) { // only insert if this is an "active" root site for the current order
2462 *out << Verbose(2) << "(" << (*runner).first << ",[" << (*runner).second.first << "," << (*runner).second.second << "])" << endl;
2463 FinalRootCandidates.insert( make_pair( (*runner).second.first, pair<int,int>((*runner).first, (*runner).second.second) ) );
2464 }
2465 } else {
2466 cerr << "Atom No. " << (*runner).second.first << " was not found in this molecule." << endl;
2467 }
2468 }
2469 // pick the ones still below threshold and mark as to be adaptively updated
2470 for(map<double, pair<int,int> >::iterator runner = FinalRootCandidates.upper_bound(pow(10.,Order)); runner != FinalRootCandidates.end(); runner++) {
2471 No = (*runner).second.first;
2472 Walker = FindAtom(No);
2473 if (Walker->AdaptiveOrder < MinimumRingSize[Walker->nr]) {
2474 *out << Verbose(2) << "Root " << No << " is still above threshold (10^{" << Order <<"}: " << runner->first << ", setting entry " << No << " of Atom mask to true." << endl;
2475 AtomMask[No] = true;
2476 status = true;
2477 } else
2478 *out << Verbose(2) << "Root " << No << " is still above threshold (10^{" << Order <<"}: " << runner->first << ", however MinimumRingSize of " << MinimumRingSize[Walker->nr] << " does not allow further adaptive increase." << endl;
2479 }
2480 // close and done
2481 InputFile.close();
2482 InputFile.clear();
2483 } else {
2484 cerr << "Unable to parse " << buffer << " file, incrementing all." << endl;
2485 while (Walker->next != end) {
2486 Walker = Walker->next;
2487 #ifdef ADDHYDROGEN
2488 if (Walker->type->Z != 1) // skip hydrogen
2489 #endif
2490 {
2491 AtomMask[Walker->nr] = true; // include all (non-hydrogen) atoms
2492 status = true;
2493 }
2494 }
2495 }
2496 Free((void **)&buffer, "molecule::CheckOrderAtSite: *buffer");
2497 // pick a given number of highest values and set AtomMask
2498 } else { // global increase of Bond Order
2499 while (Walker->next != end) {
2500 Walker = Walker->next;
2501 #ifdef ADDHYDROGEN
2502 if (Walker->type->Z != 1) // skip hydrogen
2503 #endif
2504 {
2505 AtomMask[Walker->nr] = true; // include all (non-hydrogen) atoms
2506 if ((Order != 0) && (Walker->AdaptiveOrder < Order) && (Walker->AdaptiveOrder < MinimumRingSize[Walker->nr]))
2507 status = true;
2508 }
2509 }
2510 if ((Order == 0) && (AtomMask[AtomCount] == true)) // single stepping, just check
2511 status = false;
2512
2513 if (!status)
2514 *out << Verbose(1) << "Order at every site is already equal or above desired order " << Order << "." << endl;
2515 }
2516
2517 // print atom mask for debugging
2518 *out << " ";
2519 for(int i=0;i<AtomCount;i++)
2520 *out << (i % 10);
2521 *out << endl << "Atom mask is: ";
2522 for(int i=0;i<AtomCount;i++)
2523 *out << (AtomMask[i] ? "t" : "f");
2524 *out << endl;
2525
2526 return status;
2527};
2528
2529/** Create a SortIndex to map from BFS labels to the sequence in which the atoms are given in the config file.
2530 * \param *out output stream for debugging
2531 * \param *&SortIndex Mapping array of size molecule::AtomCount
2532 * \return true - success, false - failure of SortIndex alloc
2533 */
2534bool molecule::CreateMappingLabelsToConfigSequence(ofstream *out, int *&SortIndex)
2535{
2536 element *runner = elemente->start;
2537 int AtomNo = 0;
2538 atom *Walker = NULL;
2539
2540 if (SortIndex != NULL) {
2541 *out << Verbose(1) << "SortIndex is " << SortIndex << " and not NULL as expected." << endl;
2542 return false;
2543 }
2544 SortIndex = (int *) Malloc(sizeof(int)*AtomCount, "molecule::FragmentMolecule: *SortIndex");
2545 for(int i=AtomCount;i--;)
2546 SortIndex[i] = -1;
2547 while (runner->next != elemente->end) { // go through every element
2548 runner = runner->next;
2549 if (ElementsInMolecule[runner->Z]) { // if this element got atoms
2550 Walker = start;
2551 while (Walker->next != end) { // go through every atom of this element
2552 Walker = Walker->next;
2553 if (Walker->type->Z == runner->Z) // if this atom fits to element
2554 SortIndex[Walker->nr] = AtomNo++;
2555 }
2556 }
2557 }
2558 return true;
2559};
2560
2561/** Performs a many-body bond order analysis for a given bond order.
2562 * -# parses adjacency, keysets and orderatsite files
2563 * -# performs DFS to find connected subgraphs (to leave this in was a design decision: might be useful later)
2564 * -# RootStack is created for every subgraph (here, later we implement the "update 10 sites with highest energ
2565y contribution", and that's why this consciously not done in the following loop)
2566 * -# in a loop over all subgraphs
2567 * -# calls FragmentBOSSANOVA with this RootStack and within the subgraph molecule structure
2568 * -# creates molecule (fragment)s from the returned keysets (StoreFragmentFromKeySet)
2569 * -# combines the generated molecule lists from all subgraphs
2570 * -# saves to disk: fragment configs, adjacency, orderatsite, keyset files
2571 * Note that as we split "this" molecule up into a list of subgraphs, i.e. a MoleculeListClass, we have two sets
2572 * of vertex indices: Global always means the index in "this" molecule, whereas local refers to the molecule or
2573 * subgraph in the MoleculeListClass.
2574 * \param *out output stream for debugging
2575 * \param Order up to how many neighbouring bonds a fragment contains in BondOrderScheme::BottumUp scheme
2576 * \param *configuration configuration for writing config files for each fragment
2577 */
2578void molecule::FragmentMolecule(ofstream *out, int Order, config *configuration)
2579{
2580 MoleculeListClass *BondFragments = NULL;
2581 int *SortIndex = NULL;
2582 int *MinimumRingSize = NULL;
2583 int FragmentCounter;
2584 MoleculeLeafClass *MolecularWalker = NULL;
2585 MoleculeLeafClass *Subgraphs = NULL; // list of subgraphs from DFS analysis
2586 fstream File;
2587 bool FragmentationToDo = true;
2588 Graph **FragmentList = NULL;
2589 Graph *ParsedFragmentList = NULL;
2590 Graph TotalGraph; // graph with all keysets however local numbers
2591 int TotalNumberOfKeySets = 0;
2592 atom **ListOfAtoms = NULL;
2593 atom ***ListOfLocalAtoms = NULL;
2594 bool *AtomMask = NULL;
2595
2596 *out << endl;
2597#ifdef ADDHYDROGEN
2598 *out << Verbose(0) << "I will treat hydrogen special and saturate dangling bonds with it." << endl;
2599#else
2600 *out << Verbose(0) << "Hydrogen is treated just like the rest of the lot." << endl;
2601#endif
2602
2603 // ++++++++++++++++++++++++++++ INITIAL STUFF: Bond structure analysis, file parsing, ... ++++++++++++++++++++++++++++++++++++++++++
2604
2605 // ===== 1. Check whether bond structure is same as stored in files ====
2606
2607 // fill the adjacency list
2608 CreateListOfBondsPerAtom(out);
2609
2610 // create lookup table for Atom::nr
2611 FragmentationToDo = FragmentationToDo && CreateFatherLookupTable(out, start, end, ListOfAtoms, AtomCount);
2612
2613 // === compare it with adjacency file ===
2614 FragmentationToDo = FragmentationToDo && CheckAdjacencyFileAgainstMolecule(out, configuration->configpath, ListOfAtoms);
2615 Free((void **)&ListOfAtoms, "molecule::FragmentMolecule - **ListOfAtoms");
2616
2617 // ===== 2. perform a DFS analysis to gather info on cyclic structure and a list of disconnected subgraphs =====
2618 Subgraphs = DepthFirstSearchAnalysis(out, MinimumRingSize);
2619 // fill the bond structure of the individually stored subgraphs
2620 Subgraphs->next->FillBondStructureFromReference(out, this, (FragmentCounter = 0), ListOfLocalAtoms, false); // we want to keep the created ListOfLocalAtoms
2621
2622 // ===== 3. if structure still valid, parse key set file and others =====
2623 FragmentationToDo = FragmentationToDo && ParseKeySetFile(out, configuration->configpath, ParsedFragmentList);
2624
2625 // ===== 4. check globally whether there's something to do actually (first adaptivity check)
2626 FragmentationToDo = FragmentationToDo && ParseOrderAtSiteFromFile(out, configuration->configpath);
2627
2628 // =================================== Begin of FRAGMENTATION ===============================
2629 // ===== 6a. assign each keyset to its respective subgraph =====
2630 Subgraphs->next->AssignKeySetsToFragment(out, this, ParsedFragmentList, ListOfLocalAtoms, FragmentList, (FragmentCounter = 0), false);
2631
2632 KeyStack *RootStack = new KeyStack[Subgraphs->next->Count()];
2633 AtomMask = new bool[AtomCount+1];
2634 while (CheckOrderAtSite(out, AtomMask, ParsedFragmentList, Order, MinimumRingSize, configuration->configpath)) {
2635 AtomMask[AtomCount] = true; // last plus one entry is used as marker that we have been through this loop once already in CheckOrderAtSite()
2636 // ===== 6b. fill RootStack for each subgraph (second adaptivity check) =====
2637 Subgraphs->next->FillRootStackForSubgraphs(out, RootStack, AtomMask, (FragmentCounter = 0));
2638
2639 // ===== 7. fill the bond fragment list =====
2640 FragmentCounter = 0;
2641 MolecularWalker = Subgraphs;
2642 while (MolecularWalker->next != NULL) {
2643 MolecularWalker = MolecularWalker->next;
2644 *out << Verbose(1) << "Fragmenting subgraph " << MolecularWalker << "." << endl;
2645 // output ListOfBondsPerAtom for debugging
2646 MolecularWalker->Leaf->OutputListOfBonds(out);
2647 if (MolecularWalker->Leaf->first->next != MolecularWalker->Leaf->last) {
2648
2649 // call BOSSANOVA method
2650 *out << Verbose(0) << endl << " ========== BOND ENERGY of subgraph " << FragmentCounter << " ========================= " << endl;
2651 MolecularWalker->Leaf->FragmentBOSSANOVA(out, FragmentList[FragmentCounter], RootStack[FragmentCounter], MinimumRingSize);
2652 } else {
2653 cerr << "Subgraph " << MolecularWalker << " has no atoms!" << endl;
2654 }
2655 FragmentCounter++; // next fragment list
2656 }
2657 }
2658 delete[](RootStack);
2659 delete[](AtomMask);
2660 delete(ParsedFragmentList);
2661 delete[](MinimumRingSize);
2662
2663 // free the index lookup list
2664 for (int i=FragmentCounter;i--;)
2665 Free((void **)&ListOfLocalAtoms[i], "molecule::FragmentMolecule - *ListOfLocalAtoms[]");
2666 Free((void **)&ListOfLocalAtoms, "molecule::FragmentMolecule - **ListOfLocalAtoms");
2667
2668 // ==================================== End of FRAGMENTATION ============================================
2669
2670 // ===== 8a. translate list into global numbers (i.e. ones that are valid in "this" molecule, not in MolecularWalker->Leaf)
2671 Subgraphs->next->TranslateIndicesToGlobalIDs(out, FragmentList, (FragmentCounter = 0), TotalNumberOfKeySets, TotalGraph);
2672
2673 // free subgraph memory again
2674 FragmentCounter = 0;
2675 if (Subgraphs != NULL) {
2676 while (Subgraphs->next != NULL) {
2677 Subgraphs = Subgraphs->next;
2678 delete(FragmentList[FragmentCounter++]);
2679 delete(Subgraphs->previous);
2680 }
2681 delete(Subgraphs);
2682 }
2683 Free((void **)&FragmentList, "molecule::FragmentMolecule - **FragmentList");
2684
2685 // ===== 8b. gather keyset lists (graphs) from all subgraphs and transform into MoleculeListClass =====
2686 // allocate memory for the pointer array and transmorph graphs into full molecular fragments
2687 BondFragments = new MoleculeListClass(TotalGraph.size(), AtomCount);
2688 int k=0;
2689 for(Graph::iterator runner = TotalGraph.begin(); runner != TotalGraph.end(); runner++) {
2690 KeySet test = (*runner).first;
2691 *out << "Fragment No." << (*runner).second.first << " with TEFactor " << (*runner).second.second << "." << endl;
2692 BondFragments->ListOfMolecules[k] = StoreFragmentFromKeySet(out, test, configuration);
2693 k++;
2694 }
2695 *out << k << "/" << BondFragments->NumberOfMolecules << " fragments generated from the keysets." << endl;
2696
2697 // ===== 9. Save fragments' configuration and keyset files et al to disk ===
2698 if (BondFragments->NumberOfMolecules != 0) {
2699 // create the SortIndex from BFS labels to order in the config file
2700 CreateMappingLabelsToConfigSequence(out, SortIndex);
2701
2702 *out << Verbose(1) << "Writing " << BondFragments->NumberOfMolecules << " possible bond fragmentation configs" << endl;
2703 if (BondFragments->OutputConfigForListOfFragments(out, configuration, SortIndex))
2704 *out << Verbose(1) << "All configs written." << endl;
2705 else
2706 *out << Verbose(1) << "Some config writing failed." << endl;
2707
2708 // store force index reference file
2709 BondFragments->StoreForcesFile(out, configuration->configpath, SortIndex);
2710
2711 // store keysets file
2712 StoreKeySetFile(out, TotalGraph, configuration->configpath);
2713
2714 // store Adjacency file
2715 StoreAdjacencyToFile(out, configuration->configpath);
2716
2717 // store Hydrogen saturation correction file
2718 BondFragments->AddHydrogenCorrection(out, configuration->configpath);
2719
2720 // store adaptive orders into file
2721 StoreOrderAtSiteFile(out, configuration->configpath);
2722
2723 // restore orbital and Stop values
2724 CalculateOrbitals(*configuration);
2725
2726 // free memory for bond part
2727 *out << Verbose(1) << "Freeing bond memory" << endl;
2728 delete(FragmentList); // remove bond molecule from memory
2729 Free((void **)&SortIndex, "molecule::FragmentMolecule: *SortIndex");
2730 } else
2731 *out << Verbose(1) << "FragmentList is zero on return, splitting failed." << endl;
2732
2733 *out << Verbose(0) << "End of bond fragmentation." << endl;
2734};
2735
2736/** Stores pairs (Atom::nr, Atom::AdaptiveOrder) into file.
2737 * Atoms not present in the file get "-1".
2738 * \param *out output stream for debugging
2739 * \param *path path to file ORDERATSITEFILE
2740 * \return true - file writable, false - not writable
2741 */
2742bool molecule::StoreOrderAtSiteFile(ofstream *out, char *path)
2743{
2744 stringstream line;
2745 ofstream file;
2746
2747 line << path << "/" << FRAGMENTPREFIX << ORDERATSITEFILE;
2748 file.open(line.str().c_str());
2749 *out << Verbose(1) << "Writing OrderAtSite " << ORDERATSITEFILE << " ... " << endl;
2750 if (file != NULL) {
2751 atom *Walker = start;
2752 while (Walker->next != end) {
2753 Walker = Walker->next;
2754 file << Walker->nr << "\t" << (int)Walker->AdaptiveOrder << endl;
2755 *out << Verbose(2) << "Storing: " << Walker->nr << "\t" << (int)Walker->AdaptiveOrder << "." << endl;
2756 }
2757 file.close();
2758 *out << Verbose(1) << "done." << endl;
2759 return true;
2760 } else {
2761 *out << Verbose(1) << "failed to open file " << line.str() << "." << endl;
2762 return false;
2763 }
2764};
2765
2766/** Parses pairs(Atom::nr, Atom::AdaptiveOrder) from file and stores in molecule's Atom's.
2767 * Atoms not present in the file get "0".
2768 * \param *out output stream for debugging
2769 * \param *path path to file ORDERATSITEFILEe
2770 * \return true - file found and scanned, false - file not found
2771 * \sa ParseKeySetFile() and CheckAdjacencyFileAgainstMolecule() as this is meant to be used in conjunction with the two
2772 */
2773bool molecule::ParseOrderAtSiteFromFile(ofstream *out, char *path)
2774{
2775 unsigned char *OrderArray = (unsigned char *) Malloc(sizeof(unsigned char)*AtomCount, "molecule::ParseOrderAtSiteFromFile - *OrderArray");
2776 bool status;
2777 int AtomNr;
2778 stringstream line;
2779 ifstream file;
2780 int Order;
2781
2782 *out << Verbose(1) << "Begin of ParseOrderAtSiteFromFile" << endl;
2783 for(int i=AtomCount;i--;)
2784 OrderArray[i] = 0;
2785 line << path << "/" << FRAGMENTPREFIX << ORDERATSITEFILE;
2786 file.open(line.str().c_str());
2787 if (file != NULL) {
2788 for (int i=AtomCount;i--;) // initialise with 0
2789 OrderArray[i] = 0;
2790 while (!file.eof()) { // parse from file
2791 file >> AtomNr;
2792 file >> Order;
2793 OrderArray[AtomNr] = (unsigned char) Order;
2794 //*out << Verbose(2) << "AtomNr " << AtomNr << " with order " << (int)OrderArray[AtomNr] << "." << endl;
2795 }
2796 atom *Walker = start;
2797 while (Walker->next != end) { // fill into atom classes
2798 Walker = Walker->next;
2799 Walker->AdaptiveOrder = OrderArray[Walker->nr];
2800 *out << Verbose(2) << *Walker << " gets order " << (int)Walker->AdaptiveOrder << "." << endl;
2801 }
2802 file.close();
2803 *out << Verbose(1) << "done." << endl;
2804 status = true;
2805 } else {
2806 *out << Verbose(1) << "failed to open file " << line.str() << "." << endl;
2807 status = false;
2808 }
2809 Free((void **)&OrderArray, "molecule::ParseOrderAtSiteFromFile - *OrderArray");
2810
2811 *out << Verbose(1) << "End of ParseOrderAtSiteFromFile" << endl;
2812 return status;
2813};
2814
2815/** Creates an 2d array of pointer with an entry for each atom and each bond it has.
2816 * Updates molecule::ListOfBondsPerAtom, molecule::NumberOfBondsPerAtom by parsing through
2817 * bond chain list, using molecule::AtomCount and molecule::BondCount.
2818 * Allocates memory, fills the array and exits
2819 * \param *out output stream for debugging
2820 */
2821void molecule::CreateListOfBondsPerAtom(ofstream *out)
2822{
2823 bond *Binder = NULL;
2824 atom *Walker = NULL;
2825 int TotalDegree;
2826 *out << Verbose(1) << "Begin of Creating ListOfBondsPerAtom: AtomCount = " << AtomCount << "\tBondCount = " << BondCount << "\tNoNonBonds = " << NoNonBonds << "." << endl;
2827
2828 // re-allocate memory
2829 *out << Verbose(2) << "(Re-)Allocating memory." << endl;
2830 if (ListOfBondsPerAtom != NULL) {
2831 for(int i=AtomCount;i--;)
2832 Free((void **)&ListOfBondsPerAtom[i], "molecule::CreateListOfBondsPerAtom: ListOfBondsPerAtom[i]");
2833 Free((void **)&ListOfBondsPerAtom, "molecule::CreateListOfBondsPerAtom: ListOfBondsPerAtom");
2834 }
2835 if (NumberOfBondsPerAtom != NULL)
2836 Free((void **)&NumberOfBondsPerAtom, "molecule::CreateListOfBondsPerAtom: NumberOfBondsPerAtom");
2837 ListOfBondsPerAtom = (bond ***) Malloc(sizeof(bond **)*AtomCount, "molecule::CreateListOfBondsPerAtom: ***ListOfBondsPerAtom");
2838 NumberOfBondsPerAtom = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CreateListOfBondsPerAtom: *NumberOfBondsPerAtom");
2839
2840 // reset bond counts per atom
2841 for(int i=AtomCount;i--;)
2842 NumberOfBondsPerAtom[i] = 0;
2843 // count bonds per atom
2844 Binder = first;
2845 while (Binder->next != last) {
2846 Binder = Binder->next;
2847 NumberOfBondsPerAtom[Binder->leftatom->nr]++;
2848 NumberOfBondsPerAtom[Binder->rightatom->nr]++;
2849 }
2850 for(int i=AtomCount;i--;) {
2851 // allocate list of bonds per atom
2852 ListOfBondsPerAtom[i] = (bond **) Malloc(sizeof(bond *)*NumberOfBondsPerAtom[i], "molecule::CreateListOfBondsPerAtom: **ListOfBondsPerAtom[]");
2853 // clear the list again, now each NumberOfBondsPerAtom marks current free field
2854 NumberOfBondsPerAtom[i] = 0;
2855 }
2856 // fill the list
2857 Binder = first;
2858 while (Binder->next != last) {
2859 Binder = Binder->next;
2860 ListOfBondsPerAtom[Binder->leftatom->nr][NumberOfBondsPerAtom[Binder->leftatom->nr]++] = Binder;
2861 ListOfBondsPerAtom[Binder->rightatom->nr][NumberOfBondsPerAtom[Binder->rightatom->nr]++] = Binder;
2862 }
2863
2864 // output list for debugging
2865 *out << Verbose(3) << "ListOfBondsPerAtom for each atom:" << endl;
2866 Walker = start;
2867 while (Walker->next != end) {
2868 Walker = Walker->next;
2869 *out << Verbose(4) << "Atom " << Walker->Name << " with " << NumberOfBondsPerAtom[Walker->nr] << " bonds: ";
2870 TotalDegree = 0;
2871 for (int j=0;j<NumberOfBondsPerAtom[Walker->nr];j++) {
2872 *out << *ListOfBondsPerAtom[Walker->nr][j] << "\t";
2873 TotalDegree += ListOfBondsPerAtom[Walker->nr][j]->BondDegree;
2874 }
2875 *out << " -- TotalDegree: " << TotalDegree << endl;
2876 }
2877 *out << Verbose(1) << "End of Creating ListOfBondsPerAtom." << endl << endl;
2878};
2879
2880/** Adds atoms up to \a BondCount distance from \a *Root and notes them down in \a **AddedAtomList.
2881 * Gray vertices are always enqueued in an StackClass<atom *> FIFO queue, the rest is usual BFS with adding vertices found was
2882 * white and putting into queue.
2883 * \param *out output stream for debugging
2884 * \param *Mol Molecule class to add atoms to
2885 * \param **AddedAtomList list with added atom pointers, index is atom father's number
2886 * \param **AddedBondList list with added bond pointers, index is bond father's number
2887 * \param *Root root vertex for BFS
2888 * \param *Bond bond not to look beyond
2889 * \param BondOrder maximum distance for vertices to add
2890 * \param IsAngstroem lengths are in angstroem or bohrradii
2891 */
2892void molecule::BreadthFirstSearchAdd(ofstream *out, molecule *Mol, atom **&AddedAtomList, bond **&AddedBondList, atom *Root, bond *Bond, int BondOrder, bool IsAngstroem)
2893{
2894 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::BreadthFirstSearchAdd: **PredecessorList");
2895 int *ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::BreadthFirstSearchAdd: *ShortestPathList");
2896 enum Shading *ColorList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::BreadthFirstSearchAdd: *ColorList");
2897 class StackClass<atom *> *AtomStack = new StackClass<atom *>(AtomCount);
2898 atom *Walker = NULL, *OtherAtom = NULL;
2899 bond *Binder = NULL;
2900
2901 // add Root if not done yet
2902 AtomStack->ClearStack();
2903 if (AddedAtomList[Root->nr] == NULL) // add Root if not yet present
2904 AddedAtomList[Root->nr] = Mol->AddCopyAtom(Root);
2905 AtomStack->Push(Root);
2906
2907 // initialise each vertex as white with no predecessor, empty queue, color Root lightgray
2908 for (int i=AtomCount;i--;) {
2909 PredecessorList[i] = NULL;
2910 ShortestPathList[i] = -1;
2911 if (AddedAtomList[i] != NULL) // mark already present atoms (i.e. Root and maybe others) as visited
2912 ColorList[i] = lightgray;
2913 else
2914 ColorList[i] = white;
2915 }
2916 ShortestPathList[Root->nr] = 0;
2917
2918 // and go on ... Queue always contains all lightgray vertices
2919 while (!AtomStack->IsEmpty()) {
2920 // we have to pop the oldest atom from stack. This keeps the atoms on the stack always of the same ShortestPath distance.
2921 // e.g. if current atom is 2, push to end of stack are of length 3, but first all of length 2 would be popped. They again
2922 // append length of 3 (their neighbours). Thus on stack we have always atoms of a certain length n at bottom of stack and
2923 // followed by n+1 till top of stack.
2924 Walker = AtomStack->PopFirst(); // pop oldest added
2925 *out << Verbose(1) << "Current Walker is: " << Walker->Name << ", and has " << NumberOfBondsPerAtom[Walker->nr] << " bonds." << endl;
2926 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
2927 Binder = ListOfBondsPerAtom[Walker->nr][i];
2928 if (Binder != NULL) { // don't look at bond equal NULL
2929 OtherAtom = Binder->GetOtherAtom(Walker);
2930 *out << Verbose(2) << "Current OtherAtom is: " << OtherAtom->Name << " for bond " << *Binder << "." << endl;
2931 if (ColorList[OtherAtom->nr] == white) {
2932 if (Binder != Bond) // let other atom white if it's via Root bond. In case it's cyclic it has to be reached again (yet Root is from OtherAtom already black, thus no problem)
2933 ColorList[OtherAtom->nr] = lightgray;
2934 PredecessorList[OtherAtom->nr] = Walker; // Walker is the predecessor
2935 ShortestPathList[OtherAtom->nr] = ShortestPathList[Walker->nr]+1;
2936 *out << Verbose(2) << "Coloring OtherAtom " << OtherAtom->Name << " " << ((ColorList[OtherAtom->nr] == white) ? "white" : "lightgray") << ", its predecessor is " << Walker->Name << " and its Shortest Path is " << ShortestPathList[OtherAtom->nr] << " egde(s) long." << endl;
2937 if ((((ShortestPathList[OtherAtom->nr] < BondOrder) && (Binder != Bond))) ) { // Check for maximum distance
2938 *out << Verbose(3);
2939 if (AddedAtomList[OtherAtom->nr] == NULL) { // add if it's not been so far
2940 AddedAtomList[OtherAtom->nr] = Mol->AddCopyAtom(OtherAtom);
2941 *out << "Added OtherAtom " << OtherAtom->Name;
2942 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
2943 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
2944 AddedBondList[Binder->nr]->Type = Binder->Type;
2945 *out << " and bond " << *(AddedBondList[Binder->nr]) << ", ";
2946 } else { // this code should actually never come into play (all white atoms are not yet present in BondMolecule, that's why they are white in the first place)
2947 *out << "Not adding OtherAtom " << OtherAtom->Name;
2948 if (AddedBondList[Binder->nr] == NULL) {
2949 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
2950 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
2951 AddedBondList[Binder->nr]->Type = Binder->Type;
2952 *out << ", added Bond " << *(AddedBondList[Binder->nr]);
2953 } else
2954 *out << ", not added Bond ";
2955 }
2956 *out << ", putting OtherAtom into queue." << endl;
2957 AtomStack->Push(OtherAtom);
2958 } else { // out of bond order, then replace
2959 if ((AddedAtomList[OtherAtom->nr] == NULL) && (Binder->Cyclic))
2960 ColorList[OtherAtom->nr] = white; // unmark if it has not been queued/added, to make it available via its other bonds (cyclic)
2961 if (Binder == Bond)
2962 *out << Verbose(3) << "Not Queueing, is the Root bond";
2963 else if (ShortestPathList[OtherAtom->nr] >= BondOrder)
2964 *out << Verbose(3) << "Not Queueing, is out of Bond Count of " << BondOrder;
2965 if (!Binder->Cyclic)
2966 *out << ", is not part of a cyclic bond, saturating bond with Hydrogen." << endl;
2967 if (AddedBondList[Binder->nr] == NULL) {
2968 if ((AddedAtomList[OtherAtom->nr] != NULL)) { // .. whether we add or saturate
2969 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
2970 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
2971 AddedBondList[Binder->nr]->Type = Binder->Type;
2972 } else {
2973#ifdef ADDHYDROGEN
2974 Mol->AddHydrogenReplacementAtom(out, Binder, AddedAtomList[Walker->nr], Walker, OtherAtom, ListOfBondsPerAtom[Walker->nr], NumberOfBondsPerAtom[Walker->nr], IsAngstroem);
2975#endif
2976 }
2977 }
2978 }
2979 } else {
2980 *out << Verbose(3) << "Not Adding, has already been visited." << endl;
2981 // This has to be a cyclic bond, check whether it's present ...
2982 if (AddedBondList[Binder->nr] == NULL) {
2983 if ((Binder != Bond) && (Binder->Cyclic) && (((ShortestPathList[Walker->nr]+1) < BondOrder))) {
2984 AddedBondList[Binder->nr] = Mol->AddBond(AddedAtomList[Walker->nr], AddedAtomList[OtherAtom->nr], Binder->BondDegree);
2985 AddedBondList[Binder->nr]->Cyclic = Binder->Cyclic;
2986 AddedBondList[Binder->nr]->Type = Binder->Type;
2987 } else { // if it's root bond it has to broken (otherwise we would not create the fragments)
2988#ifdef ADDHYDROGEN
2989 Mol->AddHydrogenReplacementAtom(out, Binder, AddedAtomList[Walker->nr], Walker, OtherAtom, ListOfBondsPerAtom[Walker->nr], NumberOfBondsPerAtom[Walker->nr], IsAngstroem);
2990#endif
2991 }
2992 }
2993 }
2994 }
2995 }
2996 ColorList[Walker->nr] = black;
2997 *out << Verbose(1) << "Coloring Walker " << Walker->Name << " black." << endl;
2998 }
2999 Free((void **)&PredecessorList, "molecule::BreadthFirstSearchAdd: **PredecessorList");
3000 Free((void **)&ShortestPathList, "molecule::BreadthFirstSearchAdd: **ShortestPathList");
3001 Free((void **)&ColorList, "molecule::BreadthFirstSearchAdd: **ColorList");
3002 delete(AtomStack);
3003};
3004
3005/** Adds bond structure to this molecule from \a Father molecule.
3006 * This basically causes this molecule to become an induced subgraph of the \a Father, i.e. for every bond in Father
3007 * with end points present in this molecule, bond is created in this molecule.
3008 * Special care was taken to ensure that this is of complexity O(N), where N is the \a Father's molecule::AtomCount.
3009 * \param *out output stream for debugging
3010 * \param *Father father molecule
3011 * \return true - is induced subgraph, false - there are atoms with fathers not in \a Father
3012 * \todo not checked, not fully working probably
3013 */
3014bool molecule::BuildInducedSubgraph(ofstream *out, const molecule *Father)
3015{
3016 atom *Walker = NULL, *OtherAtom = NULL;
3017 bool status = true;
3018 atom **ParentList = (atom **) Malloc(sizeof(atom *)*Father->AtomCount, "molecule::BuildInducedSubgraph: **ParentList");
3019
3020 *out << Verbose(2) << "Begin of BuildInducedSubgraph." << endl;
3021
3022 // reset parent list
3023 *out << Verbose(3) << "Resetting ParentList." << endl;
3024 for (int i=Father->AtomCount;i--;)
3025 ParentList[i] = NULL;
3026
3027 // fill parent list with sons
3028 *out << Verbose(3) << "Filling Parent List." << endl;
3029 Walker = start;
3030 while (Walker->next != end) {
3031 Walker = Walker->next;
3032 ParentList[Walker->father->nr] = Walker;
3033 // Outputting List for debugging
3034 *out << Verbose(4) << "Son["<< Walker->father->nr <<"] of " << Walker->father << " is " << ParentList[Walker->father->nr] << "." << endl;
3035 }
3036
3037 // check each entry of parent list and if ok (one-to-and-onto matching) create bonds
3038 *out << Verbose(3) << "Creating bonds." << endl;
3039 Walker = Father->start;
3040 while (Walker->next != Father->end) {
3041 Walker = Walker->next;
3042 if (ParentList[Walker->nr] != NULL) {
3043 if (ParentList[Walker->nr]->father != Walker) {
3044 status = false;
3045 } else {
3046 for (int i=0;i<Father->NumberOfBondsPerAtom[Walker->nr];i++) {
3047 OtherAtom = Father->ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
3048 if (ParentList[OtherAtom->nr] != NULL) { // if otheratom is also a father of an atom on this molecule, create the bond
3049 *out << Verbose(4) << "Endpoints of Bond " << Father->ListOfBondsPerAtom[Walker->nr][i] << " are both present: " << ParentList[Walker->nr]->Name << " and " << ParentList[OtherAtom->nr]->Name << "." << endl;
3050 AddBond(ParentList[Walker->nr], ParentList[OtherAtom->nr], Father->ListOfBondsPerAtom[Walker->nr][i]->BondDegree);
3051 }
3052 }
3053 }
3054 }
3055 }
3056
3057 Free((void **)&ParentList, "molecule::BuildInducedSubgraph: **ParentList");
3058 *out << Verbose(2) << "End of BuildInducedSubgraph." << endl;
3059 return status;
3060};
3061
3062
3063/** Looks through a StackClass<atom *> and returns the likeliest removal candiate.
3064 * \param *out output stream for debugging messages
3065 * \param *&Leaf KeySet to look through
3066 * \param *&ShortestPathList list of the shortest path to decide which atom to suggest as removal candidate in the end
3067 * \param index of the atom suggested for removal
3068 */
3069int molecule::LookForRemovalCandidate(ofstream *&out, KeySet *&Leaf, int *&ShortestPathList)
3070{
3071 atom *Runner = NULL;
3072 int SP, Removal;
3073
3074 *out << Verbose(2) << "Looking for removal candidate." << endl;
3075 SP = -1; //0; // not -1, so that Root is never removed
3076 Removal = -1;
3077 for (KeySet::iterator runner = Leaf->begin(); runner != Leaf->end(); runner++) {
3078 Runner = FindAtom((*runner));
3079 if (Runner->type->Z != 1) { // skip all those added hydrogens when re-filling snake stack
3080 if (ShortestPathList[(*runner)] > SP) { // remove the oldest one with longest shortest path
3081 SP = ShortestPathList[(*runner)];
3082 Removal = (*runner);
3083 }
3084 }
3085 }
3086 return Removal;
3087};
3088
3089/** Stores a fragment from \a KeySet into \a molecule.
3090 * First creates the minimal set of atoms from the KeySet, then creates the bond structure from the complete
3091 * molecule and adds missing hydrogen where bonds were cut.
3092 * \param *out output stream for debugging messages
3093 * \param &Leaflet pointer to KeySet structure
3094 * \param IsAngstroem whether we have Ansgtroem or bohrradius
3095 * \return pointer to constructed molecule
3096 */
3097molecule * molecule::StoreFragmentFromKeySet(ofstream *out, KeySet &Leaflet, bool IsAngstroem)
3098{
3099 atom *Runner = NULL, *FatherOfRunner = NULL, *OtherFather = NULL;
3100 atom **SonList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::StoreFragmentFromStack: **SonList");
3101 molecule *Leaf = new molecule(elemente);
3102
3103// *out << Verbose(1) << "Begin of StoreFragmentFromKeyset." << endl;
3104
3105 Leaf->BondDistance = BondDistance;
3106 for(int i=NDIM*2;i--;)
3107 Leaf->cell_size[i] = cell_size[i];
3108
3109 // initialise SonList (indicates when we need to replace a bond with hydrogen instead)
3110 for(int i=AtomCount;i--;)
3111 SonList[i] = NULL;
3112
3113 // first create the minimal set of atoms from the KeySet
3114 for(KeySet::iterator runner = Leaflet.begin(); runner != Leaflet.end(); runner++) {
3115 FatherOfRunner = FindAtom((*runner)); // find the id
3116 SonList[FatherOfRunner->nr] = Leaf->AddCopyAtom(FatherOfRunner);
3117 }
3118
3119 // create the bonds between all: Make it an induced subgraph and add hydrogen
3120// *out << Verbose(2) << "Creating bonds from father graph (i.e. induced subgraph creation)." << endl;
3121 Runner = Leaf->start;
3122 while (Runner->next != Leaf->end) {
3123 Runner = Runner->next;
3124 FatherOfRunner = Runner->father;
3125 if (SonList[FatherOfRunner->nr] != NULL) { // check if this, our father, is present in list
3126 // create all bonds
3127 for (int i=0;i<NumberOfBondsPerAtom[FatherOfRunner->nr];i++) { // go through every bond of father
3128 OtherFather = ListOfBondsPerAtom[FatherOfRunner->nr][i]->GetOtherAtom(FatherOfRunner);
3129// *out << Verbose(2) << "Father " << *FatherOfRunner << " of son " << *SonList[FatherOfRunner->nr] << " is bound to " << *OtherFather;
3130 if (SonList[OtherFather->nr] != NULL) {
3131// *out << ", whose son is " << *SonList[OtherFather->nr] << "." << endl;
3132 if (OtherFather->nr > FatherOfRunner->nr) { // add bond (nr check is for adding only one of both variants: ab, ba)
3133// *out << Verbose(3) << "Adding Bond: ";
3134// *out <<
3135 Leaf->AddBond(Runner, SonList[OtherFather->nr], ListOfBondsPerAtom[FatherOfRunner->nr][i]->BondDegree);
3136// *out << "." << endl;
3137 //NumBonds[Runner->nr]++;
3138 } else {
3139// *out << Verbose(3) << "Not adding bond, labels in wrong order." << endl;
3140 }
3141 } else {
3142// *out << ", who has no son in this fragment molecule." << endl;
3143#ifdef ADDHYDROGEN
3144 //*out << Verbose(3) << "Adding Hydrogen to " << Runner->Name << " and a bond in between." << endl;
3145 Leaf->AddHydrogenReplacementAtom(out, ListOfBondsPerAtom[FatherOfRunner->nr][i], Runner, FatherOfRunner, OtherFather, ListOfBondsPerAtom[FatherOfRunner->nr],NumberOfBondsPerAtom[FatherOfRunner->nr], IsAngstroem);
3146#endif
3147 //NumBonds[Runner->nr] += ListOfBondsPerAtom[FatherOfRunner->nr][i]->BondDegree;
3148 }
3149 }
3150 } else {
3151 *out << Verbose(0) << "ERROR: Son " << Runner->Name << " has father " << FatherOfRunner->Name << " but its entry in SonList is " << SonList[FatherOfRunner->nr] << "!" << endl;
3152 }
3153#ifdef ADDHYDROGEN
3154 while ((Runner->next != Leaf->end) && (Runner->next->type->Z == 1)) // skip added hydrogen
3155 Runner = Runner->next;
3156#endif
3157 }
3158 Leaf->CreateListOfBondsPerAtom(out);
3159 //Leaflet->Leaf->ScanForPeriodicCorrection(out);
3160 Free((void **)&SonList, "molecule::StoreFragmentFromStack: **SonList");
3161// *out << Verbose(1) << "End of StoreFragmentFromKeyset." << endl;
3162 return Leaf;
3163};
3164
3165/** Creates \a MoleculeListClass of all unique fragments of the \a molecule containing \a Order atoms or vertices.
3166 * The picture to have in mind is that of a DFS "snake" of a certain length \a Order, i.e. as in the infamous
3167 * computer game, that winds through the connected graph representing the molecule. Color (white,
3168 * lightgray, darkgray, black) indicates whether a vertex has been discovered so far or not. Labels will help in
3169 * creating only unique fragments and not additional ones with vertices simply in different sequence.
3170 * The Predecessor is always the one that came before in discovering, needed on backstepping. And
3171 * finally, the ShortestPath is needed for removing vertices from the snake stack during the back-
3172 * stepping.
3173 * \param *out output stream for debugging
3174 * \param Order number of atoms in each fragment
3175 * \param *configuration configuration for writing config files for each fragment
3176 * \return List of all unique fragments with \a Order atoms
3177 */
3178/*
3179MoleculeListClass * molecule::CreateListOfUniqueFragmentsOfOrder(ofstream *out, int Order, config *configuration)
3180{
3181 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: **PredecessorList");
3182 int *ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *ShortestPathList");
3183 int *Labels = (int *) Malloc(sizeof(int)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *Labels");
3184 enum Shading *ColorVertexList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *ColorList");
3185 enum Shading *ColorEdgeList = (enum Shading *) Malloc(sizeof(enum Shading)*BondCount, "molecule::CreateListOfUniqueFragmentsOfOrder: *ColorBondList");
3186 StackClass<atom *> *RootStack = new StackClass<atom *>(AtomCount);
3187 StackClass<atom *> *TouchedStack = new StackClass<atom *>((int)pow(4,Order)+2); // number of atoms reached from one with maximal 4 bonds plus Root itself
3188 StackClass<atom *> *SnakeStack = new StackClass<atom *>(Order+1); // equal to Order is not possible, as then the StackClass<atom *> cannot discern between full and empty stack!
3189 MoleculeLeafClass *Leaflet = NULL, *TempLeaf = NULL;
3190 MoleculeListClass *FragmentList = NULL;
3191 atom *Walker = NULL, *OtherAtom = NULL, *Root = NULL, *Removal = NULL;
3192 bond *Binder = NULL;
3193 int RunningIndex = 0, FragmentCounter = 0;
3194
3195 *out << Verbose(1) << "Begin of CreateListOfUniqueFragmentsOfOrder." << endl;
3196
3197 // reset parent list
3198 *out << Verbose(3) << "Resetting labels, parent, predecessor, color and shortest path lists." << endl;
3199 for (int i=0;i<AtomCount;i++) { // reset all atom labels
3200 // initialise each vertex as white with no predecessor, empty queue, color lightgray, not labelled, no sons
3201 Labels[i] = -1;
3202 SonList[i] = NULL;
3203 PredecessorList[i] = NULL;
3204 ColorVertexList[i] = white;
3205 ShortestPathList[i] = -1;
3206 }
3207 for (int i=0;i<BondCount;i++)
3208 ColorEdgeList[i] = white;
3209 RootStack->ClearStack(); // clearstack and push first atom if exists
3210 TouchedStack->ClearStack();
3211 Walker = start->next;
3212 while ((Walker != end)
3213#ifdef ADDHYDROGEN
3214 && (Walker->type->Z == 1)
3215#endif
3216 ) { // search for first non-hydrogen atom
3217 *out << Verbose(4) << "Current Root candidate is " << Walker->Name << "." << endl;
3218 Walker = Walker->next;
3219 }
3220 if (Walker != end)
3221 RootStack->Push(Walker);
3222 else
3223 *out << Verbose(0) << "ERROR: Could not find an appropriate Root atom!" << endl;
3224 *out << Verbose(3) << "Root " << Walker->Name << " is on AtomStack, beginning loop through all vertices ..." << endl;
3225
3226 ///// OUTER LOOP ////////////
3227 while (!RootStack->IsEmpty()) {
3228 // get new root vertex from atom stack
3229 Root = RootStack->PopFirst();
3230 ShortestPathList[Root->nr] = 0;
3231 if (Labels[Root->nr] == -1)
3232 Labels[Root->nr] = RunningIndex++; // prevent it from getting again on AtomStack
3233 PredecessorList[Root->nr] = Root;
3234 TouchedStack->Push(Root);
3235 *out << Verbose(0) << "Root for this loop is: " << Root->Name << ".\n";
3236
3237 // clear snake stack
3238 SnakeStack->ClearStack();
3239 //SnakeStack->TestImplementation(out, start->next);
3240
3241 ///// INNER LOOP ////////////
3242 // Problems:
3243 // - what about cyclic bonds?
3244 Walker = Root;
3245 do {
3246 *out << Verbose(1) << "Current Walker is: " << Walker->Name;
3247 // initial setting of the new Walker: label, color, shortest path and put on stacks
3248 if (Labels[Walker->nr] == -1) { // give atom a unique, monotonely increasing number
3249 Labels[Walker->nr] = RunningIndex++;
3250 RootStack->Push(Walker);
3251 }
3252 *out << ", has label " << Labels[Walker->nr];
3253 if ((ColorVertexList[Walker->nr] == white) || ((Binder != NULL) && (ColorEdgeList[Binder->nr] == white))) { // color it if newly discovered and push on stacks (and if within reach!)
3254 if ((Binder != NULL) && (ColorEdgeList[Binder->nr] == white)) {
3255 // Binder ought to be set still from last neighbour search
3256 *out << ", coloring bond " << *Binder << " black";
3257 ColorEdgeList[Binder->nr] = black; // mark this bond as used
3258 }
3259 if (ShortestPathList[Walker->nr] == -1) {
3260 ShortestPathList[Walker->nr] = ShortestPathList[PredecessorList[Walker->nr]->nr]+1;
3261 TouchedStack->Push(Walker); // mark every atom for lists cleanup later, whose shortest path has been changed
3262 }
3263 if ((ShortestPathList[Walker->nr] < Order) && (ColorVertexList[Walker->nr] != darkgray)) { // if not already on snake stack
3264 SnakeStack->Push(Walker);
3265 ColorVertexList[Walker->nr] = darkgray; // mark as dark gray of on snake stack
3266 }
3267 }
3268 *out << ", SP of " << ShortestPathList[Walker->nr] << " and its color is " << GetColor(ColorVertexList[Walker->nr]) << "." << endl;
3269
3270 // then check the stack for a newly stumbled upon fragment
3271 if (SnakeStack->ItemCount() == Order) { // is stack full?
3272 // store the fragment if it is one and get a removal candidate
3273 Removal = StoreFragmentFromStack(out, Root, Walker, Leaflet, SnakeStack, ShortestPathList, SonList, Labels, &FragmentCounter, configuration);
3274 // remove the candidate if one was found
3275 if (Removal != NULL) {
3276 *out << Verbose(2) << "Removing item " << Removal->Name << " with SP of " << ShortestPathList[Removal->nr] << " from snake stack." << endl;
3277 SnakeStack->RemoveItem(Removal);
3278 ColorVertexList[Removal->nr] = lightgray; // return back to not on snake stack but explored marking
3279 if (Walker == Removal) { // if the current atom is to be removed, we also have to take a step back
3280 Walker = PredecessorList[Removal->nr];
3281 *out << Verbose(2) << "Stepping back to " << Walker->Name << "." << endl;
3282 }
3283 }
3284 } else
3285 Removal = NULL;
3286
3287 // finally, look for a white neighbour as the next Walker
3288 Binder = NULL;
3289 if ((Removal == NULL) || (Walker != PredecessorList[Removal->nr])) { // don't look, if a new walker has been set above
3290 *out << Verbose(2) << "Snake has currently " << SnakeStack->ItemCount() << " item(s)." << endl;
3291 OtherAtom = NULL; // this is actually not needed, every atom has at least one neighbour
3292 if (ShortestPathList[Walker->nr] < Order) {
3293 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) {
3294 Binder = ListOfBondsPerAtom[Walker->nr][i];
3295 *out << Verbose(2) << "Current bond is " << *Binder << ": ";
3296 OtherAtom = Binder->GetOtherAtom(Walker);
3297 if ((Labels[OtherAtom->nr] != -1) && (Labels[OtherAtom->nr] < Labels[Root->nr])) { // we don't step up to labels bigger than us
3298 *out << "Label " << Labels[OtherAtom->nr] << " is smaller than Root's " << Labels[Root->nr] << "." << endl;
3299 //ColorVertexList[OtherAtom->nr] = lightgray; // mark as explored
3300 } else { // otherwise check its colour and element
3301 if (
3302#ifdef ADDHYDROGEN
3303 (OtherAtom->type->Z != 1) &&
3304#endif
3305 (ColorEdgeList[Binder->nr] == white)) { // skip hydrogen, look for unexplored vertices
3306 *out << "Moving along " << GetColor(ColorEdgeList[Binder->nr]) << " bond " << Binder << " to " << ((ColorVertexList[OtherAtom->nr] == white) ? "unexplored" : "explored") << " item: " << OtherAtom->Name << "." << endl;
3307 // i find it currently rather sensible to always set the predecessor in order to find one's way back
3308 //if (PredecessorList[OtherAtom->nr] == NULL) {
3309 PredecessorList[OtherAtom->nr] = Walker;
3310 *out << Verbose(3) << "Setting Predecessor of " << OtherAtom->Name << " to " << PredecessorList[OtherAtom->nr]->Name << "." << endl;
3311 //} else {
3312 // *out << Verbose(3) << "Predecessor of " << OtherAtom->Name << " is " << PredecessorList[OtherAtom->nr]->Name << "." << endl;
3313 //}
3314 Walker = OtherAtom;
3315 break;
3316 } else {
3317 if (OtherAtom->type->Z == 1)
3318 *out << "Links to a hydrogen atom." << endl;
3319 else
3320 *out << "Bond has not white but " << GetColor(ColorEdgeList[Binder->nr]) << " color." << endl;
3321 }
3322 }
3323 }
3324 } else { // means we have stepped beyond the horizon: Return!
3325 Walker = PredecessorList[Walker->nr];
3326 OtherAtom = Walker;
3327 *out << Verbose(3) << "We have gone too far, stepping back to " << Walker->Name << "." << endl;
3328 }
3329 if (Walker != OtherAtom) { // if no white neighbours anymore, color it black
3330 *out << Verbose(2) << "Coloring " << Walker->Name << " black." << endl;
3331 ColorVertexList[Walker->nr] = black;
3332 Walker = PredecessorList[Walker->nr];
3333 }
3334 }
3335 } while ((Walker != Root) || (ColorVertexList[Root->nr] != black));
3336 *out << Verbose(2) << "Inner Looping is finished." << endl;
3337
3338 // if we reset all AtomCount atoms, we have again technically O(N^2) ...
3339 *out << Verbose(2) << "Resetting lists." << endl;
3340 Walker = NULL;
3341 Binder = NULL;
3342 while (!TouchedStack->IsEmpty()) {
3343 Walker = TouchedStack->PopLast();
3344 *out << Verbose(3) << "Re-initialising entries of " << *Walker << "." << endl;
3345 for(int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++)
3346 ColorEdgeList[ListOfBondsPerAtom[Walker->nr][i]->nr] = white;
3347 PredecessorList[Walker->nr] = NULL;
3348 ColorVertexList[Walker->nr] = white;
3349 ShortestPathList[Walker->nr] = -1;
3350 }
3351 }
3352 *out << Verbose(1) << "Outer Looping over all vertices is done." << endl;
3353
3354 // copy together
3355 *out << Verbose(1) << "Copying all fragments into MoleculeList structure." << endl;
3356 FragmentList = new MoleculeListClass(FragmentCounter, AtomCount);
3357 RunningIndex = 0;
3358 while ((Leaflet != NULL) && (RunningIndex < FragmentCounter)) {
3359 FragmentList->ListOfMolecules[RunningIndex++] = Leaflet->Leaf;
3360 Leaflet->Leaf = NULL; // prevent molecule from being removed
3361 TempLeaf = Leaflet;
3362 Leaflet = Leaflet->previous;
3363 delete(TempLeaf);
3364 };
3365
3366 // free memory and exit
3367 Free((void **)&PredecessorList, "molecule::CreateListOfUniqueFragmentsOfOrder: **PredecessorList");
3368 Free((void **)&ShortestPathList, "molecule::CreateListOfUniqueFragmentsOfOrder: *ShortestPathList");
3369 Free((void **)&Labels, "molecule::CreateListOfUniqueFragmentsOfOrder: *Labels");
3370 Free((void **)&ColorVertexList, "molecule::CreateListOfUniqueFragmentsOfOrder: *ColorList");
3371 delete(RootStack);
3372 delete(TouchedStack);
3373 delete(SnakeStack);
3374
3375 *out << Verbose(1) << "End of CreateListOfUniqueFragmentsOfOrder." << endl;
3376 return FragmentList;
3377};
3378*/
3379
3380/** Structure containing all values in power set combination generation.
3381 */
3382struct UniqueFragments {
3383 config *configuration;
3384 atom *Root;
3385 Graph *Leaflet;
3386 KeySet *FragmentSet;
3387 int ANOVAOrder;
3388 int FragmentCounter;
3389 int CurrentIndex;
3390 int *Labels;
3391 double TEFactor;
3392 int *ShortestPathList;
3393 bool **UsedList;
3394 bond **BondsPerSPList;
3395 int *BondsPerSPCount;
3396};
3397
3398/** From a given set of Bond sorted by Shortest Path distance, create all possible fragments of size \a SetDimension.
3399 * -# loops over every possible combination (2^dimension of edge set)
3400 * -# inserts current set, if there's still space left
3401 * -# yes: calls SPFragmentGenerator with structure, created new edge list and size respective to root dist
3402ance+1
3403 * -# no: stores fragment into keyset list by calling InsertFragmentIntoGraph
3404 * -# removes all items added into the snake stack (in UniqueFragments structure) added during level (root
3405distance) and current set
3406 * \param *out output stream for debugging
3407 * \param FragmentSearch UniqueFragments structure with all values needed
3408 * \param RootDistance current shortest path level, whose set of edges is represented by **BondsSet
3409 * \param SetDimension Number of possible bonds on this level (i.e. size of the array BondsSet[])
3410 * \param SubOrder remaining number of allowed vertices to add
3411 */
3412void molecule::SPFragmentGenerator(ofstream *out, struct UniqueFragments *FragmentSearch, int RootDistance, bond **BondsSet, int SetDimension, int SubOrder)
3413{
3414 atom *OtherWalker = NULL;
3415 int verbosity = 0; //FragmentSearch->ANOVAOrder-SubOrder;
3416 int NumCombinations;
3417 bool bit;
3418 int bits, TouchedIndex, SubSetDimension, SP;
3419 int Removal;
3420 int *TouchedList = (int *) Malloc(sizeof(int)*(SubOrder+1), "molecule::SPFragmentGenerator: *TouchedList");
3421 bond *Binder = NULL;
3422 bond **BondsList = NULL;
3423
3424 NumCombinations = 1 << SetDimension;
3425
3426 // Hier muessen von 1 bis NumberOfBondsPerAtom[Walker->nr] alle Kombinationen
3427 // von Endstuecken (aus den Bonds) hinzugefᅵᅵgt werden und fᅵᅵr verbleibende ANOVAOrder
3428 // rekursiv GraphCrawler in der nᅵᅵchsten Ebene aufgerufen werden
3429
3430 *out << Verbose(1+verbosity) << "Begin of SPFragmentGenerator." << endl;
3431 *out << Verbose(1+verbosity) << "We are " << RootDistance << " away from Root, which is " << *FragmentSearch->Root << ", SubOrder is " << SubOrder << ", SetDimension is " << SetDimension << " and this means " << NumCombinations-1 << " combination(s)." << endl;
3432
3433 // initialised touched list (stores added atoms on this level)
3434 *out << Verbose(1+verbosity) << "Clearing touched list." << endl;
3435 for (TouchedIndex=SubOrder+1;TouchedIndex--;) // empty touched list
3436 TouchedList[TouchedIndex] = -1;
3437 TouchedIndex = 0;
3438
3439 // create every possible combination of the endpieces
3440 *out << Verbose(1+verbosity) << "Going through all combinations of the power set." << endl;
3441 for (int i=1;i<NumCombinations;i++) { // sweep through all power set combinations (skip empty set!)
3442 // count the set bit of i
3443 bits = 0;
3444 for (int j=SetDimension;j--;)
3445 bits += (i & (1 << j)) >> j;
3446
3447 *out << Verbose(1+verbosity) << "Current set is " << Binary(i | (1 << SetDimension)) << ", number of bits is " << bits << "." << endl;
3448 if (bits <= SubOrder) { // if not greater than additional atoms allowed on stack, continue
3449 // --1-- add this set of the power set of bond partners to the snake stack
3450 for (int j=0;j<SetDimension;j++) { // pull out every bit by shifting
3451 bit = ((i & (1 << j)) != 0); // mask the bit for the j-th bond
3452 if (bit) { // if bit is set, we add this bond partner
3453 OtherWalker = BondsSet[j]->rightatom; // rightatom is always the one more distant, i.e. the one to add
3454 //*out << Verbose(1+verbosity) << "Current Bond is " << ListOfBondsPerAtom[Walker->nr][i] << ", checking on " << *OtherWalker << "." << endl;
3455 //if ((!FragmentSearch->UsedList[OtherWalker->nr][i]) && (FragmentSearch->Labels[OtherWalker->nr] > FragmentSearch->Labels[FragmentSearch->Root->nr])) {
3456 //*out << Verbose(2+verbosity) << "Not used so far, label " << FragmentSearch->Labels[OtherWalker->nr] << " is bigger than Root's " << FragmentSearch->Labels[FragmentSearch->Root->nr] << "." << endl;
3457 *out << Verbose(2+verbosity) << "Adding " << *OtherWalker << " with nr " << OtherWalker->nr << "." << endl;
3458 TouchedList[TouchedIndex++] = OtherWalker->nr; // note as added
3459 FragmentSearch->FragmentSet->insert(OtherWalker->nr);
3460 //FragmentSearch->UsedList[OtherWalker->nr][i] = true;
3461 //}
3462 } else {
3463 *out << Verbose(2+verbosity) << "Not adding." << endl;
3464 }
3465 }
3466
3467 if (bits < SubOrder) {
3468 *out << Verbose(1+verbosity) << "There's still some space left on stack: " << (SubOrder - bits) << "." << endl;
3469 // --2-- look at all added end pieces of this combination, construct bond subsets and sweep through a power set of these by recursion
3470 SP = RootDistance+1; // this is the next level
3471 // first count the members in the subset
3472 SubSetDimension = 0;
3473 Binder = FragmentSearch->BondsPerSPList[2*SP]; // start node for this level
3474 while (Binder->next != FragmentSearch->BondsPerSPList[2*SP+1]) { // compare to end node of this level
3475 Binder = Binder->next;
3476 for (int k=TouchedIndex;k--;) {
3477 if (Binder->Contains(TouchedList[k])) // if we added this very endpiece
3478 SubSetDimension++;
3479 }
3480 }
3481 // then allocate and fill the list
3482 BondsList = (bond **) Malloc(sizeof(bond *)*SubSetDimension, "molecule::SPFragmentGenerator: **BondsList");
3483 SubSetDimension = 0;
3484 Binder = FragmentSearch->BondsPerSPList[2*SP];
3485 while (Binder->next != FragmentSearch->BondsPerSPList[2*SP+1]) {
3486 Binder = Binder->next;
3487 for (int k=0;k<TouchedIndex;k++) {
3488 if (Binder->leftatom->nr == TouchedList[k]) // leftatom is always the close one
3489 BondsList[SubSetDimension++] = Binder;
3490 }
3491 }
3492 *out << Verbose(2+verbosity) << "Calling subset generator " << SP << " away from root " << *FragmentSearch->Root << " with sub set dimension " << SubSetDimension << "." << endl;
3493 SPFragmentGenerator(out, FragmentSearch, SP, BondsList, SubSetDimension, SubOrder-bits);
3494 Free((void **)&BondsList, "molecule::SPFragmentGenerator: **BondsList");
3495 } else {
3496 // --2-- otherwise store the complete fragment
3497 *out << Verbose(1+verbosity) << "Enough items on stack for a fragment!" << endl;
3498 // store fragment as a KeySet
3499 *out << Verbose(2) << "Found a new fragment[" << FragmentSearch->FragmentCounter << "], local nr.s are: ";
3500 for(KeySet::iterator runner = FragmentSearch->FragmentSet->begin(); runner != FragmentSearch->FragmentSet->end(); runner++)
3501 *out << (*runner) << " ";
3502 InsertFragmentIntoGraph(out, FragmentSearch);
3503 //Removal = LookForRemovalCandidate(out, FragmentSearch->FragmentSet, FragmentSearch->ShortestPathList);
3504 //Removal = StoreFragmentFromStack(out, FragmentSearch->Root, FragmentSearch->Leaflet, FragmentSearch->FragmentStack, FragmentSearch->ShortestPathList,FragmentSearch->Labels, &FragmentSearch->FragmentCounter, FragmentSearch->configuration);
3505 }
3506
3507 // --3-- remove all added items in this level from snake stack
3508 *out << Verbose(1+verbosity) << "Removing all items that were added on this SP level " << RootDistance << "." << endl;
3509 for(int j=0;j<TouchedIndex;j++) {
3510 Removal = TouchedList[j];
3511 *out << Verbose(2+verbosity) << "Removing item nr. " << Removal << " from snake stack." << endl;
3512 FragmentSearch->FragmentSet->erase(Removal);
3513 TouchedList[j] = -1;
3514 }
3515 *out << Verbose(2) << "Remaining local nr.s on snake stack are: ";
3516 for(KeySet::iterator runner = FragmentSearch->FragmentSet->begin(); runner != FragmentSearch->FragmentSet->end(); runner++)
3517 *out << (*runner) << " ";
3518 *out << endl;
3519 TouchedIndex = 0; // set Index to 0 for list of atoms added on this level
3520 } else {
3521 *out << Verbose(2+verbosity) << "More atoms to add for this set (" << bits << ") than space left on stack " << SubOrder << ", skipping this set." << endl;
3522 }
3523 }
3524 Free((void **)&TouchedList, "molecule::SPFragmentGenerator: *TouchedList");
3525 *out << Verbose(1+verbosity) << "End of SPFragmentGenerator, " << RootDistance << " away from Root " << *FragmentSearch->Root << " and SubOrder is " << SubOrder << "." << endl;
3526};
3527
3528/** Creates a list of all unique fragments of certain vertex size from a given graph \a Fragment for a given root vertex in the context of \a this molecule.
3529 * -# initialises UniqueFragments structure
3530 * -# fills edge list via BFS
3531 * -# creates the fragment by calling recursive function SPFragmentGenerator with UniqueFragments structure, 0 as
3532 root distance, the edge set, its dimension and the current suborder
3533 * -# Free'ing structure
3534 * Note that we may use the fact that the atoms are SP-ordered on the atomstack. I.e. when popping always the last, we first get all
3535 * with SP of 2, then those with SP of 3, then those with SP of 4 and so on.
3536 * \param *out output stream for debugging
3537 * \param Order bond order (limits BFS exploration and "number of digits" in power set generation
3538 * \param FragmentSearch UniqueFragments structure containing TEFactor, root atom and so on
3539 * \param RestrictedKeySet Restricted vertex set to use in context of molecule
3540 * \return number of inserted fragments
3541 * \note ShortestPathList in FragmentSearch structure is probably due to NumberOfAtomsSPLevel and SP not needed anymore
3542 */
3543int molecule::PowerSetGenerator(ofstream *out, int Order, struct UniqueFragments &FragmentSearch, KeySet RestrictedKeySet)
3544{
3545 int SP, UniqueIndex, AtomKeyNr;
3546 int *NumberOfAtomsSPLevel = (int *) Malloc(sizeof(int)*Order, "molecule::PowerSetGenerator: *SPLevelCount");
3547 atom *Walker = NULL, *OtherWalker = NULL;
3548 bond *Binder = NULL;
3549 bond **BondsList = NULL;
3550 KeyStack AtomStack;
3551 atom **PredecessorList = (atom **) Malloc(sizeof(atom *)*AtomCount, "molecule::PowerSetGenerator: **PredecessorList");
3552 int RootKeyNr = FragmentSearch.Root->nr;
3553 int Counter = FragmentSearch.FragmentCounter;
3554
3555 for (int i=AtomCount;i--;) {
3556 PredecessorList[i] = NULL;
3557 }
3558
3559 *out << endl;
3560 *out << Verbose(0) << "Begin of PowerSetGenerator with order " << Order << " at Root " << *FragmentSearch.Root << "." << endl;
3561
3562 UniqueIndex = 0;
3563 if (FragmentSearch.Labels[RootKeyNr] == -1)
3564 FragmentSearch.Labels[RootKeyNr] = UniqueIndex++;
3565 FragmentSearch.ShortestPathList[RootKeyNr] = 0;
3566 // prepare the atom stack counters (number of atoms with certain SP on stack)
3567 for (int i=Order;i--;)
3568 NumberOfAtomsSPLevel[i] = 0;
3569 NumberOfAtomsSPLevel[0] = 1; // for root
3570 SP = -1;
3571 *out << endl;
3572 *out << Verbose(0) << "Starting BFS analysis ..." << endl;
3573 // push as first on atom stack and goooo ...
3574 AtomStack.clear();
3575 AtomStack.push_back(RootKeyNr);
3576 *out << Verbose(0) << "Cleared AtomStack and pushed root as first item onto it." << endl;
3577 // do a BFS search to fill the SP lists and label the found vertices
3578 while (!AtomStack.empty()) {
3579 // pop next atom
3580 AtomKeyNr = AtomStack.front();
3581 AtomStack.pop_front();
3582 if (SP != -1)
3583 NumberOfAtomsSPLevel[SP]--;
3584 if ((SP == -1) || (NumberOfAtomsSPLevel[SP] == -1)) {
3585 SP++;
3586 NumberOfAtomsSPLevel[SP]--; // carry over "-1" to next level
3587 *out << Verbose(1) << "New SP level reached: " << SP << ", creating new SP list with 0 item(s)";
3588 if (SP > 0)
3589 *out << ", old level closed with " << FragmentSearch.BondsPerSPCount[SP-1] << " item(s)." << endl;
3590 else
3591 *out << "." << endl;
3592 FragmentSearch.BondsPerSPCount[SP] = 0;
3593 } else {
3594 *out << Verbose(1) << "Still " << NumberOfAtomsSPLevel[SP]+1 << " on this SP (" << SP << ") level, currently having " << FragmentSearch.BondsPerSPCount[SP] << " item(s)." << endl;
3595 }
3596 Walker = FindAtom(AtomKeyNr);
3597 *out << Verbose(0) << "Current Walker is: " << *Walker << " with nr " << Walker->nr << " and label " << FragmentSearch.Labels[AtomKeyNr] << " and SP of " << SP << "." << endl;
3598 // check for new sp level
3599 // go through all its bonds
3600 *out << Verbose(1) << "Going through all bonds of Walker." << endl;
3601 for (int i=0;i<NumberOfBondsPerAtom[AtomKeyNr];i++) {
3602 Binder = ListOfBondsPerAtom[AtomKeyNr][i];
3603 OtherWalker = Binder->GetOtherAtom(Walker);
3604 if ((RestrictedKeySet.find(OtherWalker->nr) != RestrictedKeySet.end())
3605#ifdef ADDHYDROGEN
3606 && (OtherWalker->type->Z != 1)
3607#endif
3608 ) { // skip hydrogens and restrict to fragment
3609 *out << Verbose(2) << "Current partner is " << *OtherWalker << " with nr " << OtherWalker->nr << " in bond " << *Binder << "." << endl;
3610 // set the label if not set (and push on root stack as well)
3611 if (FragmentSearch.Labels[OtherWalker->nr] == -1) {
3612 FragmentSearch.Labels[OtherWalker->nr] = UniqueIndex++;
3613 *out << Verbose(3) << "Set label to " << FragmentSearch.Labels[OtherWalker->nr] << "." << endl;
3614 } else {
3615 *out << Verbose(3) << "Label is already " << FragmentSearch.Labels[OtherWalker->nr] << "." << endl;
3616 }
3617 if ((OtherWalker != PredecessorList[AtomKeyNr]) && (OtherWalker->nr > RootKeyNr)) { // only pass through those with label bigger than Root's
3618 FragmentSearch.ShortestPathList[OtherWalker->nr] = SP+1;
3619 *out << Verbose(3) << "Set Shortest Path to " << FragmentSearch.ShortestPathList[OtherWalker->nr] << "." << endl;
3620 if (FragmentSearch.ShortestPathList[OtherWalker->nr] < Order) { // only pass through those within reach of Order
3621 // push for exploration on stack (only if the SP of OtherWalker is longer than of Walker! (otherwise it has been added already!)
3622 if (FragmentSearch.ShortestPathList[OtherWalker->nr] > SP) {
3623 if (FragmentSearch.ShortestPathList[OtherWalker->nr] < (Order-1)) {
3624 *out << Verbose(3) << "Putting on atom stack for further exploration." << endl;
3625 PredecessorList[OtherWalker->nr] = Walker; // note down the one further up the exploration tree
3626 AtomStack.push_back(OtherWalker->nr);
3627 NumberOfAtomsSPLevel[FragmentSearch.ShortestPathList[OtherWalker->nr]]++;
3628 } else {
3629 *out << Verbose(3) << "Not putting on atom stack, is at end of reach." << endl;
3630 }
3631 // add the bond in between to the SP list
3632 Binder = new bond(Walker, OtherWalker); // create a new bond in such a manner, that bond::rightatom is always the one more distant
3633 add(Binder, FragmentSearch.BondsPerSPList[2*SP+1]);
3634 FragmentSearch.BondsPerSPCount[SP]++;
3635 *out << Verbose(3) << "Added its bond to SP list, having now " << FragmentSearch.BondsPerSPCount[SP] << " item(s)." << endl;
3636 } else *out << Verbose(3) << "Not putting on atom stack, nor adding bond, as " << *OtherWalker << "'s SP is shorter than Walker's." << endl;
3637 } else *out << Verbose(3) << "Not continuing, as " << *OtherWalker << " is out of reach of order " << Order << "." << endl;
3638 } else *out << Verbose(3) << "Not passing on, as index of " << *OtherWalker << " " << OtherWalker->nr << " is smaller than that of Root " << RootKeyNr << " or this is my predecessor." << endl;
3639 } else *out << Verbose(2) << "Is not in the restricted keyset or skipping hydrogen " << *OtherWalker << "." << endl;
3640 }
3641 }
3642 // reset predecessor list
3643 for(int i=0;i<Order;i++) {
3644 Binder = FragmentSearch.BondsPerSPList[2*i];
3645 *out << Verbose(1) << "Current SP level is " << i << "." << endl;
3646 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
3647 Binder = Binder->next;
3648 PredecessorList[Binder->rightatom->nr] = NULL; // By construction "OtherAtom" is always Bond::rightatom!
3649 }
3650 }
3651 *out << endl;
3652
3653 // outputting all list for debugging
3654 *out << Verbose(0) << "Printing all found lists." << endl;
3655 for(int i=0;i<Order;i++) {
3656 Binder = FragmentSearch.BondsPerSPList[2*i];
3657 *out << Verbose(1) << "Current SP level is " << i << "." << endl;
3658 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
3659 Binder = Binder->next;
3660 *out << Verbose(2) << *Binder << endl;
3661 }
3662 }
3663
3664 // creating fragments with the found edge sets (may be done in reverse order, faster)
3665 SP = 0;
3666 for(int i=Order;i--;) { // sum up all found edges
3667 Binder = FragmentSearch.BondsPerSPList[2*i];
3668 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
3669 Binder = Binder->next;
3670 SP ++;
3671 }
3672 }
3673 *out << Verbose(0) << "Total number of edges is " << SP << "." << endl;
3674 if (SP >= (Order-1)) {
3675 // start with root (push on fragment stack)
3676 *out << Verbose(0) << "Starting fragment generation with " << *FragmentSearch.Root << ", local nr is " << FragmentSearch.Root->nr << "." << endl;
3677 FragmentSearch.FragmentSet->clear();
3678 FragmentSearch.FragmentSet->insert(FragmentSearch.Root->nr);
3679
3680 if (FragmentSearch.FragmentSet->size() == (unsigned int) Order) {
3681 *out << Verbose(0) << "Enough items on stack already for a fragment!" << endl;
3682 // store fragment as a KeySet
3683 *out << Verbose(2) << "Found a new fragment[" << FragmentSearch.FragmentCounter << "], local nr.s are: ";
3684 for(KeySet::iterator runner = FragmentSearch.FragmentSet->begin(); runner != FragmentSearch.FragmentSet->end(); runner++) {
3685 *out << (*runner) << " ";
3686 }
3687 *out << endl;
3688 InsertFragmentIntoGraph(out, &FragmentSearch);
3689 } else {
3690 *out << Verbose(0) << "Preparing subset for this root and calling generator." << endl;
3691 // prepare the subset and call the generator
3692 BondsList = (bond **) Malloc(sizeof(bond *)*FragmentSearch.BondsPerSPCount[0], "molecule::PowerSetGenerator: **BondsList");
3693 Binder = FragmentSearch.BondsPerSPList[0];
3694 for(int i=0;i<FragmentSearch.BondsPerSPCount[0];i++) {
3695 Binder = Binder->next;
3696 BondsList[i] = Binder;
3697 }
3698 SPFragmentGenerator(out, &FragmentSearch, 0, BondsList, FragmentSearch.BondsPerSPCount[0], Order-1);
3699 Free((void **)&BondsList, "molecule::PowerSetGenerator: **BondsList");
3700 }
3701 } else {
3702 *out << Verbose(0) << "Not enough total number of edges to build " << Order << "-body fragments." << endl;
3703 }
3704
3705 // as FragmentSearch structure is used only once, we don't have to clean it anymore
3706 // remove root from stack
3707 *out << Verbose(0) << "Removing root again from stack." << endl;
3708 FragmentSearch.FragmentSet->erase(FragmentSearch.Root->nr);
3709
3710 // free'ing the bonds lists
3711 *out << Verbose(0) << "Free'ing all found lists. and resetting index lists" << endl;
3712 for(int i=Order;i--;) {
3713 *out << Verbose(1) << "Current SP level is " << i << ": ";
3714 Binder = FragmentSearch.BondsPerSPList[2*i];
3715 while (Binder->next != FragmentSearch.BondsPerSPList[2*i+1]) {
3716 Binder = Binder->next;
3717 // *out << "Removing atom " << Binder->leftatom->nr << " and " << Binder->rightatom->nr << "." << endl; // make sure numbers are local
3718 FragmentSearch.ShortestPathList[Binder->leftatom->nr] = -1;
3719 FragmentSearch.ShortestPathList[Binder->rightatom->nr] = -1;
3720 }
3721 // delete added bonds
3722 cleanup(FragmentSearch.BondsPerSPList[2*i], FragmentSearch.BondsPerSPList[2*i+1]);
3723 // also start and end node
3724 *out << "cleaned." << endl;
3725 }
3726
3727 // free allocated memory
3728 Free((void **)&NumberOfAtomsSPLevel, "molecule::PowerSetGenerator: *SPLevelCount");
3729 Free((void **)&PredecessorList, "molecule::PowerSetGenerator: **PredecessorList");
3730
3731 // return list
3732 *out << Verbose(0) << "End of PowerSetGenerator." << endl;
3733 return (FragmentSearch.FragmentCounter - Counter);
3734};
3735
3736/** Corrects the nuclei position if the fragment was created over the cell borders.
3737 * Scans all bonds, checks the distance, if greater than typical, we have a candidate for the correction.
3738 * We remove the bond whereafter the graph probably separates. Then, we translate the one component periodically
3739 * and re-add the bond. Looping on the distance check.
3740 * \param *out ofstream for debugging messages
3741 */
3742void molecule::ScanForPeriodicCorrection(ofstream *out)
3743{
3744 bond *Binder = NULL;
3745 bond *OtherBinder = NULL;
3746 atom *Walker = NULL;
3747 atom *OtherWalker = NULL;
3748 double *matrix = ReturnFullMatrixforSymmetric(cell_size);
3749 enum Shading *ColorList = NULL;
3750 double tmp;
3751 vector TranslationVector;
3752 //class StackClass<atom *> *CompStack = NULL;
3753 class StackClass<atom *> *AtomStack = new StackClass<atom *>(AtomCount);
3754 bool flag = true;
3755
3756 *out << Verbose(2) << "Begin of ScanForPeriodicCorrection." << endl;
3757
3758 ColorList = (enum Shading *) Malloc(sizeof(enum Shading)*AtomCount, "molecule::ScanForPeriodicCorrection: *ColorList");
3759 while (flag) {
3760 // remove bonds that are beyond bonddistance
3761 for(int i=NDIM;i--;)
3762 TranslationVector.x[i] = 0.;
3763 // scan all bonds
3764 Binder = first;
3765 flag = false;
3766 while ((!flag) && (Binder->next != last)) {
3767 Binder = Binder->next;
3768 for (int i=NDIM;i--;) {
3769 tmp = fabs(Binder->leftatom->x.x[i] - Binder->rightatom->x.x[i]);
3770 //*out << Verbose(3) << "Checking " << i << "th distance of " << *Binder->leftatom << " to " << *Binder->rightatom << ": " << tmp << "." << endl;
3771 if (tmp > BondDistance) {
3772 OtherBinder = Binder->next; // note down binding partner for later re-insertion
3773 unlink(Binder); // unlink bond
3774 //*out << Verbose(2) << "Correcting at bond " << *Binder << "." << endl;
3775 flag = true;
3776 break;
3777 }
3778 }
3779 }
3780 if (flag) {
3781 // create translation vector from their periodically modified distance
3782 for (int i=NDIM;i--;) {
3783 tmp = Binder->leftatom->x.x[i] - Binder->rightatom->x.x[i];
3784 if (fabs(tmp) > BondDistance)
3785 TranslationVector.x[i] = (tmp < 0) ? +1. : -1.;
3786 }
3787 TranslationVector.MatrixMultiplication(matrix);
3788 *out << Verbose(3) << "Translation vector is ";
3789 TranslationVector.Output(out);
3790 *out << endl;
3791 // apply to all atoms of first component via BFS
3792 for (int i=AtomCount;i--;)
3793 ColorList[i] = white;
3794 AtomStack->Push(Binder->leftatom);
3795 while (!AtomStack->IsEmpty()) {
3796 Walker = AtomStack->PopFirst();
3797 //*out << Verbose (3) << "Current Walker is: " << *Walker << "." << endl;
3798 ColorList[Walker->nr] = black; // mark as explored
3799 Walker->x.AddVector(&TranslationVector); // translate
3800 for (int i=0;i<NumberOfBondsPerAtom[Walker->nr];i++) { // go through all binding partners
3801 if (ListOfBondsPerAtom[Walker->nr][i] != Binder) {
3802 OtherWalker = ListOfBondsPerAtom[Walker->nr][i]->GetOtherAtom(Walker);
3803 if (ColorList[OtherWalker->nr] == white) {
3804 AtomStack->Push(OtherWalker); // push if yet unexplored
3805 }
3806 }
3807 }
3808 }
3809 // re-add bond
3810 link(Binder, OtherBinder);
3811 } else {
3812 *out << Verbose(3) << "No corrections for this fragment." << endl;
3813 }
3814 //delete(CompStack);
3815 }
3816
3817 // free allocated space from ReturnFullMatrixforSymmetric()
3818 delete(AtomStack);
3819 Free((void **)&ColorList, "molecule::ScanForPeriodicCorrection: *ColorList");
3820 Free((void **)&matrix, "molecule::ScanForPeriodicCorrection: *matrix");
3821 *out << Verbose(2) << "End of ScanForPeriodicCorrection." << endl;
3822};
3823
3824/** Blows the 6-dimensional \a cell_size array up to a full NDIM by NDIM matrix.
3825 * \param *symm 6-dim array of unique symmetric matrix components
3826 * \return allocated NDIM*NDIM array with the symmetric matrix
3827 */
3828double * molecule::ReturnFullMatrixforSymmetric(double *symm)
3829{
3830 double *matrix = (double *) Malloc(sizeof(double)*NDIM*NDIM, "molecule::ReturnFullMatrixforSymmetric: *matrix");
3831 matrix[0] = symm[0];
3832 matrix[1] = symm[1];
3833 matrix[2] = symm[3];
3834 matrix[3] = symm[1];
3835 matrix[4] = symm[2];
3836 matrix[5] = symm[4];
3837 matrix[6] = symm[3];
3838 matrix[7] = symm[4];
3839 matrix[8] = symm[5];
3840 return matrix;
3841};
3842
3843bool KeyCompare::operator() (const KeySet SubgraphA, const KeySet SubgraphB) const
3844{
3845 //cout << "my check is used." << endl;
3846 if (SubgraphA.size() < SubgraphB.size()) {
3847 return true;
3848 } else {
3849 if (SubgraphA.size() > SubgraphB.size()) {
3850 return false;
3851 } else {
3852 KeySet::iterator IteratorA = SubgraphA.begin();
3853 KeySet::iterator IteratorB = SubgraphB.begin();
3854 while ((IteratorA != SubgraphA.end()) && (IteratorB != SubgraphB.end())) {
3855 if ((*IteratorA) < (*IteratorB))
3856 return true;
3857 else if ((*IteratorA) > (*IteratorB)) {
3858 return false;
3859 } // else, go on to next index
3860 IteratorA++;
3861 IteratorB++;
3862 } // end of while loop
3863 }// end of check in case of equal sizes
3864 }
3865 return false; // if we reach this point, they are equal
3866};
3867
3868//bool operator < (KeySet SubgraphA, KeySet SubgraphB)
3869//{
3870// return KeyCompare(SubgraphA, SubgraphB);
3871//};
3872
3873/** Checking whether KeySet is not already present in Graph, if so just adds factor.
3874 * \param *out output stream for debugging
3875 * \param &set KeySet to insert
3876 * \param &graph Graph to insert into
3877 * \param *counter pointer to unique fragment count
3878 * \param factor energy factor for the fragment
3879 */
3880inline void InsertFragmentIntoGraph(ofstream *out, struct UniqueFragments *Fragment)
3881{
3882 GraphTestPair testGraphInsert;
3883
3884 testGraphInsert = Fragment->Leaflet->insert(GraphPair (*Fragment->FragmentSet,pair<int,double>(Fragment->FragmentCounter,Fragment->TEFactor))); // store fragment number and current factor
3885 if (testGraphInsert.second) {
3886 *out << Verbose(2) << "KeySet " << Fragment->FragmentCounter << " successfully inserted." << endl;
3887 Fragment->FragmentCounter++;
3888 } else {
3889 *out << Verbose(2) << "KeySet " << Fragment->FragmentCounter << " failed to insert, present fragment is " << ((*(testGraphInsert.first)).second).first << endl;
3890 ((*(testGraphInsert.first)).second).second += Fragment->TEFactor; // increase the "created" counter
3891 *out << Verbose(2) << "New factor is " << ((*(testGraphInsert.first)).second).second << "." << endl;
3892 }
3893};
3894//void inline InsertIntoGraph(ofstream *out, KeyStack &stack, Graph &graph, int *counter, double factor)
3895//{
3896// // copy stack contents to set and call overloaded function again
3897// KeySet set;
3898// for(KeyStack::iterator runner = stack.begin(); runner != stack.begin(); runner++)
3899// set.insert((*runner));
3900// InsertIntoGraph(out, set, graph, counter, factor);
3901//};
3902
3903/** Inserts each KeySet in \a graph2 into \a graph1.
3904 * \param *out output stream for debugging
3905 * \param graph1 first (dest) graph
3906 * \param graph2 second (source) graph
3907 * \param *counter keyset counter that gets increased
3908 */
3909inline void InsertGraphIntoGraph(ofstream *out, Graph &graph1, Graph &graph2, int *counter)
3910{
3911 GraphTestPair testGraphInsert;
3912
3913 for(Graph::iterator runner = graph2.begin(); runner != graph2.end(); runner++) {
3914 testGraphInsert = graph1.insert(GraphPair ((*runner).first,pair<int,double>((*counter)++,((*runner).second).second))); // store fragment number and current factor
3915 if (testGraphInsert.second) {
3916 *out << Verbose(2) << "KeySet " << (*counter)-1 << " successfully inserted." << endl;
3917 } else {
3918 *out << Verbose(2) << "KeySet " << (*counter)-1 << " failed to insert, present fragment is " << ((*(testGraphInsert.first)).second).first << endl;
3919 ((*(testGraphInsert.first)).second).second += (*runner).second.second;
3920 *out << Verbose(2) << "New factor is " << (*(testGraphInsert.first)).second.second << "." << endl;
3921 }
3922 }
3923};
3924
3925
3926/** Performs BOSSANOVA decomposition at selected sites, increasing the cutoff by one at these sites.
3927 * -# constructs a complete keyset of the molecule
3928 * -# In a loop over all possible roots from the given rootstack
3929 * -# increases order of root site
3930 * -# calls PowerSetGenerator with this order, the complete keyset and the rootkeynr
3931 * -# for all consecutive lower levels PowerSetGenerator is called with the suborder, the higher order keyset
3932as the restricted one and each site in the set as the root)
3933 * -# these are merged into a fragment list of keysets
3934 * -# All fragment lists (for all orders, i.e. from all destination fields) are merged into one list for return
3935 * Important only is that we create all fragments, it is not important if we create them more than once
3936 * as these copies are filtered out via use of the hash table (KeySet).
3937 * \param *out output stream for debugging
3938 * \param Fragment&*List list of already present keystacks (adaptive scheme) or empty list
3939 * \param &RootStack stack with all root candidates (unequal to each atom in complete molecule if adaptive scheme is applied)
3940 * \param *MinimumRingSize minimum ring size for each atom (molecule::Atomcount)
3941 * \return pointer to Graph list
3942 */
3943void molecule::FragmentBOSSANOVA(ofstream *out, Graph *&FragmentList, KeyStack &RootStack, int *MinimumRingSize)
3944{
3945 Graph ***FragmentLowerOrdersList = NULL;
3946 int NumLevels, NumMolecules, TotalNumMolecules = 0, *NumMoleculesOfOrder = NULL;
3947 int counter = 0, Order;
3948 int UpgradeCount = RootStack.size();
3949 KeyStack FragmentRootStack;
3950 int RootKeyNr, RootNr;
3951 struct UniqueFragments FragmentSearch;
3952
3953 *out << Verbose(0) << "Begin of FragmentBOSSANOVA." << endl;
3954
3955 // FragmentLowerOrdersList is a 2D-array of pointer to MoleculeListClass objects, one dimension represents the ANOVA expansion of a single order (i.e. 5)
3956 // with all needed lower orders that are subtracted, the other dimension is the BondOrder (i.e. from 1 to 5)
3957 NumMoleculesOfOrder = (int *) Malloc(sizeof(int)*UpgradeCount, "molecule::FragmentBOSSANOVA: *NumMoleculesOfOrder");
3958 FragmentLowerOrdersList = (Graph ***) Malloc(sizeof(Graph **)*UpgradeCount, "molecule::FragmentBOSSANOVA: ***FragmentLowerOrdersList");
3959
3960 // initialise the fragments structure
3961 FragmentSearch.ShortestPathList = (int *) Malloc(sizeof(int)*AtomCount, "molecule::PowerSetGenerator: *ShortestPathList");
3962 FragmentSearch.Labels = (int *) Malloc(sizeof(int)*AtomCount, "molecule::PowerSetGenerator: *Labels");
3963 FragmentSearch.FragmentCounter = 0;
3964 FragmentSearch.FragmentSet = new KeySet;
3965 FragmentSearch.Root = FindAtom(RootKeyNr);
3966 for (int i=AtomCount;i--;) {
3967 FragmentSearch.Labels[i] = -1;
3968 FragmentSearch.ShortestPathList[i] = -1;
3969 }
3970
3971 // Construct the complete KeySet which we need for topmost level only (but for all Roots)
3972 atom *Walker = start;
3973 KeySet CompleteMolecule;
3974 while (Walker->next != end) {
3975 Walker = Walker->next;
3976 CompleteMolecule.insert(Walker->GetTrueFather()->nr);
3977 }
3978
3979 // this can easily be seen: if Order is 5, then the number of levels for each lower order is the total sum of the number of levels above, as
3980 // each has to be split up. E.g. for the second level we have one from 5th, one from 4th, two from 3th (which in turn is one from 5th, one from 4th),
3981 // hence we have overall four 2th order levels for splitting. This also allows for putting all into a single array (FragmentLowerOrdersList[])
3982 // with the order along the cells as this: 5433222211111111 for BondOrder 5 needing 16=pow(2,5-1) cells (only we use bit-shifting which is faster)
3983 RootNr = 0; // counts through the roots in RootStack
3984 while ((RootNr < UpgradeCount) && (!RootStack.empty())) {
3985 RootKeyNr = RootStack.front();
3986 RootStack.pop_front();
3987 Walker = FindAtom(RootKeyNr);
3988 // check cyclic lengths
3989 if ((MinimumRingSize[Walker->GetTrueFather()->nr] != -1) && (Walker->GetTrueFather()->AdaptiveOrder+1 > MinimumRingSize[Walker->GetTrueFather()->nr])) {
3990 *out << Verbose(0) << "Bond order " << Walker->GetTrueFather()->AdaptiveOrder << " of Root " << *Walker << " greater than or equal to Minimum Ring size of " << MinimumRingSize << " found is not allowed." << endl;
3991 } else {
3992 // increase adaptive order by one
3993 Walker->GetTrueFather()->AdaptiveOrder++;
3994 Order = Walker->AdaptiveOrder = Walker->GetTrueFather()->AdaptiveOrder;
3995
3996 // initialise Order-dependent entries of UniqueFragments structure
3997 FragmentSearch.BondsPerSPList = (bond **) Malloc(sizeof(bond *)*Order*2, "molecule::PowerSetGenerator: ***BondsPerSPList");
3998 FragmentSearch.BondsPerSPCount = (int *) Malloc(sizeof(int)*Order, "molecule::PowerSetGenerator: *BondsPerSPCount");
3999 for (int i=Order;i--;) {
4000 FragmentSearch.BondsPerSPList[2*i] = new bond(); // start node
4001 FragmentSearch.BondsPerSPList[2*i+1] = new bond(); // end node
4002 FragmentSearch.BondsPerSPList[2*i]->next = FragmentSearch.BondsPerSPList[2*i+1]; // intertwine these two
4003 FragmentSearch.BondsPerSPList[2*i+1]->previous = FragmentSearch.BondsPerSPList[2*i];
4004 FragmentSearch.BondsPerSPCount[i] = 0;
4005 }
4006
4007 // allocate memory for all lower level orders in this 1D-array of ptrs
4008 NumLevels = 1 << (Order-1); // (int)pow(2,Order);
4009 FragmentLowerOrdersList[RootNr] = (Graph **) Malloc(sizeof(Graph *)*NumLevels, "molecule::FragmentBOSSANOVA: **FragmentLowerOrdersList[]");
4010
4011 // create top order where nothing is reduced
4012 *out << Verbose(0) << "==============================================================================================================" << endl;
4013 *out << Verbose(0) << "Creating KeySets of Bond Order " << Order << " for " << *Walker << ", NumLevels is " << NumLevels << ", " << (RootStack.size()-RootNr-1) << " Roots remaining." << endl;
4014
4015 // Create list of Graphs of current Bond Order (i.e. F_{ij})
4016 FragmentLowerOrdersList[RootNr][0] = new Graph;
4017 FragmentSearch.TEFactor = 1.;
4018 FragmentSearch.Leaflet = FragmentLowerOrdersList[RootNr][0]; // set to insertion graph
4019 FragmentSearch.Root = Walker;
4020 NumMoleculesOfOrder[RootNr] = PowerSetGenerator(out, Walker->AdaptiveOrder, FragmentSearch, CompleteMolecule);
4021 *out << Verbose(1) << "Number of resulting KeySets is: " << NumMoleculesOfOrder[RootNr] << "." << endl;
4022 NumMolecules = 0;
4023
4024 if ((NumLevels >> 1) > 0) {
4025 // create lower order fragments
4026 *out << Verbose(0) << "Creating list of unique fragments of lower Bond Order terms to be subtracted." << endl;
4027 Order = Walker->AdaptiveOrder;
4028 for (int source=0;source<(NumLevels >> 1);source++) { // 1-terms don't need any more splitting, that's why only half is gone through (shift again)
4029 // step down to next order at (virtual) boundary of powers of 2 in array
4030 while (source >= (1 << (Walker->AdaptiveOrder-Order))) // (int)pow(2,Walker->AdaptiveOrder-Order))
4031 Order--;
4032 *out << Verbose(0) << "Current Order is: " << Order << "." << endl;
4033 for (int SubOrder=Order-1;SubOrder>0;SubOrder--) {
4034 int dest = source + (1 << (Walker->AdaptiveOrder-(SubOrder+1)));
4035 *out << Verbose(0) << "--------------------------------------------------------------------------------------------------------------" << endl;
4036 *out << Verbose(0) << "Current SubOrder is: " << SubOrder << " with source " << source << " to destination " << dest << "." << endl;
4037
4038 // every molecule is split into a list of again (Order - 1) molecules, while counting all molecules
4039 //*out << Verbose(1) << "Splitting the " << (*FragmentLowerOrdersList[RootNr][source]).size() << " molecules of the " << source << "th cell in the array." << endl;
4040 //NumMolecules = 0;
4041 FragmentLowerOrdersList[RootNr][dest] = new Graph;
4042 for(Graph::iterator runner = (*FragmentLowerOrdersList[RootNr][source]).begin();runner != (*FragmentLowerOrdersList[RootNr][source]).end(); runner++) {
4043 for (KeySet::iterator sprinter = (*runner).first.begin();sprinter != (*runner).first.end(); sprinter++) {
4044 Graph TempFragmentList;
4045 FragmentSearch.TEFactor = -(*runner).second.second;
4046 FragmentSearch.Leaflet = &TempFragmentList; // set to insertion graph
4047 FragmentSearch.Root = FindAtom(*sprinter);
4048 NumMoleculesOfOrder[RootNr] += PowerSetGenerator(out, SubOrder, FragmentSearch, (*runner).first);
4049 // insert new keysets FragmentList into FragmentLowerOrdersList[Walker->AdaptiveOrder-1][dest]
4050 *out << Verbose(1) << "Merging resulting key sets with those present in destination " << dest << "." << endl;
4051 InsertGraphIntoGraph(out, *FragmentLowerOrdersList[RootNr][dest], TempFragmentList, &NumMolecules);
4052 }
4053 }
4054 *out << Verbose(1) << "Number of resulting molecules for SubOrder " << SubOrder << " is: " << NumMolecules << "." << endl;
4055 }
4056 }
4057 }
4058 // now, we have completely filled each cell of FragmentLowerOrdersList[] for the current Walker->AdaptiveOrder
4059 //NumMoleculesOfOrder[Walker->AdaptiveOrder-1] = NumMolecules;
4060 TotalNumMolecules += NumMoleculesOfOrder[RootNr];
4061 *out << Verbose(1) << "Number of resulting molecules for Order " << (int)Walker->GetTrueFather()->AdaptiveOrder << " is: " << NumMoleculesOfOrder[RootNr] << "." << endl;
4062 RootStack.push_back(RootKeyNr); // put back on stack
4063 RootNr++;
4064
4065 // free Order-dependent entries of UniqueFragments structure for next loop cycle
4066 Free((void **)&FragmentSearch.BondsPerSPCount, "molecule::PowerSetGenerator: *BondsPerSPCount");
4067 for (int i=Order;i--;) {
4068 delete(FragmentSearch.BondsPerSPList[2*i]);
4069 delete(FragmentSearch.BondsPerSPList[2*i+1]);
4070 }
4071 Free((void **)&FragmentSearch.BondsPerSPList, "molecule::PowerSetGenerator: ***BondsPerSPList");
4072 }
4073 }
4074 *out << Verbose(0) << "==============================================================================================================" << endl;
4075 *out << Verbose(1) << "Total number of resulting molecules is: " << TotalNumMolecules << "." << endl;
4076 *out << Verbose(0) << "==============================================================================================================" << endl;
4077
4078 // cleanup FragmentSearch structure
4079 Free((void **)&FragmentSearch.ShortestPathList, "molecule::PowerSetGenerator: *ShortestPathList");
4080 Free((void **)&FragmentSearch.Labels, "molecule::PowerSetGenerator: *Labels");
4081 delete(FragmentSearch.FragmentSet);
4082
4083 // now, FragmentLowerOrdersList is complete, it looks - for BondOrder 5 - as this (number is the ANOVA Order of the terms therein)
4084 // 5433222211111111
4085 // 43221111
4086 // 3211
4087 // 21
4088 // 1
4089
4090 // Subsequently, we combine all into a single list (FragmentList)
4091
4092 *out << Verbose(0) << "Combining the lists of all orders per order and finally into a single one." << endl;
4093 if (FragmentList == NULL) {
4094 FragmentList = new Graph;
4095 counter = 0;
4096 } else {
4097 counter = FragmentList->size();
4098 }
4099 RootNr = 0;
4100 while (!RootStack.empty()) {
4101 RootKeyNr = RootStack.front();
4102 RootStack.pop_front();
4103 Walker = FindAtom(RootKeyNr);
4104 NumLevels = 1 << (Walker->AdaptiveOrder - 1);
4105 for(int i=0;i<NumLevels;i++) {
4106 InsertGraphIntoGraph(out, *FragmentList, (*FragmentLowerOrdersList[RootNr][i]), &counter);
4107 delete(FragmentLowerOrdersList[RootNr][i]);
4108 }
4109 Free((void **)&FragmentLowerOrdersList[RootNr], "molecule::FragmentBOSSANOVA: **FragmentLowerOrdersList[]");
4110 RootNr++;
4111 }
4112 Free((void **)&FragmentLowerOrdersList, "molecule::FragmentBOSSANOVA: ***FragmentLowerOrdersList");
4113 Free((void **)&NumMoleculesOfOrder, "molecule::FragmentBOSSANOVA: *NumMoleculesOfOrder");
4114
4115 *out << Verbose(0) << "End of FragmentBOSSANOVA." << endl;
4116};
4117
4118/** Comparison function for GSL heapsort on distances in two molecules.
4119 * \param *a
4120 * \param *b
4121 * \return <0, \a *a less than \a *b, ==0 if equal, >0 \a *a greater than \a *b
4122 */
4123inline int CompareDoubles (const void * a, const void * b)
4124{
4125 if (*(double *)a > *(double *)b)
4126 return -1;
4127 else if (*(double *)a < *(double *)b)
4128 return 1;
4129 else
4130 return 0;
4131};
4132
4133/** Determines whether two molecules actually contain the same atoms and coordination.
4134 * \param *out output stream for debugging
4135 * \param *OtherMolecule the molecule to compare this one to
4136 * \param threshold upper limit of difference when comparing the coordination.
4137 * \return NULL - not equal, otherwise an allocated (molecule::AtomCount) permutation map of the atom numbers (which corresponds to which)
4138 */
4139int * molecule::IsEqualToWithinThreshold(ofstream *out, molecule *OtherMolecule, double threshold)
4140{
4141 int flag;
4142 double *Distances = NULL, *OtherDistances = NULL;
4143 vector CenterOfGravity, OtherCenterOfGravity;
4144 size_t *PermMap = NULL, *OtherPermMap = NULL;
4145 int *PermutationMap = NULL;
4146 atom *Walker = NULL;
4147 bool result = true; // status of comparison
4148
4149 *out << Verbose(3) << "Begin of IsEqualToWithinThreshold." << endl;
4150 /// first count both their atoms and elements and update lists thereby ...
4151 //*out << Verbose(0) << "Counting atoms, updating list" << endl;
4152 CountAtoms(out);
4153 OtherMolecule->CountAtoms(out);
4154 CountElements();
4155 OtherMolecule->CountElements();
4156
4157 /// ... and compare:
4158 /// -# AtomCount
4159 if (result) {
4160 if (AtomCount != OtherMolecule->AtomCount) {
4161 *out << Verbose(4) << "AtomCounts don't match: " << AtomCount << " == " << OtherMolecule->AtomCount << endl;
4162 result = false;
4163 } else *out << Verbose(4) << "AtomCounts match: " << AtomCount << " == " << OtherMolecule->AtomCount << endl;
4164 }
4165 /// -# ElementCount
4166 if (result) {
4167 if (ElementCount != OtherMolecule->ElementCount) {
4168 *out << Verbose(4) << "ElementCount don't match: " << ElementCount << " == " << OtherMolecule->ElementCount << endl;
4169 result = false;
4170 } else *out << Verbose(4) << "ElementCount match: " << ElementCount << " == " << OtherMolecule->ElementCount << endl;
4171 }
4172 /// -# ElementsInMolecule
4173 if (result) {
4174 for (flag=MAX_ELEMENTS;flag--;) {
4175 //*out << Verbose(5) << "Element " << flag << ": " << ElementsInMolecule[flag] << " <-> " << OtherMolecule->ElementsInMolecule[flag] << "." << endl;
4176 if (ElementsInMolecule[flag] != OtherMolecule->ElementsInMolecule[flag])
4177 break;
4178 }
4179 if (flag < MAX_ELEMENTS) {
4180 *out << Verbose(4) << "ElementsInMolecule don't match." << endl;
4181 result = false;
4182 } else *out << Verbose(4) << "ElementsInMolecule match." << endl;
4183 }
4184 /// then determine and compare center of gravity for each molecule ...
4185 if (result) {
4186 *out << Verbose(5) << "Calculating Centers of Gravity" << endl;
4187 DetermineCenter(CenterOfGravity);
4188 OtherMolecule->DetermineCenter(OtherCenterOfGravity);
4189 *out << Verbose(5) << "Center of Gravity: ";
4190 CenterOfGravity.Output(out);
4191 *out << endl << Verbose(5) << "Other Center of Gravity: ";
4192 OtherCenterOfGravity.Output(out);
4193 *out << endl;
4194 if (CenterOfGravity.Distance(&OtherCenterOfGravity) > threshold) {
4195 *out << Verbose(4) << "Centers of gravity don't match." << endl;
4196 result = false;
4197 }
4198 }
4199
4200 /// ... then make a list with the euclidian distance to this center for each atom of both molecules
4201 if (result) {
4202 *out << Verbose(5) << "Calculating distances" << endl;
4203 Distances = (double *) Malloc(sizeof(double)*AtomCount, "molecule::IsEqualToWithinThreshold: Distances");
4204 OtherDistances = (double *) Malloc(sizeof(double)*AtomCount, "molecule::IsEqualToWithinThreshold: OtherDistances");
4205 Walker = start;
4206 while (Walker->next != end) {
4207 Walker = Walker->next;
4208 Distances[Walker->nr] = CenterOfGravity.Distance(&Walker->x);
4209 }
4210 Walker = OtherMolecule->start;
4211 while (Walker->next != OtherMolecule->end) {
4212 Walker = Walker->next;
4213 OtherDistances[Walker->nr] = OtherCenterOfGravity.Distance(&Walker->x);
4214 }
4215
4216 /// ... sort each list (using heapsort (o(N log N)) from GSL)
4217 *out << Verbose(5) << "Sorting distances" << endl;
4218 PermMap = (size_t *) Malloc(sizeof(size_t)*AtomCount, "molecule::IsEqualToWithinThreshold: *PermMap");
4219 OtherPermMap = (size_t *) Malloc(sizeof(size_t)*AtomCount, "molecule::IsEqualToWithinThreshold: *OtherPermMap");
4220 gsl_heapsort_index (PermMap, Distances, AtomCount, sizeof(double), CompareDoubles);
4221 gsl_heapsort_index (OtherPermMap, OtherDistances, AtomCount, sizeof(double), CompareDoubles);
4222 PermutationMap = (int *) Malloc(sizeof(int)*AtomCount, "molecule::IsEqualToWithinThreshold: *PermutationMap");
4223 *out << Verbose(5) << "Combining Permutation Maps" << endl;
4224 for(int i=AtomCount;i--;)
4225 PermutationMap[PermMap[i]] = (int) OtherPermMap[i];
4226
4227 /// ... and compare them step by step, whether the difference is individiually(!) below \a threshold for all
4228 *out << Verbose(4) << "Comparing distances" << endl;
4229 flag = 0;
4230 for (int i=0;i<AtomCount;i++) {
4231 *out << Verbose(5) << "Distances: |" << Distances[PermMap[i]] << " - " << OtherDistances[OtherPermMap[i]] << "| = " << fabs(Distances[PermMap[i]] - OtherDistances[OtherPermMap[i]]) << " ?<? " << threshold << endl;
4232 if (fabs(Distances[PermMap[i]] - OtherDistances[OtherPermMap[i]]) > threshold)
4233 flag = 1;
4234 }
4235 Free((void **)&PermMap, "molecule::IsEqualToWithinThreshold: *PermMap");
4236 Free((void **)&OtherPermMap, "molecule::IsEqualToWithinThreshold: *OtherPermMap");
4237
4238 /// free memory
4239 Free((void **)&Distances, "molecule::IsEqualToWithinThreshold: Distances");
4240 Free((void **)&OtherDistances, "molecule::IsEqualToWithinThreshold: OtherDistances");
4241 if (flag) { // if not equal
4242 Free((void **)&PermutationMap, "molecule::IsEqualToWithinThreshold: *PermutationMap");
4243 result = false;
4244 }
4245 }
4246 /// return pointer to map if all distances were below \a threshold
4247 *out << Verbose(3) << "End of IsEqualToWithinThreshold." << endl;
4248 if (result) {
4249 *out << Verbose(3) << "Result: Equal." << endl;
4250 return PermutationMap;
4251 } else {
4252 *out << Verbose(3) << "Result: Not equal." << endl;
4253 return NULL;
4254 }
4255};
4256
4257/** Returns an index map for two father-son-molecules.
4258 * The map tells which atom in this molecule corresponds to which one in the other molecul with their fathers.
4259 * \param *out output stream for debugging
4260 * \param *OtherMolecule corresponding molecule with fathers
4261 * \return allocated map of size molecule::AtomCount with map
4262 * \todo make this with a good sort O(n), not O(n^2)
4263 */
4264int * molecule::GetFatherSonAtomicMap(ofstream *out, molecule *OtherMolecule)
4265{
4266 atom *Walker = NULL, *OtherWalker = NULL;
4267 *out << Verbose(3) << "Begin of GetFatherAtomicMap." << endl;
4268 int *AtomicMap = (int *) Malloc(sizeof(int)*AtomCount, "molecule::GetAtomicMap: *AtomicMap"); //Calloc
4269 for (int i=AtomCount;i--;)
4270 AtomicMap[i] = -1;
4271 if (OtherMolecule == this) { // same molecule
4272 for (int i=AtomCount;i--;) // no need as -1 means already that there is trivial correspondence
4273 AtomicMap[i] = i;
4274 *out << Verbose(4) << "Map is trivial." << endl;
4275 } else {
4276 *out << Verbose(4) << "Map is ";
4277 Walker = start;
4278 while (Walker->next != end) {
4279 Walker = Walker->next;
4280 if (Walker->father == NULL) {
4281 AtomicMap[Walker->nr] = -2;
4282 } else {
4283 OtherWalker = OtherMolecule->start;
4284 while (OtherWalker->next != OtherMolecule->end) {
4285 OtherWalker = OtherWalker->next;
4286 //for (int i=0;i<AtomCount;i++) { // search atom
4287 //for (int j=0;j<OtherMolecule->AtomCount;j++) {
4288 //*out << Verbose(4) << "Comparing father " << Walker->father << " with the other one " << OtherWalker->father << "." << endl;
4289 if (Walker->father == OtherWalker)
4290 AtomicMap[Walker->nr] = OtherWalker->nr;
4291 }
4292 }
4293 *out << AtomicMap[Walker->nr] << "\t";
4294 }
4295 *out << endl;
4296 }
4297 *out << Verbose(3) << "End of GetFatherAtomicMap." << endl;
4298 return AtomicMap;
4299};
4300
Note: See TracBrowser for help on using the repository browser.