Changeset db0833 for doc/userguide/userguide.xml

Timestamp:

May 8, 2017, 2:03:59 PM (8 years ago)

Author:

Frederik Heber <frederik.heber@…>

Branches:

ForceAnnealing_goodresults, ForceAnnealing_tocheck

Children:

485eea

Parents:

c8165c

git-author:

Frederik Heber <heber@…> (04/02/17 19:23:15)

git-committer:

Frederik Heber <frederik.heber@…> (05/08/17 14:03:59)

Message:

Added RandomPerturbationAction.

• this should be when doing structural optimization for getting given configurations out of ambivalent states, e.g. a water molecule whose three atoms define a straight line.
• note that we need to use a sensible random number distribution, the default is not doing what's expected. Added to docu.
• TESTS: Added regression test with undo/redo.
• DOCU: Added documentation to userguide.

File:

• doc/userguide/userguide.xml (modified) (1 diff)

doc/userguide/userguide.xml

@@ -879 +879 @@
         <link linkend='geometry'>Geometry</link>.</para>
         </section>
+        <section xml:id="atoms.random-perturbation">
+          <title xml:id="atoms.random-perturbation.title">Randomly perturb atoms</title>
+          <para>The positions of a set of selected atom(s) can be randomly 
+          perturbed by giving a maximum noise level..</para>
+          <programlisting>
+          ... --random-number-distribution "uniform_01"
+      --random-perturbation 0.1</programlisting>
+          <para>This will perturb all atomic positions by adding a vector
+          with components chosen randomly from the interval [-level, level].</para>
+          <para>Note that the manner these random numbers are picked depends
+          on the current random number distribution (and also on the engine),
+          see <link linkend='randomization'>Randomization</link>. The default
+          one will not give good results, therefore the example uses the 
+          uniform_01 distribution</para>
+        </section>
         <section xml:id="atoms.remove-atom">
           <title xml:id="atoms.remove-atom.title">Removing atoms</title>