Changeset dce5a3 for doc/userguide/userguide.xml

Timestamp:

Dec 21, 2025, 9:50:22 PM (3 weeks ago)

Author:

Frederik Heber <frederik.heber@…>

Branches:

Candidate_v1.7.1, stable

Children:

bc58f2

Parents:

44d5d9

git-author:

Frederik Heber <frederik.heber@…> (11/20/25 18:02:16)

git-committer:

Frederik Heber <frederik.heber@…> (12/21/25 21:50:22)

Message:

Adds AddSelectedAtomsAsFragmentAction.

• this action allows calculating the energy of an entire molecule without relying on BOSSANOVA/fragmentation. This allows to compute small molecules which are not saturated other well-captured by the fragmentation scheme.
• DOC: Adds entry in userguide.
• TEST: Adds regresssion test case.

File:

• doc/userguide/userguide.xml (modified) (4 diffs)

doc/userguide/userguide.xml

@@ -1970 +1970 @@
           . In general, this cost is prohibitive for calculating ground state energies and forces (required for molecular dynamics simulations) for larger molecules such as bio proteins. By fragmenting the molecular system and looking at fragments of fixed size, calculating the ground state energy of a
         number of fragment molecules becomes a linear scaling operation with the number of atoms. In the doctoral thesis of Frederik
-        Heber, it is explained why this is a sensible ansatz mathematically
+        Heber, it is explained why this is a sensible approach mathematically
         and shown that it delivers a very good accuracy if electrons (and
         hence interactions) are in general localized.</para>
         <para>Long-range interactions (e.g. Coulomb or van-der-Waals interactions) are artificially truncated, however,
-        with this fragmentation ansatz. It can be obtained in a perturbation manner
+        with this fragmentation approach. It can be obtained in a perturbation manner
         by sampling the resulting electronic and nuclei charge density on a
         grid, summing over all fragments, and solving the associated Poisson
@@ -1981 +1981 @@
         contained in the <link xlink:href="http://www.scafacos.org/">
             <productname>ScaFaCoS</productname>
-          </link>. This enhancement is currently implemented via another program package named <productname>JobMarket</productname> that is at the moment not available publicly (contact the author Frederik Heber if interested).</para>
+          </link>. This enhancement is implemented via another program package named <productname>JobMarket</productname>.</para>
         <para>Note that we treat hydrogen special (but can be switched off) as
         fragments are calculated as closed shell (total spin equals zero).
@@ -2044 +2044 @@
           i.e. which file formats are used for writing each fragment
           configuration (prefix is &quot;BondFragment&quot;, remember?). Here, we use
-          XYZ (mainly for checking the configurations visually) and MPQC,
+          XYZ (mainly for checking the configurations visually)  and MPQC,
           which is a very robust Hartree-Fock solver. We refer to the
           discussion of the <link linkend="fileparsers">Parsers</link>  
@@ -2065 +2065 @@
           information is inserted afterwards, i.e. it contains only the
           associations from the current fragmentation.</note>
+          </para>
+        </section>
+        <section xml:id="fragmentation.add-seletected-atoms-as-fragment">
+          <title xml:id="fragmentation.add-seletected-atoms-as-fragment.title">Adds selected atoms as one fragment</title>
+          <para>In cases where you don't want to rely on the above fragmentation#
+          scheme or it is not applicable (e.g., for ions or non-saturated
+          molecules), then you can use the whole following automation mechanism
+          with computing the selected set of atoms as a whole.
+          The following call will treat all atoms per associated molecule 
+          as one fragment. These fragments are put either into a job queue
+          (see below on how to continue there),
+          <programlisting>
+  ... --add-selected-atoms-as-fragment &quot;BondFragment&quot; \
+      --grid-level 5 \
+          </programlisting>
+          or they are stored to files.
+          <programlisting>
+  ... --add-selected-atoms-as-fragment &quot;BondFragment&quot; \
+      --grid-level 5
+      --output-types mpqc
+          </programlisting>
+          Note that the grid-level can be still be specified to also compute
+          long-range interactions in the current domain. Electrostatic
+          interactions within the molecule should generally be accounted
+          for by the employed ab-initio method.
           </para>
         </section>