From bwHPC Wiki
< JUSTUS2‎ | Software
Revision as of 14:56, 9 May 2017 by C Mosch (talk | contribs) (Version-specific information)
Jump to: navigation, search
Description Content
module load chem/adf
Availability BwForCluster_Chemistry
License Commercial - see SCM License Terms
Citing Required citations as well as module specific citations
Links ADF Homepage and product specific documentation pages
Graphical Interface ADF GUI
User forum ADF Discussion List

1 Description

The ADF Modeling Suite by SCM (Software for Chemistry & Materials) is a collection of programs for modeling chemical and physical properties of molecules and solids. The suite consists of the following components:

  • With help of the graphical user interface adfinput one can construct molecules, solids and slabs. All kinds of calculational parameters like functionals, basis sets, convergence parameters can be defined as well. The calculation and its parameters can be saved as a run-file which is suitable for inclusion in our queueing system scripts. For details see the Tutorials of SCM.
  • adfjobs can used for interactive job control. This is intended only for short test jobs running in background for some minutes. For longer jobs one should save the job as run-file which can be called from within a queueing system script. For details see the Tutorials of SCM.
  • adf is a powerful molecular DFT (density functional theory) code applicable in many areas of chemistry and materials science. One strength of ADF is in the field of inorganic chemistry.
  • band is a DFT code for periodic systems using local slater type orbital basis sets.
  • dftb provides fast approximate methods for studying large molecules and large periodic systems.
  • reaxff is a bond order based force field program for studying reaction dynamics in large complex systems.
  • The COSMO-RS method allows to predict thermodynamic properties of solutions and mixtures (LogP, VLE, pKa, ...) based on quantum mechanical data from ADF.

2 License and citing

The ADF program suite is available for all users of the cluster. When you publish any results obtained with the ADF program suite you must cite ADF and the used components properly. Therefore please see the required citations. Furthermore there might be additional component specific citations that can be found in documentation of the individual components.

3 Versions and Availability

A list of versions currently available on the bwForCluster Chemistry can be obtained from the Cluster Information System (CIS): {{#widget:Iframe |url=https://cis-hpc.uni-konstanz.de/prod.cis/Justus/chem/adf |width=99% |height=120 |border=0 }} On the command line of a particular bwHPC cluster a list of all available versions is displayed by command

$ module avail chem/adf

4 Usage

4.1 Loading the module and module help

You can load the default version of ADF with the command

$ module load chem/adf

There are no dependencies to other modules. All required components, e.g. Intel MPI, are included in the ADF module.

Additional help is available via command

$ module help chem/adf

Please read this module-specific help before using the software.

4.2 Creating input files with adfinput

Probably the easiest method to create input files is using the graphical user interface adfinput: First start and connect to a 3D-accelerated VNC session via TigerVNC running on one of our visualization nodes. Then - within the VNC window - load the adf module and run adfinput with 3D acceleration:

module load chem/adf
vglrun adfinput

With help of ADFinput molecules and periodic systems can be constructed (depending on the selected target program, i.e. adf or band). Furthermore one can control most calculational parameters like DFT methods, basis sets, convergence criteria, etc. After saving the calculation via File -> Save as with name MY_JOB, a run-shell script with name MY_JOB.run can be found on disk.

This run-shell script contains the job input and executes the appropriate ADF commands like adf, band, etc (the input is piped via STDIN into those commands). The run-script can be called from within your ADF queueing system shell script, thus the queueing system shell script still defines the queueing system parameters, manages the temporary directories, loads the module, cleans up after the end of the job, etc).

For an introduction how to use the ADF GUI please follow some of the ADF tutorials. If you want to change the job input without GUI, you can edit the run-shell scripts manually. All possible keywords and options are documented in the manuals sections of the documentation of the individual programs.

4.3 Parallel computing

In queueing system scripts the commands adf, band, ... automatically use the cores as specified when submitting the job script. Neither mpirun nor any further command line options are required.

In case of interactive single-node test jobs one can limit the number of cores via command line option -n NB_OF_CORES, for example adf -n 8.

4.4 Disk usage

Temporary files are stored below directory SCM_TMPDIR. Typically this environment variable is defined within the queing system script and points to some node-local directory below TMPDIR. In case of multi-node jobs, SCM_TMPDIR must point to the same directory on every node and must exist on every node.

When loading the adf module, SCM_TMPDIR is defined automatically and usually points to TMPDIR. When running under control of the queueing system, TMPDIR is the node-local scratch directory of the job.

Please never put temporary files of ADF on central file systems like HOME or SCRATCH or workspaces.

5 Input and queueing system examples

Links to input and queuing system example files can be found via

$ module help chem/adf

5.1 Interactive parallel example

The example shows, how to load the ADF module and run ADF on an interactive node using 4 cores:

$ mkdir ~/ADF-example-job/
$ cd ~/ADF-example-job/
$ module load chem/adf
$ adf -n 4 < $ADF_EXA_DIR/adf-Energy_H2O-all_electron-example.inp

6 Version-specific information

To list all available versions of adf run the command

$ module avail chem/adf

For specific information about version X, see the output of command

$ module help chem/adf/X

To load and use a specific ADF version X, use the command

$ module load chem/adf/X

In general it might be desirable to specify the exact version in queueing system scripts of production runs. This ensures the consistency of the final result when combing intermediate results of several runs.