ADF

Overview

Amsterdam Density Functional (ADF) is a computational chemistry program to understand and predict chemical structure and reactivity with density functional theory (DFT).

Versions 2016.107, 2016.101 and 2014.10 are installed on the CSF. This is a pre-compiled binary install, no source code is available.

Restrictions on use

Only members of Research Groups licensed to use this software may access it. All users of the software need to be in the adf unix group and all additions to the group must be approved by the Research Group leader (please copy them on your request to us).

Visitor/collaborators are not permitted to access the software.

License holders:
Prof. Nik Kaltsoyannis (adf unix group)
Prof Steve Liddle (adf-liddle unix group)

The Research Group leader holds a copy of the license terms and conditions for this software and all users must abide by it. Some guidance, which is not intended as a replacement for the license, is given below.

What may ADF be used for?

The license stipulates that:

• the software may only be used by the Research Group to which it is licensed
• the software is only for internal research purposes or non-profit research purposes
• consultancy and commercial work are excluded
• there are restrictions on how you may use any databases supplied
• use is restricted to one geographical site.

Please see clause 6 of the license for further details.

What may ADF not be used for?

You MUST not:

• use ADF for consultancy and commercial work
• make derivative works of the software or incorporate any of the software and/or data files into other programs
• decompile, reverse engineer or disassemble the software
• use the software in the development and production of weapons
• there are restrictions on how you may use any databases supplied
• use is restricted to one geographical site.

Please see clause 6 of the license for further details.

How should the program be cited?

Please see the citation requirements on the vendor website.

Export regulations including remote access

You must comply with all United States and United Kingdom export control laws and regulations controlling the export of the software, including, without limitation, all Export Administration Regulations of the United States Department of Commerce. Among other things, these laws and regulations prohibit, or require a license for, the export of certain types of software to specified countries.

Please be aware that allowing remote access from outside the United Kingdom may constitute an export.

Set up procedure

To access the software you must first load the modulefile for the version you wish to use:

Prof. Nik Kaltsoyannis’ group:

module load apps/binapps/adf/2017.103
module load apps/binapps/adf/2016.107
module load apps/binapps/adf/2016.101
module load apps/binapps/adf/2014.10

Prof. Steve Liddle’s group:

module load apps/binapps/adf/liddle-rg/2017.101

Running the application

Please do not submit any work to the system using the ADF GUI application. Use the jobscript methods shown below instead.

Serial batch job submission

It is not recommended that you run adf in serial, it performs well on multiple cores. If you really feel you need to run it in serial please ask for advice via its-ri-team@manchester.ac.uk before trying it.

Parallel batch job submission

The maximum number of cores that may be used by all users and all jobs at any one time is 64 (license restriction).

Jobs of 2 to 24 cores on a single compute node

1. You must create an input file and ensure that you terminate it with either

   end
   

   or

   eof
   

   Please refer to the ADF manual for further details about setting up input files.

2. Create a job submission file (in the same directory as your input file), for example:

   #!/bin/bash
   #$ -S /bin/bash
   #$ -cwd
   #$ -V
   #$ -pe adf-smp.pe 4
   
   $ADFBIN/adf -n $NSLOTS < input.in > output.out
   

   Note: The variable $NSLOTS ensures that the number of cores ADF uses matched the number requested in your job submission script. This also means that if you want to submit again and use a different number of cores you only need to make a change in the submission script.
   Note: The normal smp.pe must not be used – jobs will fail, but they also do not exit the batch system and cause problems to the scheduler. The pe adf-smp.pe has been set up specifically to run single node ADF single jobs.

3. Submit the job:

   qsub adf-submit.txt
   

   where adf-submit.txt is the name of your submission script.

Jobs of 48 cores or more in multiples of 24 (multi-node jobs on Infiniband connected nodes)

Please follow the same steps as for single node jobs, but the job submission script at step 4 should be:

#!/bin/bash
#$ -S /bin/bash
#$ -cwd
#$ -V
#$ -pe orte-24-ib.pe 48

$ADFBIN/adf -n $NSLOTS < input.in > output.out

The ADF jobs GUI

You may run the ADF jobs GUI on the login node for very short periods of time to view your output. Please do not submit any work to the system via this GUI – it is not configured to do this and doing so will lead to running compute work on the login node which the CSF sysadmins will kill.

You must be correctly logged in to the CSF for the GUI to work – see the note displayed when you load the modulefile – and you need to have X11 tunnelling enabled at connection time.

To start the GUI use the command:

adfjobs &

If you do not log in correctly you will be asked for license information. Close the license window that appears and use the command:

module help apps/binapps/adf/2014.10

for a reminder of the login requirements.

Further info

• The ADF website contains more information and links to documentation, tutorials and examples.

Updates

None.