Distributions with Iterative Boltzmann Inversion and LAMMPS

[Joerg-Ruediger Hill]

I have been trying to run Iterative Boltzmann Inversion with LAMMPS on a test hexane system as in the original VOTCA paper. However, the iterations stop in the first cycle since there is an obvious mismatch of the original distributions generated with csg_stat and the distributions generated in the iterations. In particular, the peaks for the B1-B2 bonds (around 2.5 Ang) show up in all distributions for the non-bonded interactions and the distribution for the B1-B2 bond is the one for the B2/B2 non-bonded interaction. Obviously, there is a problem with matching the correct interactions.

How does csg_inverse figure out where the bonds should be in the CG simulation? I have attached all my files.

[Christoph Junghans]

On Wed, Feb 28, 2024 at 06:36 'Joerg-Ruediger Hill' via votca <vo...@googlegroups.com> wrote:

I have been trying to run Iterative Boltzmann Inversion with LAMMPS on a test hexane system as in the original VOTCA paper. However, the iterations stop in the first cycle since there is an obvious mismatch of the original distributions generated with csg_stat and the distributions generated in the iterations. In particular, the peaks for the B1-B2 bonds (around 2.5 Ang) show up in all distributions for the non-bonded interactions and the distribution for the B1-B2 bond is the one for the B2/B2 non-bonded interaction. Obviously, there is a problem with matching the correct interactions.

How does csg_inverse figure out where the bonds should be in the CG simulation? I have attached all my files.

the bonds are read from the topology file, in case of lammps you will need to make a xml topology to define bonds (and exclusions for the non-bonded interactions). Without the exclusions you might get artificial peaks in the rdfs.

Christoph

--
Join us on Slack: https://join.slack.com/t/votca/signup
---
You received this message because you are subscribed to the Google Groups "votca" group.
To unsubscribe from this group and stop receiving emails from it, send an email to votca+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/votca/cdf7c2b9-b8ce-48be-8e35-9aafaaf1e42dn%40googlegroups.com.

[Joerg-Ruediger Hill]

Dear Christoph,

thanks for the quick answer. As I understood there should be bonds and angles defined in the topology file, which was indeed not the case.

However, adding them (as in the file attached) makes no difference in the outcome. If I compare the target distribution and the distribution

after the first iteration for the B1/B1 nonbond interaction I have peaks for the bonds only in the distribution from the first iteration (see figure).

Similiarly, if I do the same for the B1-B2 bond there are no peaks at around 2.5 Ang in the distribution from the first iteration.

Do I have to add some keywords to make exclusions?

Best Regards,

Jörg

[Christoph Junghans]

On Thu, Feb 29, 2024 at 07:36 'Joerg-Ruediger Hill' via votca <vo...@googlegroups.com> wrote:

Dear Christoph,

thanks for the quick answer. As I understood there should be bonds and angles defined in the topology file, which was indeed not the case.

However, adding them (as in the file attached) makes no difference in the outcome. If I compare the target distribution and the distribution

after the first iteration for the B1/B1 nonbond interaction I have peaks for the bonds only in the distribution from the first iteration (see figure).

Similiarly, if I do the same for the B1-B2 bond there are no peaks at around 2.5 Ang in the distribution from the first iteration.

Do I have to add some keywords to make exclusions?

Unfortunately not, you can check the exclusions with csg_dump —excl. 

You should also restrict the min and max of the interaction.

Christoph

Best Regards,

Jörg

Christoph Junghans schrieb am Mittwoch, 28. Februar 2024 um 16:20:58 UTC+1:

On Wed, Feb 28, 2024 at 06:36 'Joerg-Ruediger Hill' via votca <vo...@googlegroups.com> wrote:

I have been trying to run Iterative Boltzmann Inversion with LAMMPS on a test hexane system as in the original VOTCA paper. However, the iterations stop in the first cycle since there is an obvious mismatch of the original distributions generated with csg_stat and the distributions generated in the iterations. In particular, the peaks for the B1-B2 bonds (around 2.5 Ang) show up in all distributions for the non-bonded interactions and the distribution for the B1-B2 bond is the one for the B2/B2 non-bonded interaction. Obviously, there is a problem with matching the correct interactions.

How does csg_inverse figure out where the bonds should be in the CG simulation? I have attached all my files.

the bonds are read from the topology file, in case of lammps you will need to make a xml topology to define bonds (and exclusions for the non-bonded interactions). Without the exclusions you might get artificial peaks in the rdfs.

Christoph

--
Join us on Slack: https://join.slack.com/t/votca/signup
---
You received this message because you are subscribed to the Google Groups "votca" group.
To unsubscribe from this group and stop receiving emails from it, send an email to votca+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/votca/cdf7c2b9-b8ce-48be-8e35-9aafaaf1e42dn%40googlegroups.com.

--
Join us on Slack: https://join.slack.com/t/votca/signup
---
You received this message because you are subscribed to the Google Groups "votca" group.
To unsubscribe from this group and stop receiving emails from it, send an email to votca+un...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/votca/b00b29db-84ae-4a2d-97f0-722d46b46f79n%40googlegroups.com.

[Joerg-Ruediger Hill]

Dear Christoph,

thanks again for your answer. The problem was in the LAMMPS structure file. All beads have to be sorted by molecule for VOTCA to work correctly.

LAMMPS itself has no problems with other sort orders. Maybe this could be added to the VOTCA manual.

Best Regards,

Jörg

[Christoph Junghans]

On Fri, Mar 1, 2024 at 09:14 'Joerg-Ruediger Hill' via votca <vo...@googlegroups.com> wrote:

Dear Christoph,

thanks again for your answer. The problem was in the LAMMPS structure file. All beads have to be sorted by molecule for VOTCA to work correctly.

LAMMPS itself has no problems with other sort orders. Maybe this could be added to the VOTCA manual.

Ok thanks for the information, could you open an issue on GitHub with a small example, so we can try to fix that?

Thanks,

Christoph

To view this discussion on the web visit https://groups.google.com/d/msgid/votca/83980801-f98e-4f0a-a203-10aa355656cfn%40googlegroups.com.