Project

General

Profile

Bug #3049

gmx nmeig should plot a real infrared spectrum

Added by David van der Spoel 2 months ago. Updated 17 days ago.

Status:
New
Priority:
Low
Category:
analysis tools
Target version:
Affected version - extra info:
Affected version:
Difficulty:
simple
Close

Description

gmx nmeig plots an optional spectrum, however the intensities should be computed correctly as well.

History

#1 Updated by Mark Abraham about 1 month ago

  • Target version changed from 2020-infrastructure-stable to 2020-beta1

Is someone planning to work on this? It shouldn't have a target without someone planning to work on it

#2 Updated by David van der Spoel about 1 month ago

I will try and implement it this weekend (and I am assignee). Just listed it here to remind myself that the present implementation is a placeholder.

#3 Updated by David van der Spoel about 1 month ago

Upon looking into this closer it may be harder than thought for non-trivial cases. That is, systems with virtual sites or shells. Since CGenFF now uses virtual sites on all the ligands with Cl, Br or I, this is a very relevant issue.

A quick and dirty solution that ignores this could be implemented in gmx nmeig, but I would prefer to do it properly. In that case, however, it needs to be implemented in the normal modes module in mdrun (or a new derivative of that). The reason for this is that the derivative of the dipole with respect to the coordinates needs to be computed, and the dipole depends on vsites. However, since the vsites are massless they do not contribute to the vibrations and they are not stored in the hessian.

#4 Updated by Paul Bauer 17 days ago

@David, will this still get done before the beta? Otherwise I'll re-target for something later

#5 Updated by David van der Spoel 17 days ago

I did implement a hack in gmx mdrun that prints a text file that can be interpreted by a script. The issue is that we need to pass information in some way to gmx nmeig, or alternatively, print the spectrum straight from gmx mdrun whenever NMA is done but then the whole diagonalization etc. of the Hessian should move to mdrun as well.

Do you have any preference? I would say printing an optional text file from mdrun and optionally reading that in gmx nmeig would be preferred.

Either way, it will not make it to the beta :(

Also available in: Atom PDF