Estimating diffusion constant for AWH simulation

User discussions
1

GROMACS version:2024.3
GROMACS modification: No

Hi
I want to run a protein in water simulation to estimate some PMF using AWH method.
As i know, Diffusion constant (D) is a key .mdp file entry to secure a fast and yet accurate enough simulation. Gromacs reference manual on this subject reads:

…For good convergence, it should be chosen as large as possible (while maintaining a stable system) giving large initial bias updates and fast initial transitions… It may be a good idea to run a short trial simulation and after the first covering…An accurate estimate of the diffusion can be obtained from an AWH simulation with the [gmx awh] tool.

but it does not elaborate on how to extract the estimated diffusion constant from an earlier AWH simulation data using [gmx awh] command line tool.
i will appreciate if someone can help me on this.

Ps. I already have the .edr & log files of the initial AWH simulation with some rounds of covering and wish to use it to estimate D for a fresh AWH simulation.

kind regards,
roozi

2

The reference manual describes this, formula (360). The diffusion is given by D(lambda)=kT eta^-1(lambda), where eta is the friction metric. The friction metric can be obtained with gmx awh -fric