Re: [AMBER] calculate potential energies with imin=5

From: David A Case <>
Date: Fri, 15 Oct 2021 08:09:02 -0400

On Fri, Oct 15, 2021, Richard Kullmann wrote:
>I am currently doing Hamiltonian replica exchange MD simulations. Now
>that I want to analyze the data using MBAR, I need potential energy
>information of all the replicas evaluated with all the different
>prmtop files. According to the AMBER 21 manual, I can do this kind of
>post-processing by performing a 1-step minimization (maxcyc=1) of my
>trajectory with imin=5. However, when I compare the obtained energies
>with the output in the mdout file, I am always a little off, up to a a
>couple hundreds kcal/mol in either direction, even after additionally
>setting ntp and/or ntc values in the input file.

The energies printed in the mdout file are one time step off from the
coordinates. So what you see is expected: to get the energies that exactly
correspond to a given snapshot in the trajectory, you need to use imin=5 or
equivalent functionality in cpptraj.

This behavior is explained more fully somewhere, but I can't seem to find it
right now. Dan or others may know of a fuller explanation. The origin is
in how the MD algorithm (originally a leapfrog code) is implemented in


AMBER mailing list
Received on Fri Oct 15 2021 - 05:30:03 PDT
Custom Search