Re: [AMBER] MBAR energies when using soft-core potential in pmemd TI

From: Brian Radak <>
Date: Mon, 10 May 2021 09:21:56 -0400

Are you using ifmbar_net = 1 (not default)? The "star output" certainly
indicates large numbers, but perhaps not always as large as one thinks. The
net keyword switches to printing differences, which are the actual MBAR
input and generally quite a bit smaller.


On Mon, May 10, 2021, 9:00 AM David A Case <> wrote:

> On Mon, May 10, 2021, Hosein Geraili Daronkola wrote:
> >
> >When I use soft-core potential for TI, in GPU pmemd 2018, for many of the
> >intermediate states I get MBAR energies for either lambda 0 or 1, which is
> >either very different from the other lambdas, sometimes very positive, and
> >sometimes "**********". If I just ignore this issue my result looks fine
> >with TI calculations.
> There were bugs in the MBAR printouts in Amber18, but I don't know much
> about the details. Maybe Ray Luo or Charlie Lin will recognize the
> symptom,
> or have a different suggestion.
> It's probably worth upgrading to Amber20 (and apply all the updates).
> Or, you might try applying Amber20, update.7 to your 2018 code, either by
> hand or by using patch. It only involves a one-line change in
> mdin_ctrl_dat.F90. Of course, you problem might have some other origin.
> ....dac
> _______________________________________________
> AMBER mailing list
AMBER mailing list
Received on Mon May 10 2021 - 06:30:04 PDT
Custom Search