Re: [AMBER] Convergence value for GMAX

From: David A Case <>
Date: Thu, 17 Oct 2019 13:36:04 -0400

On Thu, Oct 17, 2019, Anuja Ware wrote:

>I am new to amber. I have set my system following the tutorials and the
>mailing list. In my study, I want to perform QM/MM calculations and only
>want to compare energies of the same system with the different redox state
>of a cofactor.
> Initially, I have set my system, assigned parameters for cofactors and run
>MM minimization to relax the system. Input file is as follows:
> imin=1,
> maxcyc=10000,
> ncyc=15000,
> ntmin=1
> cut=12.0,
> ntb=0,
> ntr=0,
> &end
>The final output of minimization is as follows:
> 10000 -1.7204E+04 1.9128E-01 6.5236E+00 FE 12904
>So, my question is, can I consider that my system is relaxed sufficiently
>with this GMAX value? or I have to check the parameters or another thing
>for Fe to further lower the GMAX?

You need to think about what you mean by "relaxed sufficiently": if you
are planning to continue with MD (with or without QM/MM), you should be

>What should be the values of GMAX or RMS to consider the fully minimized

[Aside: the following is probably not relevant to your situation:]

"Fully minimized" (to me) means exactly that: RMS values down to
something close to machine precision (10^-12). You would want that for
subsequent normal mode calculations, although in practice, getting the
RMS down to 10^6 is usually fine.

Note that you can't get this without setting the cutoff to a value
like 999 that is larger than the extent of your system. You will also
want to invoke xmin minimization.

Note that *very* few types of calculations require full minimization,
but it is possible.o

...good luck...dac

AMBER mailing list
Received on Thu Oct 17 2019 - 11:00:02 PDT
Custom Search