Re: [AMBER] Estimating time in AMD

From: Adrian Roitberg <>
Date: Mon, 08 Oct 2012 11:10:48 +0200


The issue you raise in very interesting.The easy answer is that you
cannot scale the time step. This is because you use a 'real' time step
in a modified potential energy surface (amd surface) and that cannot be
easily related to a change in 'time'.

If you run long enough, you can see how long it takes to jump over
barriers with regular md and amd, and that should scale trivially with
barrier height. That gives you a scaled time scale. However, the scaling
factor depends very much on your position in the surface. Your minima
are higher in energy in AMD, but the barriers do not change (in absolute
energy). So, you do not have a global scaling factor.

As a recipe, unless you can really validate things, you should NOT use
amd to derive time scales for processes. Just stick with the
thermodynamics of the problem and you are safe.


On 10/8/12 10:25 AM, Romelia Salomon wrote:
> Dear Soumya
> The amd.log file contains the AMD pertinent information produced along the
> calculation. It is printed every time the trajectory is printed, as those
> are the configuration points that will be available for analysis.
> The information being printed is:
> 1) The total potential energy for that step (without the extra added by
> AMD, to keep in reference with mdout)
> 2) the total dihedral energy for that step (without the extra added by
> AMD, to keep in reference with mdout)
> 3) factor by which total forces are scaled
> 4) factor by which dihedral forces are scaled
> 5) energy added by AMD to the total potential energy
> 6) energy added by AMD to the dihedral energies
> I hope this helps.
> Romelia
>> Dear Amber Users,
>> I wanted to calculate the modified time step Δt* in the accelerated MD
>> which one can find by using the equation Δt x boost factor. How can one
>> get
>> the value for boost factor?
>> Thanks
>> Soumya
>> _______________________________________________
>> AMBER mailing list

                            Dr. Adrian E. Roitberg
                Quantum Theory Project, Department of Chemistry
                            University of Florida
AMBER mailing list
Received on Mon Oct 08 2012 - 02:30:02 PDT
Custom Search