hi Adrian and others,
we indeed benefit from your team's work. We went ahead using 4 fs in
Amber14 pmemd GPU and got this paper:
"Folding simulations for proteins with diverse topologies are accessible in
days with a physics-based force field and implicit solvent"
http://pubs.acs.org/doi/abs/10.1021/ja5032776
(sadly we published this paper 1/2 year before yours so can not cite it)
So far we run extremely long MD/REMD (up to 50 microsecond with single run
and few ms aggregated data). 4 fs + pmemd.cuda works really well with GB
simulation (have not extensively run TIPP3 yet). And the speeds are
amazingly fast. (tables in supp).
PS (to others): like Adrian said, it is very straight forward to use 4fs.
Just use ParmEd (we used default option) to update masses
(via HMassRepartitioning command) in topology file and go ahead changing dt
= 0.002 to dt = 0.004 (and enjoy the speeding up).
Best
Hai
On Mon, Mar 16, 2015 at 5:40 PM, Adrian Roitberg <roitberg.ufl.edu> wrote:
> Dear all,
> I usually do not like doing this, but I think our recent paper would
> benefit many of you.
> Please take a look at pubs.acs.org/doi/abs/10.1021/ct5010406
>
> "Long Time Step Molecular Dynamics through Hydrogen Mass Repartitioning
>
> Chad W. Hopkins , Scott Le Grand , Ross C Walker , and Adrian E. Roitberg"
>
> In this paper, we show how to manipulate the masses of atoms in Amber,
> in order to be able to use a time step of 4 fs, while retaining the same
> thermodynamic properties (and essentially the same kinetic properties)
> as you would get when using regular masses and a dt=0.002
>
> So, you gain a factor of 2x with a very small change.
>
> If you use Parmed, you can use the HMassRepartitioning command to change
> your prmtop to the 'new' masses, and then just change your mdin to dt=0.004
>
> Please use it and report back successes and failures !
>
> Thanks
> adrian
>
> --
> Dr. Adrian E. Roitberg
> Professor.
> Department of Chemistry
> University of Florida
> roitberg.ufl.edu
> 352-392-6972
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Mar 16 2015 - 17:30:03 PDT
