Dear Amber experts,
I have a short question regarding the random number generator for
initial velocities (TEMPi) on GPUs:
I ran two MD simulations in a periodic solvent box starting from a
minimized structure, with weak-temperature coupling (NTT=1) and an
initial temperature of 50 K (TEMPi=50.0); for each simulation, I used
an explicit, but different random number seed (ig=NNN). I used the
executable pmemd.cuda_DPFP (Amber20, patch12) on a A40-GPU, since my
original intention was to test whether system relaxation might be save
on GPUs.
The resulting trajectories of the two runs, however, were identical,
although different random seeds were used.
Is the ig-seed not used on GPUs, or for NTT=1 in general?
Maybe I misunderstood the section 21.6.7 in the Amber manual about
temperature regulation...
I did not check ig=-1 up to now, though.
Best regards,
Anselm
Bioinformatik | NHR.FAU
Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU)
Germany
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Feb 04 2022 - 06:30:02 PST
