Hello Amberites,
I’ve been working with students to use MD simulations produced by
pmemd.CUDA to estimate diffusion constants of small molecules (from to
simple sugars) as an class exercise to compare water and solvent models
available by default. These simulations feature multiple solutes in a large
water box and are propagated for 10 ns. I note that the solutes are
separated by 2 A more than the cutoff distance to minimize their
interactions, and we observe no aggregation over the short simulation
timescale.
Interestingly, despite this being in line with literature procedures, the
resulting diffusion constants calculated with the velocity autocorrelation
function or diffusion analysis in cpptraj did not match literature nor was
there an expected correlation between the molecule size and diffusion
constant. This was alarming, so we have now varied the simulations using
NVE, NPT, or NVT setups, turning off shake, decreasing or increasing the
time step or sampling rate, turning off jfastw, etc. in a seemingly endless
variation of molecular dynamics simulations. All this and no correlation
observed.
While the student learned nicely how to automate the execution, analysis,
and reporting of molecular dynamics results, the overall problem remains so
I thought to pick the community’s collective brain as to why these
simulations fail to reproduce diffusion constants where it has easily been
done in the past. Is there some obvious approximation made for water (also
chloroform and methanol) models in amber or pmemd.cuda that I am missing?
Kindly,
David Poole
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jun 08 2023 - 11:30:02 PDT
