Re: [AMBER] integration algorithm of sander? And reference of sander?

From: David A Case <>
Date: Mon, 13 Apr 2020 15:21:55 -0400

On Mon, Apr 13, 2020, Erdem Yeler wrote:

>1) What is the integration algorithm is it velocity, Verlet or leap-frog?

Actually, both and neither. For a simple system, the two methods will
give equivalent trajectories in the limit of infinite precison
arithmetic. Once you add SHAKE (bond length constraints) and any sort
of thermostat or barostat, then the distinctions between the two methods
become quite technical and harder to just label as one or the other.

Sander's implementation is closer to velocity Verlet, since we
propagate velocities in two (half-time) steps, rather than as a single
(full-time) step, as in the original Leapfrog method. But the external
interface still uses velocities on the half-step, for backwards
compatibility with the original leapfrog implementation, which dates
back to about 1979.

...hope this helps...dac

AMBER mailing list
Received on Mon Apr 13 2020 - 12:30:03 PDT
Custom Search