Re: [AMBER] Leapfrog algorithm

From: David A Case <>
Date: Wed, 18 Jun 2014 14:34:09 -0600

On Wed, Jun 18, 2014, Jason Swails wrote:
> The leapfrog algorithm _is_ used to integrate the equations of motion in
> sander and pmemd.

Since this has come up several times recently: the situation is more complex.
Sander and pmemd integrators are closer to velocity Verlet than to leapfrog,
as the way the kinetic energy is estimated comes from velocity Verlet.
Restart files make things look like leapfrog (for historical reasons).

In terms of roundoff errors, sander and pmemd are neither "pure" leapfrog nor
"pure" velocity Verlet.

Bottom line, if you really care about the differences between leapfrog and vV,
you need to study the code itself.


AMBER mailing list
Received on Wed Jun 18 2014 - 14:00:02 PDT
Custom Search