Re: [AMBER] Transformation from cartesian to torsion space

From: David Cerutti <>
Date: Mon, 22 Jan 2018 16:39:50 -0500

You can check the various codes--perhaps the easiest to access is mdgx,
based on the DLPoly implementation. Look in the file BondContrib.c, which
is #include'd multiple times in Bonded.c. On line 471 of BondContrib.c,
there is the comment "// Transform the dihedral forces to cartesian
coordinates." It's all a bunch of chain rules--the transformation for
angular forces is a bit easier to hash through, but in each case you can
verify that the fomulaic sums of forces on all atoms add to zero. The
conversion of two-body forces to Cartesian space also rests on a chain
rule, but it's so simple and intuitive that the technicality is often


On Mon, Jan 22, 2018 at 3:00 PM, Siddhartha Jain <> wrote:

> Hi,
> So from what I've been reading, AMBER force field has terms that depend on
> the dihedral angle. However if you ask for gradients via sander, you get
> gradients for the cartesian space. So I'm guessing that in the function to
> compute the energy, there's a transformation from cartesian to torsion
> space?
> What transformation is used?
> Thanks
> --Sid
> --
> _______________________________________________
> AMBER mailing list
AMBER mailing list
Received on Mon Jan 22 2018 - 14:00:02 PST
Custom Search