Re: [AMBER-Developers] a note regarding qm/mm shake

From: Timothy J Giese <>
Date: Fri, 26 Feb 2016 14:02:42 -0500

On 02/26/2016 01:13 PM, David A Case wrote:
> On Thu, Feb 25, 2016, Ross Walker wrote:

>>> (Actually, there's something more subtle here because the offset
>>> seems to apply to the X-H bonds, in particular, and not the heavy-atom
>>> bonds, but I haven't been able to perform a complete pathology)
> With ntc=2, Amber only shakes bonds to hydrogens. Generally, that is all that
> anyone really uses, since shaking heavy atom bonds is not supported for
> parallel runs, or for qmmm.

At the time that I wrote the above quote, I was mystified that setbon
deleted the X-Y and X-H bonds, but the arrays in shake were not offset
by the total number of deleted bonds. I had failed to realize that the
X-Y and X-H data are stored in two sets of arrays. Now that I realized
that, I guess it makes perfect sense that the X-H arrays were only
offset by the number of deleted X-H bonds.

There also appear to be an issue with offsets in the conp array, i.e.
x(l50), but I'll refer interested readers to the bugzilla report that
Ross kindly setup.

AMBER-Developers mailing list
Received on Fri Feb 26 2016 - 11:30:04 PST
Custom Search