This might have been a little short-sightedness with respect to using the
parabolic restraint shortcuts on torsions, but there may be some unintended
consequences if you only make that exact change. I will look at the code
carefully and make sure that this change won't break anything else. Namely,
one alternative might be instead of:
if ( abs(k0) > zerner ) then
to use:
if ( abs(k0) > zerner .and. (abs(r0) > zerner .or. resttype(i) < 3) ) then
I don't want to break anything or needlessly sacrifice any robustness
against user error.
In the mean time, it's possible to get the exact behavior you desire for the
torsion restrained to 0 by setting r1 = -180.0, r2 = 0.0, r3 = 0.0, and r4 =
180.0, and then rk2 = k0 and rk3 = k0 while r0 and k0 remain 0.0.
Matthew Seetin
University of Rochester Medical Center
-----Original Message-----
From: amber-developers-bounces.ambermd.org
[mailto:amber-developers-bounces.ambermd.org] On Behalf Of case
Sent: Saturday, December 12, 2009 4:14 PM
To: amber-developers.ambermd.org
Subject: [AMBER-Developers] restraint bug?
Comments from people who might know something about this are welcome...dac
----- Forwarded message from Tyler Luchko <tluchko.rci.rutgers.edu> -----
Date: Fri, 11 Dec 2009 14:11:34 -0500
From: Tyler Luchko <tluchko.rci.rutgers.edu>
Subject: restraint bug
I forgot to mention what I think is a bug with the new restraint syntax
(r0 and k0). Line 5092 of nmr.f
if (abs(r0) > zerner .and. abs(k0) > zerner ) then
should be
if ( abs(k0) > zerner ) then
or else you cannot restrain to zero. There are work-arounds; one could
set r0=zerner*10 (1e-6) or r0 = 360, in my case for the torsion
restraints. However, I thought I would just change the code but check
with you that I haven't missed some logic in the current form first.
Tyler
----- End forwarded message -----
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Sat Dec 12 2009 - 22:00:02 PST
