Summary:
It appears that with the "class2" dihedral style, molecules move under
the influence of only the positive half of the potential function
described in the documentation. (ie. U(phi) for positive phi).
Conformations with negative phi values are treated the same as
conformations with positive phi values, and fall into minima located
at either +phi1 or -phi1. (This is most clear in test #2 below. Test
files attached.)
I know people who use this dihedral style and are not aware of this
limitation. For now at least, can we insert a warning into the
documentation?
---------- Test 1 ----------
The following input script commands should result in a dihedral potential of
U(phi) = 10.0 * (1 - cos(phi - (-60)))
(angles in degrees)
This should have a minima at
phi1 = -60 degrees
test_dihedral_class2.tar.gz (3.45 KB)
(The test file I attached is for test 1. It's easy to modify it for test 2)
After forwarding this report to someone else, I got to thinking that
in retrospect, I could have been more helpful elaborating what I meant
here.
To do that, change the line that says:
dihedral_coeff 1 10.0 -60 0.0 -60 0.0 -60 #<--test1
to
dihedral_coeff 1 10.0 60 0.0 60 0.0 60 #<--test2
You can alter the initial geometry of the system to start out with
either a positive or negative dihedral angle.
For a positive 45 degree starting dihedral angle, line 42 of
"butane.data" should be:
4 1 2 0.0 13.00 11.00 9.00
(In this case annealing usually results in a conformation with a +60
degree twist.)
For a negative 45 angle, it should be
4 1 2 0.0 13.00 11.00 11.00
(In this case it usually falls into a local minima with a -60 degree twist)
There should be only one minima at 60 degrees for these coeff settings.)
I hope that helps.
Andrew