Hi Jason,
We have taken into account the 1/2 factor that you mentioned. With regards
to the mass weighting, we are deriving down from Gaussian’s Hessian
(Hartree/bohr_radius^2) into the final kcal/(mol.ang^2) as required for
Amber. However, we’re seeing discrepancies through the literature. Some
publications are showing derived force constants which are within the same
kind of kcal/mol.Ang^2 you would expect for Amber (e.g., 200-500 range),
but others are showing bond stretch values in the region of 20-100.
Neither set of publications mention mass weighting. Whilst testing our own
methodology, we can reproduce the former value range *without* mass
weighting, and the later *with* mass weighting.
It is interesting that you introduced the idea of the mass weighted units
- I’m not quite clear how the units would end up being cancelled out from
the final calculation to derive force constants.
Thanks
Anthony
Dr Anthony Nash
Department of Chemistry
University College London
On 09/02/2016 12:39, "Jason Swails" <jason.swails.gmail.com> wrote:
>On Tue, Feb 9, 2016 at 6:56 AM, Thomas Collier <t.collier.12.ucl.ac.uk>
>wrote:
>
>> Dear All,
>>
>> I am currently trying to derive force constants for a new residue in
>> amber. I have derived my force constants, but they are significantly
>> different to analogous atom types already present in amber. However if I
>> do not mass weight the Hessian, the values obtained are close to those
>> present for analogous atom types in the parm13.dat file.
>>
>> My question is are the values for the force constants present in Amber
>> parm**.dat files derived from the mass weighted Hessian? Or is the
>> atomic mass taken into account later, when running LeaP?
>>
>
>I believe it is taken from the mass-weighted Hessian (isn't that the only
>way the units work out correctly?). Did you account for the fact that
>Amber "force constants" are not really force constants, but rather
>half-force constants? That is, the m-w Hessian should give you k such
>that
>U=1/2*k*(x-xeq)^2, whereas the force constants in the Amber parameter
>files
>correspond to U=k*(x-xeq)^2. So they should be half as large (give or
>take) as the values you derive from QM frequency calculations.
>
>HTH,
>Jason
>
>--
>Jason M. Swails
>BioMaPS,
>Rutgers University
>Postdoctoral Researcher
>_______________________________________________
>AMBER mailing list
>AMBER.ambermd.org
>http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Feb 09 2016 - 06:00:03 PST
