[gmx-users] question about torsional potential fitting

[Victor Rosas Garcia]

Hello everybody,

I have calculated ab initio energies for a torsional potential that I
want to fit to a Fourier series. I have calculated the difference
between the ab initio energies and the forcefield with the torsion
potential set to zero (lets call that "diff curve").  This gives me a
somewhat jagged periodic function. I can then get a cosine series that
parallels the shape of the diff curve, but it is displaced along the
energy axis.
see this plot

https://drive.google.com/file/d/0B2K7fedV_ZFzM3hXZVhyS2VvYnM/view?usp=sharing

If I shift the plot of the series, by adding a constant term, the fit
doesn't look so bad, to me at least, as in the following plot

https://drive.google.com/file/d/0B2K7fedV_ZFzRmlpcnZhckEtdUE/view?usp=sharing

However, I cannot add that constant term in the itp file for my
molecule. Is it enough if the energies calculated by the Fourier
series are parallel to the diff curve?

Best regards

Victor