periodic dihedral extrapolation

[marvinbernhardt]

When dealing with dihedrals I ran into problems with the extrapolation. potential_to_gromacs.sh would extrapolate both sides to 180°. If the new potential is then interpolated and the force is calculated, one obtains a very high force for x=180.

This pull request is the solution I came up with. It will ignore the gradient of the last points on the right side and extrapolate linearly to the same value one finds on the left side.

[junghans]

So we set the force to same value on both side (https://github.com/votca/csg/blob/master/share/scripts/inverse/table_to_xvg.pl#L101)
but the value is too high, correct?

[codecov]

Codecov Report

Merging #576 into master will not change coverage.
The diff coverage is n/a.

@@          Coverage Diff           @@
##           master    #576   +/-   ##
======================================
  Coverage    62.2%   62.2%           
======================================
  Files         124     124           
  Lines        7722    7722           
======================================
  Hits         4805    4805           
  Misses       2917    2917           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update b02ead2...693b716. Read the comment docs.

[junghans]

@marvinbernhardt I am all for merging this as dihedrals aren't very well supported currently. Have a look at my comments above. Do dihedrals always need to go from 0 to 180?

[junghans]

@votca-bot changelog: add periodic extrapolation for dihedrals

[marvinbernhardt]

So we set the force to same value on both side (https://github.com/votca/csg/blob/master/share/scripts/inverse/table_to_xvg.pl#L101)
but the value is too high, correct?

Yes, precisely

[marvinbernhardt]

Ups, sorry reverted those codacy changes. Will change again.

[junghans]

@marvinbernhardt I undid the last two commits, was there a special reason to split the option parsing? I think we can alllow periodic for the left as well, even though we don't have a use case yet.