GitLab

Bug fixes

Added serialization proxy for CustomCentroidBondForce

Update CustomHbondForce.h

Add OpenCL to Appveyor and fix nosetests

Minor changes to doxygen comments

Fixed error in energy computation

Support nucleic acids in AMOEBA 2009

Add CONECT bonds for multi-model pdb files 

Fixed bug in getState() when enforcing periodic boundary conditions

Increased SHAKE iterations when positioning extra particles

Amber switching

 Conflicts:
	wrappers/python/simtk/openmm/app/amberprmtopfile.py

In fixing the merge conflict, I went ahead and fixed up the switchDistance logic
to match what I did in CharmmPsfFile.

Deleted GBVIForce

Fix #1213 addInteractionGroup should accept `set` values as parameters

apparently SWIG typemaps maps Python's list but not set with std::set.
Instead, Python's `set` is mapped with `std::unordered_set`.

The original code of `swigInputBuilder.py` is kindly in a mess. Too much
lines in a single method `writeMethods` which causes debug becomes
difficult. Also there are a lot of warnings about
[PEP8](https://www.python.org/dev/peps/pep-0008/).

A refactoring in the future may be a good idea.

Residue templates can specify atoms by name instead of index

AMD docstring fix

Deploy requires AWS region