GitLab

Add elapsed time column in statedatareporter

Disabled PME stream on CUDA 7

Eliminate scipy RuntimeWarning when reading NetCDF restart files

support gen-pairs=no in Gromacs

This was actually fully supported already, since it parsed pairtypes and
assigned the exception parameters correctly from there. All this does is throw
an exception when pair parameters are not defined and gen-pairs=no.

Due to a bug that was uncovered in mdtraj with scipy's NetCDF implementation
regarding mmapping, scipy started issuing the following warning if any NetCDF
variable objects had references remaining after closing the file:

/.../lib/python2.7/site-packages/scipy/io/netcdf.py:287:
RuntimeWarning: Cannot close a netcdf_file opened with mmap=True, when
netcdf_variables or arrays referring to its data still exist. All data arrays
obtained from such files refer directly to data on disk, and must be copied
before the file can be cleanly closed. (See netcdf_file docstring for more
information on mmap.)

We were already copying data to avoid segfaults on some scipy versions, but
there were still references to variables that went out of scope shortly after
the NetCDF file was closed.  Simply deleting these references gets rid of the
warning.

Improved accuracy of energy computation

Raise exception when processing Gromacs molecule with virtual sites

Moved the MTS and aMD integrators into the main repository

Virtual sites are never created by the OpenMM system, and their information is
not parsed from the Gromacs topology file. For now just throw an exception
rather than let them run, since TIP4Pew behaves *very* strangely...

Tabulated erfc() to improve accuracy of energies

Fix Gromacs #define replacement and add support for Gromacs 5

Gromacs uses the preprocessor to replace previously #define'd tokens in strings.
This is used, for instance, to override certain parameters without having to
introduce new file types.  This is how Gromacs implemented ff99SB-ILDN (i.e., by
simply replacing the chi torsion for isoleucine, leucine, aspartate, and
asparagine with a define token).  So OpenMM *was* missing some in-line
parameters that were defined.

The immediate consequence I see is that whenever someone tried to run
ff99SB-ILDN with OpenMM, what they were *really* getting was ff99SB. Not sure
where else, if anywhere, this kind of thing was happening.

Switched to a new Windows pthreads library

Conflicts:
	libraries/pthreads/include/pthread.h

Implemented runForClockTime()

Added checkpointing functions to Simulation

Update setup.py

Improved accuracy of correction for exclusions with PME

mingw has STRUCT_TIMESPEC

Implemented select() in custom expressions

fix mingw build

Fixed failing test cases.  Increased default minimizer tolerance to 10.