GitLab

Updated documentation for Lepton

Performance improvements to CPU implementation of custom forces

Take 1-4 nonbonded scaling parameters from the amber topology file

Write CONECT records to PDB files

present, defaulting to 1.2 (SCEE) and 2.0 (SCNB) if they're not present. If
values for scee and scnb are provided to readAmberSystem they will override any
values in the topology file, but a warning will be printed (since ignoring the
scaling parameters in the topology file really changes the force field).

Prevent segfaults when a simulation blows up

Created optimized CPU platform

correct minor bug in GBn screening parameter determination.

instead of gb is intentional.

was:

if symbl[0] == ('c' or 'C'):
   ...do stuff...

This logic will match only 'c' as ('c' or 'C') evaluates to 'c' while
('c' and 'C') evaluates to 'C':

>>> 'c' or 'C'
'c'
>>> 'C' == ('c' or 'C')
False

The correct logic is either

if symbl[0].upper() == 'C':
   ...do stuff...

or

if symbl[0] in ('C', 'c'):
   ...do stuff...

(Alternatives can be [ if symbl[0] in 'cC' ] or [ if symbl[0].lower() == 'c'])

Separated hardware query functions into a separate file.  CPU platform checks for the correct hardware.

Python records where OpenMM is installed, so it doesn't need to be specified with environment variables at runtime

Lots of code cleanup related to constraints in the reference platform