GitLab

* add tests for correctness of step and time written in XTC and DCD

* improve tests

* improve xtc tests

* fix XTC/DCD time/step writing

* different approach by changing the reporters to not pass currentStep as firstStep but instead interval

* undo change

* Remove includes for Dart

* Fix CMake error

* Fixed broken link

* Pin to cmake<4.0

* Try to make the link checker stop complaining

Correctly assign CMAPs to torsions across multiple CMAPTorsionForce sections

* Began splitting CommonKernels into multiple files

* Moved two more kernels into separate files

* Moved two integrators into separate files

* Fix compilation error on Windows

* reset overflowed state energies at the alchemical endpoints

* address formatting, complete clash test

* Fixed indentation

---------
Co-authored-by: Peter Eastman <peter.eastman@gmail.com>

* Replace pthreads with C++ threads

* Try to fix CI errors

* Try including -pthread linker option

Check all wildcard combinations for dihedrals and impropers.  Fixes #4828.

Add PME optimizer and rigid body scripts

* In CharmmPsfFile, add lonepair exceptions even when no Drude particles
  are present

* In ForceField, correctly update the indices of virtual site exclusion
  reference particles in a residue when applying a patch to it

* Properly handle duplicate atom type definitions without elements

* Correct the description of a test in a docstring

* Correctly handle existing type without element and duplicate type with element

* Try to update actions/cache to v4

* Ensure that in_prec will never be uninitialized

* Add a test case that uses a system size below the compression threshold

* Compare and track openmm.unit.BaseUnit objects by name

* Retrieve _conversion_factors as a class attribute

* Include more cases in TestForceField.test_NBFix

* Reduce LennardJonesGenerator NBFIX memory consumption

* Restored test case to use multiple elements to avoid ambiguity

This is similar to https://github.com/openmm/openmm/pull/4528 but was
missed out. When the membrane and protein are combined a new topology
is created. This invalidates the residueTemplates dict so it must be
recreated before calling addSolvent.

* CPU platform checkpoints random number generator

* Fix Windows compilation error

* Another Windows compilation error

* Use Intel OpenCL for CI

* Set environment variables

* Try to get CI to run

* Debugging

* Debugging

* Fixes for Intel OpenCL

* updateParametersInContext() can modify parameter offsets

* Reordering respects parameter offsets

* Implemented for CUDA and HIP