Releases · lanl/hippynn

What's Changed

Add timeplot scaling setting by @shinkle-lanl in #58
add features and example for using them by @lubbersnick in #62
add weights for loss functions by @lubbersnick in #63
Updated Cusp Regularization by @sakibmatin in #59
remove need for backwards compatibility operation in feature nodes by @lubbersnick in #65
energy initialization by @shinkle-lanl in #66
Deploy versioneer 0.29 by @lubbersnick in #64
Ensemble creation feature by @lubbersnick in #68
Added write_h5 function by @bnebgen-LANL in #70
fix coordinate wrapping issue by @shinkle-lanl in #73
Add custom MD driver and improve pairfinder memory performance by @shinkle-lanl in #71
add documentation for new md code by @shinkle-lanl in #76
add documentation for new md code by @shinkle-lanl in #77
06102024 add optimizer/ with batch FIRE algorithm implemented with py… by @amateurcat in #78
update docstring format in molecular_dynamics/md.py for compatibility with sphinx by @shinkle-lanl in #79
fix imports issue causing error when building docs by @shinkle-lanl in #80
minor changes to MD code & its documentation by @shinkle-lanl in #81
Enabled removing leading comments from json DB by @bnebgen-LANL in #83
Added variable to limit pytorch memory in lammps interface by @bnebgen-LANL in #85
[WIP] Add custom triton kernels by @cagrikymk in #84
Update Optimizer by @amateurcat in #82
Fix bug with custom kernels not working on non-default device. by @lubbersnick in #87
Add database features by @lubbersnick in #86
update MLIAP to better find unavailable devices. by @lubbersnick in #91
Improve speed of bond order regression by @lubbersnick in #92
Add units to lammps interface and fix kokkos+cpu bug. by @lubbersnick in #90
Check cuda device capability for triton kernel initialization. by @sakibmatin in #93
Add pytorch lightning trainer by @lubbersnick in #99
Add coarse-graining example by @shinkle-lanl in #98
change setup to include pyproject.toml file by @shinkle-lanl in #97
Many documentation updates, small tweaks to database interface. by @lubbersnick in #100
Check cuda - torch.cuda.get_device_capability by @mgt16-LANL in #102
Add the documentation and examples for hyperparameter tuning by @tautomer in #103
Many improvements to documentation, some workflow improvements by @lubbersnick in #105
Fix documentation workflow by @lubbersnick in #106
Custom kernel improvements by @lubbersnick in #107
fix bug with saving batched MD results by @shinkle-lanl in #109
clarify language on units for custom MD algorithms, remove default units by @shinkle-lanl in #112
new example by @shinkle-lanl in #108
change method of custom MD driver for handling units by @shinkle-lanl in #115
add missing import by @shinkle-lanl in #116
fix StressForceNode by @shinkle-lanl in #110
Fix some issues with the speed tests in custom kernels by @lubbersnick in #117
improve ensembling robustness and documentation by @lubbersnick in #119
fix error handling for custom kernel benchmarking by @lubbersnick in #120
Add atomization-consistent training by @lubbersnick in #123
add species indexing node by @shinkle-lanl in #128
Kdtree fix by @shinkle-lanl in #130
new Langevin MD integrator implemented by @shinkle-lanl in #129
Add batched prediction to molecular dynamics by @miketynes in #134
Add LAMMPS MD option to coarse graining example by @shinkle-lanl in #135
add easy method to swap pairfinder on GraphModules and Predictors by @shinkle-lanl in #133
Fix torch.load weights_only argument compatibility by @shinkle-lanl in #136
add backwards compatibility to add_safe_globals calls by @lubbersnick in #138