Correctly count acceptance rate when maximum box move size is zero.
Correctly count acceptance rate when maximum particle move is is.
Add convex_spheropolyhedron_union shape class.
When building with ENABLE_CUDA=on, CUDA 8.0 is now a minimum.
Bundle libgetar v0.7.0 and remove sqlite3 dependency.
Improvements are only expected with NVLINK hardware.
Node-local acceleration is implemented for a subset of kernels.
Run on the first n GPUs (Pascal and above). Execute with -gpu=0,1.,n-1 command line option to
Support multi-GPU execution on dense nodes using CUDA managed.
To obtain output on some batch job systems
Flush message output so that python -u is no longer required.
Accept mpi4py communicators in context.initialize.
Improve rigid body pressure calculation performance with multi-GPU execution.
Improve some angular degrees of freedom integrators with multi-GPU execution.
Improve integrate.nve and integrate.npt performance with multi-GPU execution.
Fix a sporadic data corruption / bus error issue when data structures are dynamically resized in simulations that use unified memory (multi-GPU, or with -DALWAYS_USE_MANAGED_MEMORY=ON compile time option).
Display correct cell list statistics with the -DALWAYS_USE_MANAGED_MEMORY=ON compile option.
Improve rigid body performance with multi-GPU execution.
Improve charge.pppm performance with multi-GPU execution.
Improve pair.table performance with multi-GPU execution.
Fix illegal memory access in NeighborListGPU with -DALWAYS_USE_MANAGED_MEMORY=ON on single GPUs.
Fix issue where an initially cubic box can become non-cubic with integrate.npt() and randomize_velocities().
#Download osx image resizer 0.9.2 for mac. generator
Fix a bug in the MPCD Gaussian random number generator that could lead to NaN values.
Send messages to replaced sys.stdout and sys.stderr streams.
HPMC integrators raise errors when restore_state=True and state information is missing.
Fix error when importing jit before hpmc.
CMake version 2.8.10.1 is now a minimum requirement for compiling from source.
Use random123 for random number generation.
Refactor handling of MPI_Comm inside library.
Fix a bug which could lead to invalid MPI simulations with nlist.cell() and nlist.stencil().
Fix a bug where rigid body energy and pressure deviated on the first time step after run().
Validate snapshot input in restore_snapshot.
Fix a problem where HOOMD could not be imported in mpi4py jobs.
Allow simulations with 0 bonds to specify bond potentials.
Fix compile errors with LLVM 8 and -DBUILD_JIT=on.
A block force, a constant force, and a sine force are implemented.
mpcd.stream includes an external mpcd.force acting on the MPCD particles.
MPCD supports virtual particle filling in bounded geometries through the set_filler method of mpcd.stream classes.
Users can implement other geometries as a plugin.
allows for simulations in parallel-plate channels.
Exclude neighbors that belong to the same floppy molecule.
Document upcoming feature removals and API changes.
Report detailed CUDA errors on initialization.
Fix plug-in builds when ENABLE_TBB or ALWAYS_USE_MANAGED_MEMORY CMake parameters are set.
Correct typos in HPMC example snippets.
Correctly document the functional form of the Buckingham pair potential.
#Download osx image resizer 0.9.2 for mac. download
Clarify singularity image download documentation.
Improve documentation of lattice_field and frenkel_ladd_energy classes.
Fix documentation about interoperability between md.mode_minimize_fire() and MPI.
Fix a bug in md.integrate.langevin() and md.integrate.bd() where on the GPU the value of gamma would be ignored.
test_group_rigid.py is run out of the md module.
mpcd.integrate supports integration of MD (solute) particles with bounce-back rules in MPCD streaming geometries.
_pore allows for simulations through parallel-plate (lamellar) pores.
Add _shape to save particle shape information in GSD files.