Skip to content

Latest commit

 

History

History
249 lines (165 loc) · 5.52 KB

ChangeLog.md

File metadata and controls

249 lines (165 loc) · 5.52 KB
Raw

TODO

  • Make TabulatedEAM work for alloys. Probably useful to rename TabulatedEAM to FuncflEAM and TabulatedAlloyEAM to SetflEAM to distinguish by file format.

Change log

v0.10.2

  • Update Python interface for recent changes in ASE
  • Bug fix: Wrong neighbor list construction for small negative positions (#30)
  • Bug fix: Default system of units in DFT-D3 when using Python or LAMMPS (#27)

v0.10.1

  • Select elements in SlidingT

v0.10.0

  • DFT-D3 dispersion correction
  • Variable-charge model

v0.9.2

  • OpenMP parallelization of the grid portion of PME

v0.9.1

  • OpenMP parallelization of LJCut

v0.9.0

  • Variable charge model

v0.8.0

  • Support for DFTB3.

v0.7.0

  • Updated to support latest ASE master.
  • Fixed random hangs when running with LAMMPS.
  • Bug fix in C-H interaction of screened REBO2 potential when running within LAMMPS.

v0.6.0

  • Updated for ASE 3.15.
  • Bug fix in SCC-DFTB.

v0.5.6

  • Fixed memory leak.

v0.5.5

  • Improved introspection of the electronic structure obtained in tight-binding calculations.

v0.5.4

  • Fixed segfault in tight-binding materials database on Mac OS X.

v0.5.3

  • Support for ASE's Atoms.celldisp parameter
  • Bug fix: Atoms outside the simulation cell are now treated correctly in periodic and nonperiodic cells.

v0.5.2

  • Removed LAPACK dependency for everything that does not use tight-binding (in particular the LAMMPS moduls)
  • Updated LAMMPS pair_style for LAMMPS >= 07Sep16
  • Bug fix: Proper inclusion of numpy extra_link_args in setup.py
  • Bug fix: Fixed problem with tilted orthorhombic cells

v0.5.1

  • Bug fix: Set element charge from skf file. Only affects tight-binding runs with Slater-Koster tables from dftb.org

v0.5.0

  • Particle mesh Ewald Coulomb solver
  • Regression fix in empirical bond-order potentials (dimers were handle incorrectly) - regression introduced in 0.4.4

v0.4.6

  • More robust EAM (low and high density configurations)
  • More robust version extraction for LAMMPS and standalone code (uses versioneer)

v0.4.5

  • Maintenance: Fixed problem with passing c_null_ptr through c_f_pointer when using XLF

v0.4.4

  • Bug fix: Occasional NaNs in bond-order potentials
  • Maintenance: Fixed segfault on BlueGene

v0.4.3

  • Preparation for PyPI release
  • Minor fixes to tight-binding solver (when used with MPI domain decomposition)

v0.4.2

  • Support for new ASE calculator interface (thanks to James Kermode)

v0.4.1

  • Regression fix: Charge extrapolation did not work because charges were overriden with initial charges at every step. This was introduced when changing to the new ASE Calculator class.

v0.4.0

  • Python 3 compatibility.

  • Python interface based on new ASE Calculator base class.

  • Tight binding: Removed necessity for an 'elements.dat' file. Code auto-detects elements from files found in the database directory.

  • Tight binding: Added support for d, sd and pd electronic configurations.

  • LAMMPS interface automatically checks git fingerprint to ensure compatibility between pair_style and Atomistica library.

  • Fixed proper stopping of timer when errors are raised or passed in some parts of the code.

v0.3.2

  • General mechanism for object introspection from Python

  • Exposed NOTB internals to Python (Hamiltonian, overlap and density matrices, eigenvalues and eigenvectors if LAPACK solver is used)

  • Made NOTB per-bond analysis tools available from Python (added get_per_bond_property method)

  • Some bug fixes to parts of the standalone code

v0.3.1

  • Implemented charge extrapolation for SCC NOTB

  • Fixed a couple of OpenMP related regression that broke compilation.

  • Fixed a regression that lead to wrong unit conversion for Hubbard-Us.

  • Fixed a regression that lead to OutputEnergy not being called.

  • Coulomb solvers now have an energy_and_forces rather than a potential_and_field function.

v0.3.0

  • Added standalone molecular dynamics code. Source is in src/standalone and Makefiles are in build_standalone.

  • Dipatch generator now parses the additional 'features' key in the @meta data. Features can be 'mask', 'per_at' and 'per_bond'. This enables passing of mask, epot/wpot_per_at and epot/wpot_per_bond parameters to the potential. (These parameters can be omitted from the interface if unused now.)

  • Added mask parameter that can be used to turn on/off contribution from individual atoms. Contributions are additive, i.e. epot(mask) + epot(.not. mask) gives the total energy.

  • Fixed TabulatedEAM. Never worked properly but does now for unary systems.

  • Size of internal neighbor lists (EAM potentials and BOPs) is now computed on-the-fly.

v0.2.5

  • Corrected another bug regarding handling of PBCs in ASE.

v0.2.4

  • Corrected handling of PBCs in ASE.

v0.2.3

  • Fixed buffer overflow bug in atom type handling in LAMMPS interface.

v0.2.2

  • Added harmonic spring (with cutoff) potential.
  • Added double harmonic potential. (Two harmonic springs at different distance, useful to model an SC solid that is isotropically elastic.)

v0.2.1

  • Added proper handling of PBCs to ASE interface. (All systems were periodic so far independent of ASEs PBC setting.)

v0.2.0

  • Added support for non-orthogonal tight-binding (NOTB) with charge self-consistency (SCC).

  • Added Coulomb solvers (required for SCC-NOTB).

v0.1.3

  • Compatibility with IBM XL compiler suite on the BlueGene architecture.