GAtor 1.1

    • Fitness functions: energy-based, cluster-based
    • Selection: roulette wheel, tournament
    • Crossover operators: Standard, Symmetric
    • Mutation operators: lattice strain, molecular rotations, molecular translations, molecular permutations, permutation-rotation, permutation-reflection
    • Geometry optimization and energy evaluation: FHI-aims
    • Parallelization: subprocessing of FHI-aims, spawning of independent GA replicas which communicate through I/O, multiprocessing within each replica for child generation.
    • References: J. Chem. Theory Comput., 14, 2246 (2018)Faraday Discussions, 211, 61 (2018)
New Features:
    • Python 3:All features from GAtor 1.0 written in Python 2 have been ported to Python 3 in the 1.1 release.
    • GAtor Package:GAtor 1.1 is now installed as Python Package, along with all of its external dependencies, usingsetuptoolsthrough thesetup.py file. Also distributed is the cgenarris library and the ibslib molecular crystal productivity Python library.
    • Fast Structure Checks:Faster structure checks which are responsible for determining if new child structures are physically reasonable molecular crystal structures have been ported from Genarris 2.0 into GAtor 1.1. These structure checks take into account the reduction of distances between intermolecular hydrogen bonds, as described in Genarris 2.0.
    • Radial Symmetry Functions (RSF) descriptor for clustering

Genarris 2.0

  • Python: 3; Libraries: pymatgen, spglib, ASE, sci-kit learn, SciPy, mpi4py, PyTorch
  • Descriptors: radial distribution function (RDF), relative coordinate descriptor (RCD)
  • Energy evaluations: FHI-aims
  • Parallelization: MPI (via mpi4py)
  • Workflows: Robust
  • Reference: Comput. Phys. Commun. 250, 107170 (2020)
New Features:
  • Machine-learned model for volume estimation
  • Generation in special Wyckoff positions of space groups
  • Updated settings for strong hydrogen bonds
  • Streamlined execution of user-defined workflows

Ogre 2

New Features:
  • Lattice matching for inorganic epitaxial interfaces
  • Miller index scan to find the best domain-matched interface
  • Geometric score function for fast surface matching
  • Bayesian optimization to find the optimal in-plane registry and interfacial distance
  • Ranking score function
  • Automated surface and interface model construction
  • Automated surface passivation
  • Automated interface energy evaluation

Archived Versions

GAtor 1.0

  • Python: 2.7; Libraries: pymatgen, spglib, sci-kit learn
  • Fitness functions: energy-based, cluster-based
  • Descriptors: radial distribution function (RDF), relative coordinate descriptor (RCD), lattice parameter-based descriptor
  • Selection: roulette wheel, tournament
  • Crossover operators: Standard, Symmetric
  • Mutation operators: lattice strain, molecular rotations, molecular translations, molecular permutations, permutation-rotation, permutation-reflection
  • Geometry optimization and energy evaluation: FHI-aims
  • Parallelization: subprocessing of FHI-aims, spawning of independent GA replicas which communicate through I/O, multiprocessing within each replica for child generation.
  • References: J. Chem. Theory Comput., 14, 2246 (2018); Faraday Discussions, 211, 61 (2018)

Genarris 1.0

  • Python: 2.7; Libraries: pymatgen, spglib, ase, aimsutils, sci-kit learn
  • Descriptors: radial distribution function (RDF), relative coordinate descriptor (RCD)
  • Energy evaluations with the Harris approximaiton: FHI-aims
  • Parallelization: subprocessing of FHI-aims, spawning of independent Genarris partitions (replicas) which communicate through I/O, multiprocessing within each replica for executing procedures such as structure generation batch
  • Workflows: Diverse, Random, Energy, Rigorous
  • Reference: J. Chem. Phys., 148, 241701 (2018)

Ogre 1.0

  • Python: 3; Libraries: pymatgen, ASE.
  • Energy evaluations: FHI-aims.
  • Generates slab models of molecular crystal surfaces without breaking intra-molecular covalent bonds.
  • Graph representation used to identify molecules.
  • Surfaces cleaved along user-defined Miller indices with user-defined number of layers and vacuum space.
  • Automatic identification of all symmetry non-equivalent surfaces and all possible surface terminations for given Miller indices/
  • OgreSWAMP utilities streamline the calculation of surface energies and Wulff shapes.
  • Reference: J. Chem. Phys. 152, 244122 (2020)

Ogre 1.1

  • Python: 3; Libraries: pymatgen, ASE.
  • Energy evaluations: FHI-aims.
  • Generates slab models of molecular crystal surfaces without breaking intra-molecular covalent bonds.
  • Graph representation used to identify molecules.
  • Surfaces cleaved along user-defined Miller indices with user-defined number of layers and vacuum space.
  • Automatic identification of all symmetry non-equivalent surfaces and all possible surface terminations for given Miller indices/
  • OgreSWAMP utilities streamline the calculation of surface energies and Wulff shapes.
  • Reference: J. Chem. Phys. 152, 244122 (2020)
New Features:
  • Accelerated slab cleaving
  • Wulffmaker tool for plotting Wulff shapes
  • Updated algorithm for finding unique planes
  • Bug fixes