chemparseplot.parse.trajectory.hdf5

HDF5 NEB reader for ChemGP output.

Added in version 1.2.0.

Reads neb_result.h5 (final converged path) and neb_history.h5 (per-step optimization history) produced by ChemGP’s Julia NEB optimizer.

These files contain pre-computed f_para and rxn_coord, so profile data can be read directly without recomputing the Henkelman-Jonsson tangent.

Module Contents

Functions

_read_path_group

Extract arrays from an HDF5 path group.

_reconstruct_atoms

Rebuild ASE Atoms from flattened positions and metadata.

load_neb_result

Read a ChemGP neb_result.h5 file.

load_neb_history

Read a ChemGP neb_history.h5 file.

result_to_profile_dat

Convert a result HDF5 to a (5, N) profile array.

result_to_atoms_list

Reconstruct ASE Atoms from a result HDF5 file.

history_to_profile_dats

Convert a history HDF5 to a list of (5, N) profile arrays.

history_to_landscape_df

Convert a history HDF5 to a landscape DataFrame.

Data

log

API

chemparseplot.parse.trajectory.hdf5.log

‘getLogger(…)’

chemparseplot.parse.trajectory.hdf5._read_path_group(grp: h5py.Group) → dict

Extract arrays from an HDF5 path group.

Parameters:

grp – An h5py Group containing images, energies, gradients, f_para, and rxn_coord datasets.

Returns:

Dictionary with numpy arrays for each field.

chemparseplot.parse.trajectory.hdf5._reconstruct_atoms(images: numpy.ndarray, atomic_numbers: numpy.ndarray | None, cell: numpy.ndarray | None, gradients: numpy.ndarray, energies: numpy.ndarray) → list[ase.Atoms]

Rebuild ASE Atoms from flattened positions and metadata.

Each image’s positions are (ndof,) where ndof = 3 * n_atoms. A SinglePointCalculator is attached with energy and forces (-gradients).

Parameters:

  • images – Array of shape (n_images, ndof).

  • atomic_numbers – Optional array of shape (n_atoms,).

  • cell – Optional array of shape (9,) (flattened 3x3 cell).

  • gradients – Array of shape (n_images, ndof).

  • energies – Array of shape (n_images,).

Returns:

List of ASE Atoms objects with calculators attached.

chemparseplot.parse.trajectory.hdf5.load_neb_result(h5_file: str) → dict

Read a ChemGP neb_result.h5 file.

Added in version 1.2.0.

Parameters:

h5_file – Path to the HDF5 result file.

Returns:

Dictionary with keys path (from :func:_read_path_group), convergence (dict of arrays), and metadata (dict of scalars and optional arrays).

chemparseplot.parse.trajectory.hdf5.load_neb_history(h5_file: str) → list[dict]

Read a ChemGP neb_history.h5 file.

Added in version 1.2.0.

Parameters:

h5_file – Path to the HDF5 history file.

Returns:

List of dicts (one per optimization step), sorted by step number. Each dict has the same structure as :func:_read_path_group.

chemparseplot.parse.trajectory.hdf5.result_to_profile_dat(h5_file: str) → numpy.ndarray

Convert a result HDF5 to a (5, N) profile array.

Added in version 1.2.0.

Layout matches eOn .dat format: [index, rxn_coord, energy, f_para, zeros].

Reads directly from pre-computed HDF5 fields – no tangent recomputation.

Parameters:

h5_file – Path to neb_result.h5.

Returns:

Array of shape (5, n_images).

chemparseplot.parse.trajectory.hdf5.result_to_atoms_list(h5_file: str) → list[ase.Atoms]

Reconstruct ASE Atoms from a result HDF5 file.

Added in version 1.2.0.

Parameters:

h5_file – Path to neb_result.h5.

Returns:

List of Atoms with SinglePointCalculators attached.

chemparseplot.parse.trajectory.hdf5.history_to_profile_dats(h5_file: str) → list[numpy.ndarray]

Convert a history HDF5 to a list of (5, N) profile arrays.

Added in version 1.2.0.

One array per optimization step, sorted by step number.

Parameters:

h5_file – Path to neb_history.h5.

Returns:

List of arrays, each of shape (5, n_images).

chemparseplot.parse.trajectory.hdf5.history_to_landscape_df(h5_file: str, ira_kmax: float = 1.8)

Convert a history HDF5 to a landscape DataFrame.

Added in version 1.2.0.

Reconstructs Atoms per step for RMSD coordinates, uses pre-computed f_para for synthetic gradients.

Parameters:

  • h5_file – Path to neb_history.h5.

  • ira_kmax – kmax factor for IRA alignment.

Returns:

Polars DataFrame with columns [r, p, grad_r, grad_p, z, step].