bluepyemodel.tools.checkpoint_hdf5

HDF5 checkpoint reader for BluePyOpt optimisation results.

Provides read_checkpoint_h5() which returns a dict compatible with all BluePyEModel analysis functions (store_best_model, optimisation plot, evolution_parameters_density, check_optimisation_state).

Also provides convert_checkpoint() for pickle → HDF5 conversion.

Functions

convert_checkpoint(pickle_path[, ...])

Convert a BluePyOpt pickle checkpoint to an HDF5 file.

detect_optimizer(cp)

Auto-detect the optimizer type from checkpoint keys.

read_checkpoint_h5(h5_path)

Read an HDF5 checkpoint and return a dict mimicking a pickle checkpoint.
convert_checkpoint(pickle_path, output_path=None, optimizer_override=None)

Convert a BluePyOpt pickle checkpoint to an HDF5 file.

Parameters:

  • pickle_path (str) – Path to the .pkl checkpoint file.

  • output_path (str) – Path for the output .h5 file (default: same name with .h5).

  • optimizer_override (str) – Force optimizer type (“IBEA”, “CMA_SO”, “CMA_MO”).

Returns:

Path to the created HDF5 file.

Return type:

str

detect_optimizer(cp)

Auto-detect the optimizer type from checkpoint keys.

read_checkpoint_h5(h5_path)

Read an HDF5 checkpoint and return a dict mimicking a pickle checkpoint.

Compatible with all BluePyEModel analysis consumers:

  • store_best_model

  • optimisation() plot

  • evolution_parameters_density()

  • access_point.__str__

  • check_optimisation_state

Returns:

(run_dict, seed)