Data Processing#

Utilities for merging and processing experiment results.

Utilities for merging and post-processing experiment results stored as NumPy .npz archives.

online_estimators.data.processing.merge_npz(files, noise_labels, out='merged.npz')[source]#

Concatenate several .npz experiment result files along the trajectory axis and save a single merged archive.

All files must share the same set of keys. Keys whose first dimension matches len(traj_labels) are treated as trajectory-indexed and concatenated along axis=0. All remaining keys are copied verbatim from the first file.

Parameters:

files (sequence of str or Path) – Paths to the .npz archives.
noise_labels (sequence of str) – Human-readable noise level for each file (e.g. ["low", "high"]). Must have the same length as files.
out (str or Path) – Destination path for the merged archive.

Raises:

ValueError – If the files have different key sets or incompatible trailing shapes.

Return type:

None