Scientific array metadata

NumPy, hierarchical scientific formats, and matrix files return layout and catalog fields. Some also include column statistics when data can be sampled as a 2-D table — see when there are no columns.

npy_metadata

Field	Meaning
byte_count	File size
format_version, dtype, shape, fortran_order	ArrayLayoutSummary
header_region_bytes, data_offset, data_region_bytes, expected_data_bytes_from_dtype	Payload bounds
row_count, column_count, stats_rows_sampled, columns	Optional when rank ≤ 2 and numeric dtype
tensor3d	Rank-3 plane min/max/mean/stdev (capped)

npz_metadata

Field	Meaning
byte_count	Archive size
zip_entry_count	ZIP member count
npy_entries_scanned, npy_entries	Per-member summaries (same fields as npy_metadata per entry)

hdf5_metadata

Field	Meaning
byte_count	File size
superblock_version	HDF5 superblock version
root_member_count, root_dataset_count, root_attribute_count	Root group summary
groups_visited, datasets_visited	Walk counters
datasets	{ path, shape, datatype_class, inspect_error? }[]
walk_truncated, parse_error	Caps or parse failures

netcdf_metadata

Field	Meaning
byte_count	File size
netcdf4_model	NetCDF-4 / HDF5-backed
global_attributes	{ name, value }[]
dimension_count, dimensions_sample	Dimension names and lengths
variable_count, variables	{ name, shape, dimension_names, vartype, attributes? }[]
metadata_truncated, parse_error	Caps or parse failures

mtx_metadata

Field	Meaning
byte_count	File size
storage	sparse or dense
num_stored_values, symmetry	Matrix Market header
numeric_stats_include_implicit_zeros	Sparse column stats include implicit zeros
row_count, column_count, columns	Logical matrix size and column profiles
parse_error	Parse failure message

mat_metadata

Field	Meaning
byte_count	File size
mat_format	e.g. v7 classic; v7.3 HDF5 flagged
variable_count, variables_scanned	Variable catalog
entries	Per-variable { name, layout, global?, struct_subtree?, value? }
file_parse_error	Top-level parse error

Non-scalar entries may include a global block (whole-array stats) instead of per-column profiles.

zarr_metadata

Field	Meaning
byte_count	Store size on disk
root_group_metadata	Root group Zarr version (2 / 3)
array_entries_scanned, array_entries	Per-array { name, layout, common?, tensor3d?, num_chunks? }

tetration_metadata (.tet)

Field	Meaning
byte_count	File size
layout_version, dataset_count	Container header
chunk_count, raw_chunk_count, zstd_chunk_count, history_event_count	Chunk index summary
datasets	{ name, dtype, shape, chunk_shape, layout, common?, tensor3d?, query_error? }[]
datasets_stats_skipped, parse_error	Query/stat caps or parse errors

See tetration-docs for the .tet on-disk layout.

Related: Column statistics, Metadata overview.