zarr-matlab¶

Read and write Zarr v3 arrays natively in MATLAB — no Python required.

Zarr is a chunked, compressed array storage format for scientific data, designed for cloud storage and parallel I/O. zarr-matlab is a pure-MATLAB implementation of the Zarr v3 specification with byte-level interoperability against zarr-python: data written by either implementation reads back identically in the other, verified by CI on every commit.

z = zarr.create("weather.zarr", [720 1440], "single", ...
    Path="temp", ChunkShape=[180 360], ...
    Codecs={zarr.codecs.ZstdCodec(5)}, FillValue=single(NaN));

z(:, :) = single(20 + 5 * randn(720, 1440));   % write with normal indexing
block = z(1:100, end-99:end);                  % reads only the chunks it needs

Installation¶

Toolbox (recommended): download zarr-matlab-<version>.mltbx from the latest release and double-click it. Prebuilt zstd/blosc/crc32c codecs for Linux, Windows, and Apple Silicon are included.

From source: clone the repo and add its root (the folder containing +zarr) to your MATLAB path. Build the compression codecs once with run tools/build_mex.m (needs a C compiler plus libzstd/libblosc).

Requires MATLAB R2022b or newer with a JVM (used for gzip).

Feature highlights¶

Arrays and groups with attributes and dimension_names, accessed with ordinary MATLAB indexing
Every Zarr v3 data type: integers, floats (incl. float16), complex, variable-length strings and bytes, datetime64/timedelta64
Codecs: zstd, blosc (lz4/zstd/… with shuffle/bitshuffle), gzip, crc32c checksums, transpose
Sharding: many small chunks per stored object, with efficient ranged partial reads
Stores: local directories, in-memory, zip archives, read-only HTTP(S)
Consolidated metadata for single-read hierarchy opens on remote storage

How this fits the MATLAB Zarr landscape¶

MathWorks' official Zarr support covers format v2 only. scalableminds/zarr-matlab binds the zarrs Rust implementation. This library is a pure-MATLAB v3 implementation with complete dtype coverage and CI-verified zarr-python round trips.

Documentation is tested

Every MATLAB code block in these pages is executed by the test suite (tests/TestDocs.m) on every commit — examples cannot silently rot.