Dask Integration

The pyinterp.dask module provides helpers that compute the Statistics & Binning containers in parallel on Dask arrays. Each helper processes blocks independently using the same C++ kernels exposed by the in-memory API, then merges the per-block results with the += operator implemented by the underlying holders.

Dask is an optional dependency; install it with pip install dask[array].

Overview

The functions in this module are thin wrappers that mirror the eager classes of pyinterp. The table below maps each Dask helper to the in-memory counterpart it produces:

Dask helper	Returns	Eager equivalent
descriptive_statistics()	pyinterp.DescriptiveStatistics	pyinterp.DescriptiveStatistics
tdigest()	pyinterp.TDigest	pyinterp.TDigest
binning1d()	pyinterp.Binning1D	pyinterp.Binning1D
binning2d()	pyinterp.Binning2D	pyinterp.Binning2D
histogram2d()	pyinterp.Histogram2D	pyinterp.Histogram2D

Descriptive statistics & quantiles

descriptive_statistics(values[, weights, ...])	Compute descriptive statistics on a dask array.
tdigest(values[, weights, axis, ...])	Compute quantile estimates on a dask array using T-Digest.

Binning containers

binning1d(x, z, binning[, weights])	Accumulate values into 1D bins from a dask array.
binning2d(x, y, z, binning[, simple])	Accumulate values into 2D bins from dask arrays.

Histograms

histogram2d(x, y, z, histogram)

Accumulate values into a 2D histogram from dask arrays.

See also

• Statistics & Binning — eager (in-memory) versions of the containers returned by the helpers above.