papyri
{ } Raw JSON

bundles / scipy 1.17.1 / scipy / _lib / array_api_extra / _lib / _lazy / lazy_apply

function

scipy._lib.array_api_extra._lib._lazy:lazy_apply

source: /scipy/_lib/array_api_extra/_lib/_lazy.py :57

Signature

def   lazy_apply ( func : Callable[P, Array | ArrayLike | Sequence[Array | ArrayLike]] ,  * args : Array | complex | None ,  shape : tuple[int | None, ...] | Sequence[tuple[int | None, ...]] | None = None ,  dtype : DType | Sequence[DType] | None = None ,  as_numpy : bool = False ,  xp : ModuleType | None = None ,  ** kwargs : P.kwargs )  →  Array | tuple[Array, ...]

Summary

Lazily apply an eager function.

Extended Summary

If the backend of the input arrays is lazy, e.g. Dask or jitted JAX, the execution of the function is delayed until the graph is materialized; if it's eager, the function is executed immediately.

Parameters

func : callable

The function to apply.

It must accept one or more array API compliant arrays as positional arguments. If as_numpy=True, inputs are converted to NumPy before they are passed to func. It must return either a single array-like or a sequence of array-likes.

func must be a pure function, i.e. without side effects, as depending on the backend it may be executed more than once or never.

*args : Array | int | float | complex | bool | None

One or more Array API compliant arrays, Python scalars, or None's.

If as_numpy=True, you need to be able to apply numpy.asarray to non-None args to convert them to NumPy; read notes below about specific backends.

shape : tuple[int | None, ...] | Sequence[tuple[int | None, ...]], optional

Output shape or sequence of output shapes, one for each output of func. Default: assume single output and broadcast shapes of the input arrays.

dtype : DType | Sequence[DType], optional

Output dtype or sequence of output dtypes, one for each output of func. dtype(s) must belong to the same array namespace as the input arrays. Default: infer the result type(s) from the input arrays.

as_numpy : bool, optional

If True, convert the input arrays to NumPy before passing them to func. This is particularly useful to make NumPy-only functions, e.g. written in Cython

or Numba, work transparently with array API-compliant arrays.

Default: False.

xp : array_namespace, optional

The standard-compatible namespace for args. Default: infer.

**kwargs : Any, optional

Additional keyword arguments to pass verbatim to func. They cannot contain Array objects.

Returns

: Array | tuple[Array, ...]

The result(s) of func applied to the input arrays, wrapped in the same array namespace as the inputs. If shape is omitted or a single tuple[int | None, ...], return a single array. Otherwise, return a tuple of arrays.

Raises

: ValueError

When xp=jax.numpy, the output shape is unknown (it contains None on one or more axes) and this function was called inside jax.jit.

: RuntimeError

When xp=sparse and auto-densification is disabled.

: Exception (backend-specific)

When the backend disallows implicit device to host transfers and the input arrays are on a non-CPU device, e.g. on GPU.

Notes

JAX

This allows applying eager functions to jitted JAX arrays, which are lazy. The function won't be applied until the JAX array is materialized. When running inside jax.jit, shape must be fully known, i.e. it cannot contain any None elements.

Using this with as_numpy=False is particularly useful to apply non-jittable JAX functions to arrays on GPU devices. If as_numpy=True, the jax:transfer_guard may prevent arrays on a GPU device from being transferred back to CPU. This is treated as an implicit transfer.

PyTorch, CuPy

If as_numpy=True, these backends raise by default if you attempt to convert arrays on a GPU device to NumPy.

Sparse

If as_numpy=True, by default sparse prevents implicit densification through numpy.asarray. This safety mechanism can be disabled.

Dask

This allows applying eager functions to Dask arrays. The Dask graph won't be computed until the user calls compute() or persist() down the line.

The function name will be prominently visible on the user-facing Dask dashboard and on Prometheus metrics, so it is recommended for it to be meaningful.

lazy_apply doesn't know if func reduces along any axes; also, shape changes are non-trivial in chunked Dask arrays. For these reasons, all inputs will be rechunked into a single chunk.

The outputs will also be returned as a single chunk and you should consider rechunking them into smaller chunks afterwards.

If you want to distribute the calculation across multiple workers, you should use dask.array.map_blocks, dask.array.map_overlap, dask.array.blockwise, or a native Dask wrapper instead of lazy_apply.

Dask wrapping around other backends

If as_numpy=False, func will receive in input eager arrays of the meta namespace, as defined by the ._meta attribute of the input Dask arrays. The outputs of func will be wrapped by the meta namespace, and then wrapped again by Dask.

See also

dask.array.blockwise
dask.array.map_blocks
dask.array.map_overlap
jax.pure_callback
jax.transfer_guard

Aliases

  • scipy.differentiate.xpx.lazy_apply