Source code for impy.random

from __future__ import annotations
import numpy as np
from typing import Callable, Literal, TypeVar
import functools
from .arrays import ImgArray, LazyImgArray
from .arrays.bases import MetaArray
from .array_api import xp
from .core import asarray
from .axes import AxesLike

def wraps(npfunc):
    def _wraps(ipfunc):
        ipfunc.__doc__ = npfunc.__doc__
        return ipfunc
    return _wraps

def __getattr__(name: str):
    xpfunc = getattr(xp.random, name)
    @wraps(xpfunc)
    def _func(*args, **kwargs) -> ImgArray:
        name = kwargs.pop("name", xpfunc.__name__)
        axes = kwargs.pop("axes", None)
        out = xp.asnumpy(xpfunc(*args, **kwargs))
        return asarray(out, name=name, axes=axes)
    return _func


def _normalize_like(size, name, axes, like: MetaArray | LazyImgArray | None):
    if like is not None:
        name = name or like.name
        axes = axes or like.axes
        size = size or like.shape
    return size, name, axes
    

[docs]@wraps(np.random.random) def random( size, *, name: str = None, axes: str = None, like: MetaArray | LazyImgArray =None, ) -> ImgArray: size, name, axes = _normalize_like(size, name, axes, like) name = name or "random" return asarray(xp.asnumpy(xp.random.random(size)), name=name, axes=axes)
[docs]@wraps(np.random.normal) def normal( loc: float = 0.0, scale: float = 1.0, size=None, *, name: str = None, axes: str = None, like: MetaArray | LazyImgArray | None = None, ) -> ImgArray: size, name, like = _normalize_like(size, name, axes, like) name = name or f"normal({loc}, {scale})" return asarray(xp.asnumpy(xp.random.normal(loc, scale, size)), name=name, axes=axes)
[docs]def random_uint8( size: int | tuple[int], *, name: str = None, axes: str = None, like: MetaArray | LazyImgArray =None, ) -> ImgArray: """ Return a random uint8 image, ranging 0-255. Parameters ---------- size : int or tuple of int Image shape. name : str, optional Image name. axes : str, optional Image axes. Returns ------- ImgArray Random Image in dtype ``np.uint8``. """ size, name, like = _normalize_like(size, name, axes, like) arr = xp.random.randint(0, 255, size, dtype=np.uint8) name = name or "random_uint8" return asarray(xp.asnumpy(arr), name=name, axes=axes)
[docs]def random_uint16( size, *, name: str = None, axes: str = None, like: MetaArray | LazyImgArray =None, ) -> ImgArray: """ Return a random uint16 image, ranging 0-65535. Parameters ---------- size : int or tuple of int Image shape. name : str, optional Image name. axes : str, optional Image axes. Returns ------- ImgArray Random Image in dtype ``np.uint16``. """ size, name, like = _normalize_like(size, name, axes, like) arr = xp.random.randint(0, 65535, size, dtype=np.uint16) name = name or "random_uint16" return asarray(xp.asnumpy(arr), name=name, axes=axes)
[docs]def default_rng(seed) -> ImageGenerator: """Get the default random number generator.""" return ImageGenerator(xp.random.default_rng(seed))
[docs]class ImageGenerator: def __init__(self, rng: xp.random.Generator): self._rng = rng
[docs] def standard_normal( self, size: int | tuple[int, ...] | None = None, dtype = None, *, axes: AxesLike | None = None, name: str | None = None, like: MetaArray | LazyImgArray =None, ) -> ImgArray: size, name, axes = _normalize_like(size, name, axes, like) arr = self._rng.standard_normal(size=size, dtype=dtype) if np.isscalar(arr): return arr return asarray(arr, axes=axes, name=name)
[docs] def standard_exponential( self, size: int | tuple[int, ...] | None = None, dtype = None, method: Literal["zig", "inv"] = None, *, axes: AxesLike | None = None, name: str | None = None, like: MetaArray | LazyImgArray =None, ) -> ImgArray: size, name, axes = _normalize_like(size, name, axes, like) arr = self._rng.standard_exponential(size=size, dtype=dtype, method=method) if np.isscalar(arr): return arr return asarray(arr, axes=axes, name=name)
[docs] def random( self, size: int | tuple[int, ...] | None = None, dtype = None, *, axes: AxesLike | None = None, name: str | None = None, like: MetaArray | LazyImgArray =None, ) -> ImgArray: size, name, axes = _normalize_like(size, name, axes, like) arr = self._rng.random(size=size, dtype=dtype) if np.isscalar(arr): return arr return asarray(arr, axes=axes, name=name)
[docs] def normal( self, loc: float | np.ndarray = 0., scale: float | np.ndarray = 1., size: int | tuple[int, ...] | None = None, *, axes: AxesLike | None = None, name: str | None = None, like: MetaArray | LazyImgArray =None, ) -> ImgArray: size, name, axes = _normalize_like(size, name, axes, like) arr = self._rng.normal(loc=loc, scale=scale, size=size) if np.isscalar(arr): return arr return asarray(arr, axes=axes, name=name)
[docs] def poisson( self, lam: float, size: int | tuple[int, ...] | None = None, *, axes: AxesLike | None = None, name: str | None = None, like: MetaArray | LazyImgArray =None, ) -> ImgArray: size, name, axes = _normalize_like(size, name, axes, like) arr = self._rng.poisson(lam=lam, size=size) if np.isscalar(arr): return arr return asarray(arr, axes=axes, name=name)
[docs] def random_uint8( self, size: int | tuple[int], *, name: str = None, axes: str = None, like: MetaArray | LazyImgArray =None, ) -> ImgArray: size, name, axes = _normalize_like(size, name, axes, like) arr = self._rng.integers(0, 255, size, dtype=np.uint8) name = name or "random_uint8" return asarray(xp.asnumpy(arr), name=name, axes=axes)
[docs] def random_uint16( self, size, *, name: str = None, axes: str = None, like: MetaArray | LazyImgArray =None, ) -> ImgArray: size, name, axes = _normalize_like(size, name, axes, like) arr = self._rng.integers(0, 65535, size, dtype=np.uint16) name = name or "random_uint16" return asarray(xp.asnumpy(arr), name=name, axes=axes)
del wraps