MLXBackend

class MLXBackend(device='cpu', float_dtype=None, complex_dtype=None, int_dtype=None, overflow_safe_dtype=None, **kwargs)

Bases: NumpyFFTWBackend

A mlx-based matching backend.

Methods

MLXBackend.abs(*args, **kwargs)	Compute the absolute of array elements.
MLXBackend.add(x1, x2[, out])	Element-wise addition of arrays.
MLXBackend.arange(*args, **kwargs)	Arange values in evenly spaced interval.
MLXBackend.argsort(*args, **kwargs)	Compute the indices to sort a given input array.
MLXBackend.astype(arr, dtype)	Change the datatype of arr.
MLXBackend.build_fft(fast_shape, ...)	Build fft builder functions.
MLXBackend.center_of_mass(arr[, cutoff])	Computes the center of mass of a numpy ndarray instance using all available elements.
MLXBackend.clip(*args, **kwargs)	Clip elements of arr.
MLXBackend.compute_convolution_shapes(...)	Computes regular, optimized and fourier convolution shape.
MLXBackend.concatenate(*args, **kwargs)	Join a sequence of objects along an existing axis.
MLXBackend.datatype_bytes(dtype)	Return the number of bytes occupied by a given datatype.
MLXBackend.device_count()	Returns the number of available GPU devices.
MLXBackend.divide(*args, **kwargs)	Element-wise division of arrays.
MLXBackend.dot(*args, **kwargs)
MLXBackend.einsum(*args, **kwargs)	Compute the einstein notation based summation.
MLXBackend.eps(dtype)	Returns the minimal difference representable by dtype.
MLXBackend.extract_center(arr, newshape)	Extract the centered portion of an array based on a new shape.
MLXBackend.fill(arr, value)	Fills arr in-place with a given value.
MLXBackend.free_cache()	Free cached objects allocated by backend.
MLXBackend.from_sharedarr(arr)	Returns an array of given shape and dtype from shared memory location.
MLXBackend.full(shape, fill_value[, dtype])	Returns an array filled with fill_value of specified shape and dtype.
MLXBackend.get_available_memory()	Returns the available memory available for computations in bytes.
MLXBackend.get_fundamental_dtype(arr)	Given an array instance, returns the corresponding fundamental python type, i.e., int, float or complex.
MLXBackend.identity(*args, **kwargs)
MLXBackend.indices(arr)	Creates an array representing the index grid of an input.
MLXBackend.max(*args, **kwargs)	Compute the maximum of array elements.
MLXBackend.max_filter_coordinates(...)	Identifies local maxima in score_space separated by min_distance.
MLXBackend.max_score_over_rotations(scores, ...)	Update elements in max_scores and rotations where scores is larger than max_scores with score and rotation_index, respectivelty.
MLXBackend.maximum(*args, **kwargs)	Compute the element wise maximum of arr1 and arr2.
MLXBackend.mean(*args, **kwargs)	Compute the mean of array elements.
MLXBackend.min(*args, **kwargs)	Compute the minimum of array elements.
MLXBackend.minimum(*args, **kwargs)	Compute the element wise minimum of arr1 and arr2.
MLXBackend.mod(arr1, arr2[, out])	Element-wise modulus of arrays.
MLXBackend.multiply(x1, x2[, out])	Element-wise multiplication of arrays.
MLXBackend.norm_scores(arr, exp_sq, sq_exp, ...)	Normalizes arr by the standard deviation ensuring numerical stability.
MLXBackend.power(*args, **kwargs)	Compute the n-th power of an array.
MLXBackend.repeat(*args, **kwargs)	Repeat each array element a specified number of times.
MLXBackend.reshape(*args, **kwargs)
MLXBackend.reverse(arr)	Reverse the order of elements in an array along all its axes.
MLXBackend.rigid_transform(arr, rotation_matrix)	Performs a rigid transformation.
MLXBackend.roll(a, shift, axis, **kwargs)	Roll array elements along a specified axis.
MLXBackend.set_device(device_index)	Context manager that sets active compute device device for operations.
MLXBackend.size(arr)	Compute the number of elements of arr.
MLXBackend.sqrt(*args, **kwargs)	Compute the square root of array elements.
MLXBackend.square(*args, **kwargs)	Compute the square of array elements.
MLXBackend.stack(*args, **kwargs)	Join a sequence of objects along a new axis.
MLXBackend.std(arr, axis)	Compute the standad deviation of array elements.
MLXBackend.subtract(*args, **kwargs)	Element-wise subtraction of arrays.
MLXBackend.sum(*args, **kwargs)	Compute the sum of array elements.
MLXBackend.to_backend_array(arr)	Convert a numpy array instance to backend array type.
MLXBackend.to_cpu_array(arr)	Convert an array of a given backend to a CPU array of that backend.
MLXBackend.to_numpy_array(arr)	Convert an array of given backend to a numpy array.
MLXBackend.to_sharedarr(arr[, ...])	Converts an array to an object shared in memory.
MLXBackend.tobytes(arr)	Compute the bytestring representation of arr.
MLXBackend.topk_indices(arr, k)	Determinces the indices of largest elements.
MLXBackend.topleft_pad(arr, shape[, padval])	Returns an array that has been padded to a specified shape with a padding value at the top-left corner.
MLXBackend.transpose(arr)	Compute the transpose of arr.
MLXBackend.tril_indices(*args, **kwargs)	Compute indices of upper triangular matrix
MLXBackend.unique(*args, **kwargs)	Find the unique elements of an array.
MLXBackend.unravel_index(indices, shape)	Convert flat index to array indices.
MLXBackend.where(*args, **kwargs)	Return elements from input depending on condition.
MLXBackend.zeros(shape[, dtype])	Returns an aligned array of zeros with specified shape and dtype.