Creation

Creation#

zeros#

def zeros(shape: 'ShapeLike', *, dtype: 'DType | None' = None, device: 'Device | None' = None, is_traced: 'bool' = False):

Return a tensor of shape filled with zeros.

Parameters

  • shape – Output shape.

  • dtype – Element dtype. Uses the current default if None.

  • device – Target device. Uses the current default if None.

Returns

Zero-valued tensor of the given shape and dtype.

ones#

def ones(shape: 'ShapeLike', *, dtype: 'DType | None' = None, device: 'Device | None' = None, is_traced: 'bool' = False):

Return a tensor of shape filled with ones.

Parameters

  • shape – Output shape.

  • dtype – Element dtype. Uses the current default if None.

  • device – Target device. Uses the current default if None.

Returns

One-valued tensor of the given shape and dtype.

full#

def full(shape: 'ShapeLike', value: 'Number', *, dtype: 'DType | None' = None, device: 'Device | None' = None, is_traced: 'bool' = False):

Return a tensor of shape filled with value.

Parameters

  • shape – Output shape.

  • value – Fill value (scalar).

  • dtype – Element dtype. Uses the current default if None.

  • device – Target device. Uses the current default if None.

Returns

Tensor with all elements equal to value.

constant#

def constant(value: 'NestedArray | Number', *, dtype: 'DType | None' = None, device: 'Device | None' = None):

Create a tensor from a Python scalar, list, or NumPy array.

Scalars and 0-d arrays are wrapped with the default dtype unless dtype is specified. Multi-dimensional arrays/lists are converted via DLPack without copying.

Parameters

  • value – Python int, float, bool, complex, list, or np.ndarray.

  • dtype – Target element dtype. Inferred from value if None.

  • device – Target device. Uses the current default if None.

Returns

A realized Tensor wrapping value.

arange#

def arange(start: 'int' = 0, stop: 'int | None' = None, step: 'int' = 1, *, dtype: 'DType | None' = None, device: 'Device | None' = None):

Return a 1-D tensor with evenly spaced values in [start, stop).

When called with a single positional argument, it is treated as stop and start defaults to 0, matching NumPy / PyTorch semantics.

Parameters

  • start – Start of the interval (inclusive). Default: 0.

  • stop – End of the interval (exclusive). If None, start is used as stop and start becomes 0.

  • step – Spacing between values. Default: 1.

  • dtype – Element dtype. Uses the current default if None.

  • device – Target device. Uses the current default if None.

Returns

1-D tensor of length ceil((stop - start) / step).

uniform#

def uniform(shape: 'ShapeLike' = (), low: 'float' = 0.0, high: 'float' = 1.0, *, dtype: 'DType | None' = None, device: 'Device | None' = None):

Return a tensor of shape with values sampled from U(low, high).

Parameters

  • shape – Output shape.

  • low – Lower bound of the uniform distribution.

  • high – Upper bound of the uniform distribution.

  • dtype – Element dtype. Uses the current default if None.

  • device – Target device. Uses the current default if None.

Returns

Tensor with elements drawn uniformly at random from [low, high).

normal#

def normal(shape: 'ShapeLike' = (), mean: 'float' = 0.0, std: 'float' = 1.0, *, dtype: 'DType | None' = None, device: 'Device | None' = None):

Return a tensor of shape with values sampled from N(mean, std²).

Also accessible as nabla.normal.

Parameters

  • shape – Output shape.

  • mean – Mean of the Gaussian distribution. Default: 0.0.

  • std – Standard deviation. Default: 1.0.

  • dtype – Element dtype. Uses the current default if None.

  • device – Target device. Uses the current default if None.

Returns

Tensor with elements drawn from a normal distribution.

gaussian#

def gaussian(shape: 'ShapeLike' = (), mean: 'float' = 0.0, std: 'float' = 1.0, *, dtype: 'DType | None' = None, device: 'Device | None' = None):

Return a tensor of shape with values sampled from N(mean, std²).

Also accessible as nabla.normal.

Parameters

  • shape – Output shape.

  • mean – Mean of the Gaussian distribution. Default: 0.0.

  • std – Standard deviation. Default: 1.0.

  • dtype – Element dtype. Uses the current default if None.

  • device – Target device. Uses the current default if None.

Returns

Tensor with elements drawn from a normal distribution.

hann_window#

def hann_window(window_length: 'int', *, periodic: 'bool' = True, dtype: 'DType | None' = None, device: 'Device | None' = None):

Return a 1-D Hann (raised cosine) window of length window_length.

The window follows the convention used in signal processing: w[n] = 0.5 * (1 - cos(2π n / N)) where N is the window size.

Parameters

  • window_length – Number of points in the window.

  • periodic – If True (default), generates a periodic window for use in spectral analysis. If False, generates a symmetric window.

  • dtype, default: float32 – Element dtype. Defaults to float32.

  • device – Target device. Uses the current default if None.

Returns

1-D Tensor of shape (window_length,).

triu#

def triu(x: 'Tensor', k: 'int' = 0) -> 'Tensor':

Return the upper triangular part of a matrix (or batch of matrices).

Elements below the k-th diagonal are zeroed out.

Parameters

  • x – Input tensor of shape (*, M, N).

  • k – Diagonal offset. k=0 (default) is the main diagonal, k>0 is above it, k<0 is below.

Returns

Tensor of the same shape as x with the lower triangle zeroed.

tril#

def tril(x: 'Tensor', k: 'int' = 0) -> 'Tensor':

Return the lower triangular part of a matrix (or batch of matrices).

Elements above the k-th diagonal are zeroed out.

Parameters

  • x – Input tensor of shape (*, M, N).

  • k – Diagonal offset. k=0 (default) is the main diagonal, k>0 is above it, k<0 is below.

Returns

Tensor of the same shape as x with the upper triangle zeroed.

zeros_like#

def zeros_like(x: 'Tensor') -> 'Tensor':

Return a zero tensor with the same shape, dtype, device, and sharding as x.

Parameters

  • x – Reference tensor.

Returns

Tensor of zeros matching all metadata of x.

ones_like#

def ones_like(x: 'Tensor') -> 'Tensor':

Return a ones tensor with the same shape, dtype, device, and sharding as x.

Parameters

  • x – Reference tensor.

Returns

Tensor of ones matching all metadata of x.

full_like#

def full_like(x: 'Tensor', value: 'Number') -> 'Tensor':

Return a tensor filled with value, matching x’s shape, dtype, device, and sharding.

Parameters

  • x – Reference tensor.

  • value – Scalar fill value.

Returns

Tensor filled with value, with the same metadata as x.

Contents