permute

Return tensor with permuted axes.

permute

permute(tensor: Tensor, order: Sequence[int]) -> Tensor

Return a new tensor with permuted axes according to the provided order.

This functional version creates a fully independent tensor by cloning all data blocks. For an efficient version that shares underlying tensors, use tensor.permute(order, in_place=False).

Parameters:

Name	Type	Description	Default
tensor	Tensor	The input tensor to permute.	required
order	Sequence[int]	Sequence of integer axes specifying the new ordering. Must be a permutation of range(len(tensor.indices)).	required

Returns:

Type	Description
Tensor	A new tensor instance with reordered indices and cloned permuted blocks.

Raises:

Type	Description
ValueError	If order is not a valid permutation.

Notes

For non-Abelian (SU2) tensors, permutation involves R-symbols that transform the outer multiplicity (OM) indices. The weights are updated by matrix multiplication with the R-symbol: new_weights = old_weights @ R.

Examples:

>>> from nicole import permute, Tensor
>>> # Assuming t is a 3-index tensor with itags [a, b, c]
>>> t_perm = permute(t, [2, 0, 1])  # Reorder to [c, a, b]

Description

Returns a fully independent tensor with axes reordered according to the specified permutation. All data blocks are deep-copied, so the result is completely isolated from the original. For a memory-efficient version that shares storage with the original, use Tensor.permute() (default in_place=False). To modify the tensor in place, use Tensor.permute(order, in_place=True).

See Also

• Tensor.permute: Method form (shared-storage default, or in-place)
• transpose: Reverse axis order
• Examples: Manipulation

Notes

The order parameter must be a valid permutation of range(len(tensor.indices)). Each block is transposed accordingly.