legendre_decomp.utils

Utility functions

Functions

default_B(shape, order[, xp])	Vectorized implementation of the default B tensor.
ring_decomposition_B(shape)	Construct B tensor for ring decomposition.
index_channel_B(shape)	２次の近似から、index - channel 結合を切断する.

Module Contents

legendre_decomp.utils.default_B(shape, order, xp=np)

Vectorized implementation of the default B tensor.

Parameters:

• shape (Sequence[int]) – Shape of the corresponding X tensor.

• order (int) – Order of the B tensor.

• xp (ModuleType) – Array module, either numpy (CPU) or cupy (CUDA/GPU)

Returns:

Default B tensor of specified order.

Return type:

array-like

legendre_decomp.utils.ring_decomposition_B(shape)

Construct B tensor for ring decomposition.

Parameters:

shape (Sequence[int]) – Shape of the corresponding X tensor.

Returns:

B tensor for ring decomposition.

Return type:

list[tuple[int, Ellipsis]]

legendre_decomp.utils.index_channel_B(shape)

２次の近似から、index - channel 結合を切断する.

Parameters:

shape (Sequence[int]) – Shape of the corresponding X tensor.

Returns:

B tensor with broken index-channel bond for larger than 2D.

Return type:

list[tuple[int, Ellipsis]]