TuckerTensorTrain.randn#

static t3toolbox.tucker_tensor_train.TuckerTensorTrain.randn(shape, tucker_ranks, tt_ranks, stack_shape=(), use_jax=False)#

def randn(
        shape: Tuple[int, ...],
        tucker_ranks: Tuple[int, ...],
        tt_ranks: Tuple[int, ...],
        stack_shape: Tuple[int, ...] = (),
        use_jax: bool = False,
) -> 'TuckerTensorTrain':

Construct a Tucker tensor train with random cores. Core entries are i.i.d. draws from N(0,1).

Parameters:

shape (Sequence[int]) – Shape of the TuckerTensorTrain. len(shape)=d.
tucker_ranks (Sequence[int], optional) – Tucker ranks. len(tucker_ranks)=d. Default (tucker_ranks=None): all Tucker ranks equal 1 .
tt_ranks (Sequence[int], optional) – TT ranks. len(tt_ranks)=d+1. Default (tt_ranks=None): all TT ranks equal 1.
stack_shape (Sequence[int], optional) – Stack shape. Default (stack_shape=()): No stacking.
use_jax (bool, optional) – Cores are jax arrays if True, and numpy arrays if False. (default: use_jax=False)

Returns:

Random TuckerTensorTrain with the desired shape and ranks.

Return type:

TuckerTensorTrain

Examples

>>> import numpy as np
>>> import t3toolbox.tucker_tensor_train as t3
>>> np.random.seed(0)
>>> shape = (14, 15, 16)
>>> tucker_ranks = (4, 5, 6)
>>> tt_ranks = (1, 3, 2, 1)
>>> stack_shape = (2, 3)
>>> x = t3.TuckerTensorTrain.randn(shape, tucker_ranks, tt_ranks, stack_shape=stack_shape) # random cores
>>> print(x.structure == (shape, tucker_ranks, tt_ranks, stack_shape))
True
>>> print(np.any(x.tucker_cores[0] != 0.0))  # cores are filled with N(0,1) draws, not zeros
True