cr.sparse.data.sparse_normal_representations

cr.sparse.data.sparse_normal_representations(key, D, K, S=1)[source]

Generates a set of sparse model vectors with normally distributed non-zero entries.

  • Each vector is K-sparse.

  • The non-zero basis indexes are randomly selected and shared among all vectors.

  • The non-zero values are normally distributed.

Parameters

  • key – a PRNG key used as the random key.

  • D (int) – Dimension of the model space

  • K (int) – Number of non-zero entries in the sparse model vectors

  • S (int) – Number of sparse model vectors (default 1)

Returns

A tuple consisting of (i) a matrix of sparse model vectors (ii) an index set of locations of non-zero entries

Return type

(jax.numpy.ndarray, jax.numpy.ndarray)

Example

>>> key = random.PRNGKey(1)
>>> X, Omega = sparse_normal_representations(key, 6, 2, 3)
>>> print(X.shape)
(6, 3)
>>> print(Omega)
[1 5]
>>> print(X)
[[ 0.          0.          0.        ]
[ 0.07545021 -1.0032069  -1.1431499 ]
[ 0.          0.          0.        ]
[ 0.          0.          0.        ]
[ 0.          0.          0.        ]
[-0.14357079  0.59042295 -1.43841705]]