LRFreeFormCov

class glimix_core.cov.LRFreeFormCov(n, m)

General semi-definite positive matrix of low rank, K = LLᵀ.

The covariance matrix K is given by LLᵀ, where L is a n×m matrix and n≥m. Therefore, K will have rank(K) ≤ m.

Example

>>> from glimix_core.cov import LRFreeFormCov
>>> cov = LRFreeFormCov(3, 2)
>>> print(cov.L)
[[1. 1.]
 [1. 1.]
 [1. 1.]]
>>> cov.L = [[1, 2], [0, 3], [1, 3]]
>>> print(cov.L)
[[1. 2.]
 [0. 3.]
 [1. 3.]]
>>> cov.name = "F"
>>> print(cov)
LRFreeFormCov(n=3, m=2): F
  L: [[1. 2.]
      [0. 3.]
      [1. 3.]]

__init__(n, m)

Constructor.

Parameters

• n (int) – Covariance dimension.

• m (int) – Upper limit of the covariance matrix rank.

Methods

__init__(n, m)	Constructor.
fix()	Disable parameter optimisation.
gradient()	Derivative of the covariance matrix over the lower triangular, flat part of L.
listen(func)	Listen to parameters change.
unfix()	Enable parameter optimisation.
value()	Covariance matrix.

Attributes

L	Matrix L from K = LLᵀ.
Lu	Lower-triangular, flat part of L.
name	Name of this function.
nparams	Number of parameters.
shape	Array shape.