bayespy.inference.vmp.transformations.RotationOptimizer

class bayespy.inference.vmp.transformations.RotationOptimizer(block1, block2, D)[source]

Optimizer for rotation parameter expansion in state-space models

Rotates one model block with \mathbf{R} and one model block with \mathbf{R}^{-1}.

Parameters:
  • block1 (rotator object) – The first rotation parameter expansion object

  • block2 (rotator object) – The second rotation parameter expansion object

  • D (int) – Dimensionality of the latent space

References

[7], [6]

__init__(block1, block2, D)[source]

Methods

__init__(block1, block2, D)

rotate([maxiter, check_gradient, verbose, ...])

Optimize the rotation of two separate model blocks jointly.