mr2.algorithms.varimax

mr2.algorithms.varimax(phi: Tensor, gamma: float = 1.0, n_iterations: int = 20) → Tensor[source]

Apply an orthogonal Varimax rotation.

Parameters:

• phi (Tensor) – Input matrix of shape (*batch, n_components, n_channels). Rotation is performed across the n_components dimension.

• gamma (float, default: 1.0) – Power parameter for the orthomax criterion. 1.0 gives Varimax.

• n_iterations (int, default: 20) – Number of rotation iterations.

Returns:

Rotated matrix with the same shape as phi.