azula.guidance.jfps

Jacobian-Free Posterior Sampling (JFPS) internals.

Classes

JFPSDenoiser

Creates a JFPS denoiser module.

Descriptions

class azula.guidance.jfps.JFPSDenoiser(denoiser, y, A, cov_y, cov_x, solver='cg', iterations=1)[source]

Creates a JFPS denoiser module.

Parameters:

  • denoiser (Denoiser) – A denoiser \(q_\phi(X \mid X_t)\).

  • y (Tensor) – An observation \(y \sim \mathcal{N}(A(x), \Sigma_y)\), with shape \((*, D)\).

  • A (Callable[[Tensor], Tensor]) – The forward operator \(x \mapsto A(x)\).

  • cov_y (Covariance) – The noise covariance \(\Sigma_y\).

  • cov_x (Covariance) – The signal covariance \(\Sigma_x\).

  • solver (Literal['cg', 'gmres']) – The linear solver name.

  • iterations (int) – The number of solver iterations.