DPSDataFidelity#

class deepinv.sampling.DPSDataFidelity(denoiser=None)[source]#

Bases: NoisyDataFidelity

The DPS data-fidelity term.

This corresponds to the $p(y|x)$ prior as proposed in Diffusion Probabilistic Models.

Parameters:

denoiser (deepinv.models.Denoiser) – Denoiser network.

$${\mathrm{\nabla }}_x\log p(y|x)={(\mathrm{Id}+{\mathrm{\nabla }}_{x}D(x))}^{\mathrm{\top }}{A}^{\mathrm{\top }}(y-A\left(D(x)\right))$$

Note

The preconditioning term is computed with automatic differentiation.

Parameters:

denoiser (deepinv.models.Denoiser) – Denoiser network

forward(x, y, physics, sigma, clip=False)[source]#

Returns the loss term $d(A\left({\mathrm{D}}_x\left(\sigma \right)\right),y)$.

Parameters:

• x (torch.Tensor) – input image

• y (torch.Tensor) – measurements

• physics (deepinv.physics.Physics) – forward operator

• sigma (float) – standard deviation of the noise.

• clip (bool) – whether to clip the output of the denoiser to the range [-1, 1].

Returns:

(torch.Tensor) loss term.

Return type:

Tensor

grad(x, y, physics, sigma, *args, **kwargs)[source]#

Parameters:

• x (torch.Tensor) – Current iterate.

• y (torch.Tensor) – Input data.

• physics (deepinv.physics.Physics) – physics model

• sigma (float) – Standard deviation of the noise.

Returns:

(torch.Tensor) score term.

Return type:

Tensor

Examples using DPSDataFidelity:#

Posterior Sampling for Inverse Problems with Stochastic Differential Equations modeling.

Posterior Sampling for Inverse Problems with Stochastic Differential Equations modeling.