CompositeOp.py 🚧

Source: LION/operators/CompositeOp.py

Warning

This file has not yet received a complete narrative and docstring audit. Its public source-level API is listed automatically below.

Composite linear operator A = Phi Psi^{-1} and its adjoint for compressed sensing.

class CompositeOp(wavelet: Operator, phi: PhotocurrentMapOp, device: str | torch.device | None = None)

Composite linear operator \(A = \Phi \Psi^{-1}\) and its adjoint.

Parameters ———- wavelet : Wavelet2D Wavelet transform object. phi : PhotocurrentMapOp Photocurrent mapping operator. device : str or torch.device, optional Device for computations. If None, uses wavelet.device.

References ———- .. [Koutsourakis2021] G. Koutsourakis, A. Thompson, and J. C. Blakesley, “Toward Megapixel Resolution Compressed Sensing Current Mapping of Photovoltaic Devices Using Digital Light Processing”, Solar RRL, 5(11):2100467, 2021. doi:10.1002/solr.202100467

CompositeOp.forward(self, w: torch.Tensor) → torch.Tensor

Apply the forward projection A = Phi Psi^{-1}.

Note

Prefer calling the instance of the CompositeOp operator as operator(w) over directly calling this method. See this PyTorch discussion.

CompositeOp.adjoint(self, r: torch.Tensor) → torch.Tensor

Apply A^T = Psi Phi^T.

Parameters ———- r : torch.Tensor Residual in measurement space, shape (M,).

Returns ——- g : torch.Tensor Gradient in wavelet coefficient space, shape (Nw,).

CompositeOp.domain_shape(self) → tuple[int, ...]

Return the shape of the wavelet coefficient domain.

CompositeOp.range_shape(self) → tuple[int, ...]

Return the shape of the measurement range.