pdft.coherence.dense_operator#
- pdft.coherence.dense_operator(basis)[source]#
The basis as an explicit matrix acting on
vec(image).Column j is the transform of the image that is 1 at pixel j and 0 elsewhere, so the result is (2^m 2^n) x (2^m 2^n).
This is a diagnostic, not a code path: it costs N^2 transforms of an N-pixel image and is meant for small m, n. Nothing in training needs it —
certify_flat_modulusgives the parameter-space guarantee without forming it.- Return type: