MesoNet: convolution and linear-algebra reading map

convolution / SVD / eigenvalues reading pile. keep shapes + indexing next to the intuition, otherwise i lose track of what the matrix is even acting on

February 1–3 reference cluster: convolution, finite geometric series, Kronecker delta, conjugate transpose, row echelon form, SVD, orthogonal matrices, eigenvalues and gradients. Most of the theorem-development content in this section is embedded figures; text extraction mainly preserves the reading list.

Saved explanation: intuitive convolution. Use it alongside the model's actual indexing convention rather than assuming every convolution diagram matches the implementation.

Reference questions: what are the operator's input/output dimensions, boundary convention and transpose? Which expression is a convolution, which is matrix multiplication, and which is an elementwise product? Recover the original figures before attaching a derivation or theorem claim to this date.

Tangent to retain: the following February reading cluster includes pseudoinverses, deconvolution and Hadamard products; these can become a separate undated math-reference card when the image notes are transcribed.

Convolutional spiking architecture