Part 3 of a series on the Fourier transform: the time-frequency tradeoff, multiresolution analysis, and training a minimal wavelet neural operator on Burgers' equation in JAX.

Part 2 of a series on the Fourier transform: the discrete transform as a matrix, the Cooley–Tukey trick, aliasing, spectrograms of actual music, and spectral methods for PDEs.

Part 1 of a series on the Fourier transform: where it came from, square waves as a sum smooth sinusoidal waves, and Epicycles.

Inside a synthetic-data prototype that scores public-finance payments for fraud risk: a JAX logistic regression classifier, an anomaly detector, and a graph-diffusion 'guilt by association' feature -- with the spectral math and the plots that come out of it.