—
Shared normalized rank u; exploratory correspondence.
The wave is a monochromatic scalar simulation, with sources 1 metre behind the observation plane. U(x,y) = Σ A exp(i(2πr/λ + phase))/r. It is not a full vector electromagnetic solver.
Auto-solve estimates source C position, amplitude and phase from 210 noisy complex field samples, with A, B and wavelength known. A grid search and local refinement minimize least-squares error; 70 separate samples provide a held-out check. The search uses observations, never the true C parameters. It does not guarantee the global optimum. Noise is added to each field component; the slider is its absolute standard deviation in relative amplitude units.
Prime data: two verified blocks of 77,777 primes. p = 30q + r, for r = 1, 7, 11, 13, 17, 19, 23, 29. Exceptions 2, 3, 5 remain in the source. Eight degree-1 splines are fitted by least squares; every fifth rank is withheld. Residual units are q units. These are fits to observed primes, not predictions of untested primes.
The all-pairs view resamples every signal onto 256 equally spaced normalized positions u. Prime ranks, a wave cross-section and an audio time window do not share a physical axis. Correlation here is an exploratory visual comparison, not evidence of causation. “Best XY” removes a linear trend, standardizes both signals and fits their axes; it is a defined view, not a universal optimality claim.
The sphere, plane and helix are display mappings of the selected signal. Their displacement, color and camera controls change the rendering, not the underlying evidence. Frequency bands are not isolated U2 members or stems.
Buttons export a receipt containing current parameters, data hashes, residuals and solver status. Original wave and spline tools and the exact prime decoder are included under reference/.