The Golden Line: Compiling Constant-in-n Retrieval via Whitening-Based Resonant Fields in Internal Time τ (Paper 6 of 7 in the τ-Epoch Series)

This preprint (Paper 6 of 7 in the τ-Epoch series) presents a compiled retrieval mechanism in which dataset-size dependence is moved to a one-time compilation step, while the query phase is executed as a single projection y=qM with runtime independent of dataset size n for fixed dimensions. The method compiles a fixed-size field M using inverse-square-root whitening W=A^{-1/2} and reports reproducible scaling evidence, speedups versus scan baselines, whitening diagnostics \|W^\top A W-I\|_F, and conditioning measurements. The scope is intentionally narrow and does not claim equivalence to approximate nearest neighbor systems or exact softmax attention.