7. Finite Fourier And Circulant Algebra
Claim boundary proved finite algebra Python diagnostics AI performance separate no quality claim
Finite Fourier analysis is the spectral language of a finite circle. A cyclic character turns addition on ZMod n into multiplication by a phase. Circular convolution is the corresponding circulant token mixer.
The Lean spine is:
Circle.Core.FiniteFourierCircle.Applications.CirculantCircle.Applications.CirculantSpectral
The public imports are:
import Circle.Core
import Circle.Applications.Public
Proved Facts
The current formal layer proves:
- character values are
nth roots of unity; - finite shifts compose additively;
- Fourier coefficients are additive in the signal;
- shifting a signal multiplies each Fourier coefficient by the matching phase;
- circular convolution is shift-equivariant, commutative, and additive;
- each Fourier coefficient of
c ⋆ xfactors into the matching coefficient ofctimes the matching coefficient ofx.
Executable Companion
from circle_math.core import (
circular_convolution,
spectral_aliasing_report,
spectral_convolution_report,
)
signal = [1, 2, 0, -1]
kernel = [2, 0, 1, 0]
print(circular_convolution(kernel, signal))
print(spectral_convolution_report(kernel, signal).passed)
print(spectral_aliasing_report(4, [-1, 0, 3, 4, 7]))Why It Connects
The same circle algebra shows up in:
- finite coils and stride orbits;
- DFT and roots of unity;
- circular convolution and circulant matrices;
- RoPE and other phase-bank positional encodings;
- Fourier features and harmonic networks;
- cyclic sparse-attention patterns.
The proof layer certifies algebraic structure. It does not certify neural model quality, speed, memory use, training stability, or deployment behavior.