Skip to content
back to the quest log
Fourier Sparsity Testing repository with exact definitions and theorem scope
rare∎ researchshipped

Fourier Sparsity Testing

Exact finite testing, no hidden gap

A Lean 4 development of the exact finite core of Boolean Fourier sparsity testing and a proved nonadaptive exhaustive tester.

role
Formalization author
org
Independent research
when
2026

What it is

The repository formalizes Boolean functions on the finite cube, exact Fourier coefficients and support, Hamming distance, farness from sparse functions, and a deterministic nonadaptive query model.

It proves a perfect tester that queries the whole cube, computes the exact spectrum, and decides sparsity with precisely 2^n queries.

That theorem is intentionally not presented as the sought dimension-independent randomized result. The publication records the unresolved gap as part of the deliverable.

where the effort went

  • Fourier94
  • Proof94
  • Queries90
  • Auditability96

the numbers

Query count
2^n
Proof escapes
0
Coefficients
Exact ℚ
Lean
4.33.1

built with

  • Lean 4
  • Mathlib
  • Lake
  • Fourier analysis

delivery record

What I made

I wanted a checked baseline where every definition and query count is explicit before attempting the much harder randomized sparsity-testing bound.

  • 01

    Exact Fourier coefficients and support on the Boolean cube

  • 02

    Hamming distance and farness from the sparse-function class

  • 03

    Deterministic nonadaptive oracle-query model

  • 04

    Perfect exhaustive tester with an exact 2^n query count

  • 05

    Audited Lean build with the open randomized gap clearly marked

Hard problems

The constraints mattered as much as the finished interface.

01field note

Distinguishing a baseline from the open result

problem
A perfect tester sounds stronger than it is when its query count is exponential in dimension.
response
The theorem states the exact query set and count, while the README separates it from any dimension-independent randomized claim.
02field note

Making Fourier definitions executable

problem
Analytic notation must become finite data that Lean can compute and reason about without approximation.
response
Coefficients, support, distance, and sparsity are expressed as finite rational sums and decidable predicates.

Screens

after shipping

What stayed with me

  • A formal baseline makes the distance between an exact exhaustive argument and a sublinear tester impossible to hand-wave away.

  • Query complexity belongs in the theorem interface, not only in explanatory prose.