Skip to content

af271/ic-theory

 
 

Repository files navigation

IcTheory

Lean 4 + Mathlib formalization workspace for A theory of incremental compression.

Status

The development has moved past the initial foundation layer.

Current compiled coverage includes:

  • foundations for finite bitstrings, pairing/prefix encodings, internal encodings, and explicit logarithmic-slack relations
  • a computability and complexity substrate built around a bitstring-facing universal machine, conditional complexity, prefix complexity, prefix information, finite prefix descriptions, and symmetry-of-information tooling
  • compression definitions and Section 2 results for features, descriptive maps, singleton features, and the universal-feature branch
  • Section 3.1 and 3.2 results, including the conditional-complexity characterizations of shortest features/maps, the prefix upper bounds from Lemma 3.2, and the conditional bridge K(x | r) ≤ C(x | r) + O(log C(x | r))
  • Section 3.3 machinery, including a paper-facing Lemma 3.3
  • a packaged Theorem 3.2 by cases
  • a Theorem 3.3 reduction layer for equation (16), with the residual-side prefix bridge proved concretely and the remaining blocker isolated to the shortest-feature versus prefix-shortest gap

The default library target builds with lake build.

Module Guide

  • IcTheory.Basics: bitstrings, finite enumerators, prefix/pair encodings, internal encodings, and logarithmic bound infrastructure
  • IcTheory.Computability: the universal machine wrapper and the complexity/information layer
  • IcTheory.Compression: feature theory, Section 2, Section 3.1 to 3.3 developments, and the current Theorem 3.2 / Theorem 3.3 packaging
  • IcTheory.Sanity: consistency checks and small integration lemmas

Build

lake build

Project Notes

  • PLAN.md
  • docs/model-decisions.md
  • docs/theorem-map.md

About

Lean formalization of the Theory of Incremental Compression

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages

  • Lean 100.0%