Skip to content

ryanncode/nf-sketches

Repository files navigation

NF Sketches: High-Throughput Combinatory Execution Engine

A proof-theoretic engine utilizing Lean 4 for $O(V+E)$ topological bounds checking to enable acyclic combinatory graph reduction.


NF Sketches provides a bifurcated architecture designed to natively process saturated states, highly entangled sets, and self-referential graph evaluations. By utilizing Lean 4 for $O(V+E)$ topological bounds checking, the engine translates syntax into a Directed Acyclic Graph (DAG). It then algorithmically compiles the verified logic into a flat operational domain of untyped $S$ and $K$ combinators.

This repository functions as the theoretical laboratory—utilizing Lean's trusted kernel to formally prove algorithmic stratification, K-Iteration limits, and topological boundaries. It conducts the rigorous diagnostic work required to secure cryptographic certainty before the concepts are independently forged into the custom bare-metal engine.


Setup and Building

This repository contains multiple interconnected Lean 4 projects.

Quick Build

To execute a fast build of a specific module (e.g. the REPL sandbox), navigate to its directory and run:

lake build

Full Verification Build

Running the NfValidate module performs full verification, compiling the entire library along with all 1000+ line soundness proofs. This takes significantly longer as Lean must mathematically verify every theorem:

lake build NfValidate

Updating Lean and Mathlib

Because these projects rely heavily on mathlib, keeping your compiler and cache perfectly synced is critical to avoid multi-hour source compilations. Execute the provided shell script at the root to automate this update process across all projects:

./update-lean.sh

This script syncs the lean-toolchain, fetches the latest mathlib commit, downloads pre-compiled binaries via lake exe cache get, and runs a full build.


Core Modules

The repository divides into distinct modules, separating core logical constraints from interactive exploration and bare-metal compilation.

1. nf-validate: The Core Stratification Library

An algorithmic syntax checker reducing set theory validation into a geometric shortest-path routing problem. Utilizing a modified Bellman-Ford traversal, it mathematically traps non-terminating logical loops as negative-weight cycles, securing a soundness proof entirely independent of traditional model-theoretic structures.

2. parse-strat: The Interactive ITP Sandbox

A tactile Read-Eval-Print Loop (REPL) environment acting as an Interactive Theorem Prover. It allows users to explore weak stratification constraints using classical Natural Deduction tactics (intro, rewrite, cut). By dynamically running eval, users can instantly parse topological shift and detect boundaries like Extensionality Collisions in real-time.

3. seq-embed: The Formal Diagnostic Tool

A deep embedding of Gentzen's Sequent Calculus in Lean that restricts comprehension rules strictly to formulas verified by nf-validate. The reduceCut engine intentionally forces logical detours through impredicative substitutions (e.g., the Impredicative Singleton or Kuratowski Pair bounds) to mathematically witness normalization breakdowns and graph collapse.

  • Extensionality Collisions: It verifies Marcel Crabbé's observation that strict global typing destroys substitution closure. By upgrading to weak stratification, the engine restores substitution closure but surfaces the Extensionality Collision, mathematically proving that classical identity ($x = y$) requires an infinite oracle and otherwise triggers unbounded topological recursion.

4. untyped-comb: Untyped Combinatory Logic

Constructs a flat, variable-free combinatory execution environment ($S$, $K$, $I$, $U$).

  • Algorithmic T-Weaking: Synthesizes Forster's $T$-operator ($x \mapsto \iota"x$) as dynamic geometric stabilizers based on the NfValidate integer map.
  • Acyclic Flattening: Uses Kosaraju's algorithm to flatten semantic cycles.
  • Topologically-Guided Reduction: Uses Minimum Cycle Mean (MCM) bounds to enforce K-Iteration limits, safely evaluating paradoxical self-reference (like $V \in V$ or the Russell Set) by forcing terminal states without stack exhaustion.

5. CombLib: Native Operational Semantics

This untyped execution environment processes advanced, paradox-heavy operations natively without memory exhaustion or dependency on external well-ordering axioms:

  • Recursive Self-Models ($V \in V$): Uses the Y-combinator to natively evaluate Quine Atoms ($\Omega = {\Omega}$), safely relying on MCM halting bounds to intercept infinite topological recursion.
  • Choice-Free Cardinal Arithmetic: Implements Transfinite Calculus (e.g., $|A| + |B|$ and $\aleph_0$) entirely via pure $\lambda \to S, K$ bracket abstraction, dynamically triggering $T$-functor injections to track disjoint union topologies across structural boundaries.
  • Physical Observables: Extracts Standard Model empirical metrics (Rest Mass, Resonance) directly from measuring the bounded recursion cost of topological particle knots (electron_knot).
  • Holographic Data Indexing: Compiles absolute complement queries ($V \setminus A$) into $O(1)$ exclusion-first search algorithms, isolating unresolvable contradictions across distributed data swarms in strictly linear time.

6. ExPrograms: Native Verification

The repository provides a suite of executable Lean scripts that organically evaluate these bounds in real-time logic loops:

  • The Russell Paradox Evaluator: Geometrically limits $R \in R$.
  • Native Transfinite Arithmetic: Synthesizes $T$-functor mappings natively on $\aleph_0 + \aleph_0$.
  • Stratified Yoneda Embedding: Confirms the categorical self-representation natively bounds to $T(F(V))$ without graph collapse, validating Pseudo-Cartesian closure.
  • Agentic Workflows & Simulation: Evaluates autonomous logic routing via Sandboxed Loops (AgenticReflection), and physically calculates Resonance magnitude collisions via ColliderSimulation to avoid uncomputable infinities.

Contrast Against HoTT

While Homotopy Type Theory (HoTT) escapes the micro-level iterative construction of ZFC sets via generative spatial types, it retains an infinite, ascending tower of universes (U0:U1:U2...) to prevent self-referential paradoxes. The ontology remains fundamentally layered, introducing heavy computational friction when calculating generative events.

Syntactic Monism provides a truly flat alternative. The topology of the NF universe is singular and natively accommodates a Universal Set containing itself. The system averts logical collapse by shifting the burden of consistency entirely away from the ontological structure onto the syntax via stratification constraints. By regulating the grammar of the query rather than compartmentalizing the fabric of the data itself, syntactic monism allows a logic engine to physically process reality as a unified, cohesive network without artificial hierarchical scaffolding.


License

This project is licensed under the GNU Affero General Public License v3 (AGPLv3) - see the LICENSE file for details.

About

A high-throughput combinatory logic engine, utilizing topological routing to execute non-well-founded algorithms natively on bare-metal hardware.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors