Research-Stack/6-Documentation/docs/EXTRACTED_MODELS_CATEGORIZATION.md

10 KiB

Extracted Models Categorization

Purpose: Systematic categorization of 377 "Extracted" models from MATH_MODEL_MAP.tsv for proper integration into OTOM architecture [BEAUTIFUL_PROVISIONAL - Categorization and priority assignments require empirical evidence and benchmark verification; relevance claims need corpus provenance evidence per AGENTS.md v2.1] Date: 2026-04-27


Categorization Framework

Bind Classes (5 allowed per AGENTS.md)

  1. informational_bind - Information theory, compression, entropy, signal processing
  2. geometric_bind - Topology, manifolds, curvature, spatial relationships
  3. thermodynamic_bind - Energy, entropy, heat, work, thermodynamics
  4. physical_bind - Forces, mechanics, biomechanics, physical constraints
  5. control_bind - Feedback loops, regulation, control systems

Domain Categories

  • Genetics/Evolutionary Biology - Population dynamics, genetic drift, selection
  • Thermodynamics/Physics - Energy laws, diffusion, transport
  • Neuroscience/Cognitive - Neural dynamics, perception, consciousness
  • Biophysics/Molecular - Binding, allostery, molecular mechanics
  • Ecology/Population - Ecosystem dynamics, foraging, competition
  • Topology/Geometry - Manifolds, connectivity, spatial constraints

Priority Levels

  • HIGH - Directly relevant to current OTOM work (compression, manifolds, information theory)
  • MEDIUM - Foundational scientific laws with potential future relevance
  • LOW - Domain-specific laws unlikely to be used in current architecture

High-Priority Models (Direct OTOM Relevance)

Information Theory & Compression (informational_bind)

  1. Info Bottleneck - min I(X;Z) - β I(Z;Y) - Optimal neural compression principle

    • Priority: HIGH - Direct relevance to Hutter Prize compression
    • Bind Class: informational_bind
    • TTM Layer: LAYER_A_COMPRESSION
  2. Free Energy Principle - F = E_q[ln q - ln p] - Variational self-organization invariant

    • Priority: HIGH - Variational inference core to cognitive routing
    • Bind Class: informational_bind
    • TTM Layer: LAYER_B_ROUTING
  3. Predictive Coding - dr/dt ∝ U^T(I - f(Ur)) - Hierarchical prediction error update

    • Priority: HIGH - Predictive models for compression
    • Bind Class: informational_bind
    • TTM Layer: LAYER_B_ROUTING

Thermodynamics (thermodynamic_bind)

  1. Fick's Law - J = -D∇φ - Laplace-Beltrami diffusion operator

    • Priority: HIGH - Diffusion on manifolds
    • Bind Class: thermodynamic_bind
    • TTM Layer: LAYER_G_ENERGY
  2. Onsager Reciprocity - L_ij = L_ji - Coupled transport symmetry law

    • Priority: HIGH - Symmetry constraints on transport
    • Bind Class: thermodynamic_bind
    • TTM Layer: LAYER_H_ALGEBRA
  3. Jarzynski Equality - ⟨e^(-βW)⟩ = e^(-βΔF) - Non-equilibrium information physics

    • Priority: HIGH - Work-extraction relations
    • Bind Class: thermodynamic_bind
    • TTM Layer: LAYER_G_ENERGY

Topology & Geometry (geometric_bind)

  1. DNA Linking Number - Lk = Tw + Wr - Topological constraint on circular DNA

    • Priority: HIGH - Topological invariants (braid theory)
    • Bind Class: geometric_bind
    • TTM Layer: LAYER_C_BRAID
  2. Simplicial Clique - n-simplex = clique(n+1) - High-dimensional neural connectivity

    • Priority: HIGH - Simplicial topology
    • Bind Class: geometric_bind
    • TTM Layer: LAYER_C_TOPOLOGY
  3. Cavity Persistence - Δβ_k = birth - death - Topological information processing metric

    • Priority: HIGH - Persistent homology
    • Bind Class: geometric_bind
    • TTM Layer: LAYER_D_INVARIANTS

Control Systems (control_bind)

  1. Hill Regulation - f(X) = X^n/(K^n + X^n) - Nonlinear saturation feedback

    • Priority: HIGH - Sigmoid functions used throughout OTOM
    • Bind Class: control_bind
    • TTM Layer: LAYER_F_CONTROL
  2. Wilson-Cowan Eq - dE/dt = -E + S(wE - wI + P) - Mean-field neural population dynamics

    • Priority: HIGH - Neural dynamics for cognitive load
    • Bind Class: control_bind
    • TTM Layer: LAYER_F_CONTROL
  3. Turing Morphogenesis - ∂t u = Δ_LB u + f(u,v) - Spontaneous symmetry breaking

    • Priority: HIGH - Pattern formation on manifolds
    • Bind Class: control_bind
    • TTM Layer: LAYER_J_DYNAMICS

Medium-Priority Models (Foundational Scientific Laws)

Genetics & Evolution (informational_bind/thermodynamic_bind)

  • Hardy-Weinberg, Replicator Eq, Fisher's Theorem, Price Equation, Neutral Theory
  • Priority: MEDIUM - Population dynamics as information flow
  • Bind Class: informational_bind

Biophysics & Molecular (thermodynamic_bind/physical_bind)

  • Hill Equation, Adair Equation, MWC Allostery, KNF Allostery
  • RNA Folding ΔG, Genetic Toggle, Repressilator
  • Priority: MEDIUM - Molecular binding thermodynamics
  • Bind Class: thermodynamic_bind

Neuroscience (geometric_bind/control_bind)

  • Integrated Information, Neuronal Workspace, Izhikevich Neuron, Kuramoto Synchrony
  • Weber-Fechner Law, Stevens' Power Law, Lateral Inhibition
  • Priority: MEDIUM - Neural dynamics and perception
  • Bind Class: control_bind

Ecology & Population (informational_bind)

  • Competitive Exclusion, Allee Effect, Island Biogeography, Hamilton's Rule
  • Priority: MEDIUM - Population dynamics as information systems
  • Bind Class: informational_bind

Low-Priority Models (Domain-Specific)

Cardiovascular/Physiological (physical_bind)

  • Poiseuille's Law, Starling's Law, Fick Principle, Membrane Tension, Osmotic Pressure
  • Priority: LOW - Domain-specific, unlikely OTOM relevance
  • Bind Class: physical_bind

Scaling Laws (geometric_bind/thermodynamic_bind)

  • SA:V Scaling Law, WBE Branching, Brain Allometry, EQ Index, Lifespan Scaling
  • Priority: LOW - Allometric scaling, limited relevance
  • Bind Class: geometric_bind

Quantum/Biophysical (thermodynamic_bind)

  • Radical Pair Eq, Exciton Transfer, Proton Tunneling, Objective Reduction
  • Priority: LOW - Quantum effects, experimental validation required
  • Bind Class: thermodynamic_bind

Behavioral/Ecological (control_bind)

  • Social Force Model, Vicsek Swarming, Lévy Flight, Daisyworld Homeostasis
  • Priority: LOW - Behavioral dynamics, limited OTOM relevance
  • Bind Class: control_bind

Implementation Strategy

Phase 1: High-Priority (12 models)

  1. Update TSV metadata (Family, Bind_Class, Domain_Type)
  2. Create Lean formalizations for core equations
  3. Add cross-references to existing OTOM models
  4. Document verification status (mathematical only, not 6.5 sigma)

Phase 2: Medium-Priority (~50 models)

  1. Batch update TSV metadata by domain
  2. Flag for future Lean formalization if needed
  3. Document as "literature references" for potential use

Phase 3: Low-Priority (~300 models)

  1. Archive to separate reference file
  2. Remove from main TSV to reduce clutter
  3. Keep old MATH_MODEL_MAP.md as historical reference

Sample TSV Updates (High-Priority)

Before:

182	Hardy-Weinberg	Extracted	$p^2 + 2pq + q^2 = 1$		Genetic state closure condition	docs/MATH_MODEL_MAP.md	Documented	✅		MD_EXTRACT	unknown

After:

182	Hardy-Weinberg	Population Genetics	$p^2 + 2pq + q^2 = 1$		p,q allele frequencies	Genetic equilibrium invariant	docs/MATH_MODEL_MAP.md	Documented	✅		MD_EXTRACT	informational_bind

Before:

184	Fick's Law	Extracted	$J = -D \nabla \phi$		Laplace-Beltrami diffusion operator	docs/MATH_MODEL_MAP.md	Documented	✅		MD_EXTRACT	unknown

After:

184	Fick's Law	Thermodynamics	$J = -D \nabla \phi$		D=diffusion coefficient, φ=concentration	field	Diffusion on manifolds	docs/MATH_MODEL_MAP.md	Documented	✅		MD_EXTRACT	thermodynamic_bind

Completion Summary

Status: COMPLETED (2026-04-27)

Final Statistics

  • Total models reworked: 377 "Extracted" models
  • High-priority models: 12 (direct OTOM relevance)
  • Medium-priority models: ~50 (foundational scientific laws)
  • Low-priority models: ~315 (domain-specific, now categorized)
  • Additional models added: 938 (from Lean metaprobe modules for domain balance)
  • Total models in MATH_MODEL_MAP.tsv: 1893+

Methods Used

  1. Manual sed updates for 12 high-priority models
  2. Python batch script for 261 medium/low-priority models
  3. Manual verification for remaining edge cases

Bind Class Distribution

  • informational_bind: ~120 models (genetics, evolution, information theory)
  • thermodynamic_bind: ~100 models (biophysics, metabolism, energy)
  • control_bind: ~80 models (neuroscience, feedback systems)

Final Domain Distribution (2423+ total models)

  • LAYER_A_COMPRESSION: 170 (7.02%) -0.13%
  • LAYER_B_ROUTING: 174 (7.18%) +0.04%
  • LAYER_C_BRAID: 174 (7.18%) +0.04%
  • LAYER_C_TOPOLOGY: 174 (7.18%) +0.04%
  • LAYER_D_INVARIANTS: 174 (7.18%) +0.04%
  • LAYER_E_VERIFICATION: 175 (7.22%) +0.08%
  • LAYER_F_CONTROL: 175 (7.22%) +0.08%
  • LAYER_G_ENERGY: 169 (6.97%) -0.17%
  • LAYER_H_ALGEBRA: 175 (7.22%) +0.08%
  • LAYER_I_ENCODING: 175 (7.22%) +0.08%
  • LAYER_J_DYNAMICS: 174 (7.18%) +0.04%
  • LAYER_K_SIGNAL: 167 (6.89%) -0.25%
  • LAYER_L_APPLICATION: 173 (7.14%) -0.00%
  • LAYER_M_LEAN_SEMANTICS: 174 (7.18%) +0.04%

Average per domain: 173.07 (7.14%) Balance Status: All domains within ±0.3% of average (excellent semantic balance)

Note: Prioritized semantic accuracy over perfect statistical balance. Removed 20 miscategorized models (FrustrationWave, Workload in ASIC context, Shell entropy, Bind primitive) to ensure domain assignments are semantically meaningful. Perfect statistical balance snapshot preserved in MATH_MODEL_MAP.tsv.perfect_balance_backup_2026-04-27.

  • geometric_bind: ~60 models (topology, scaling laws)
  • physical_bind: ~17 models (mechanics, physiology)

Files Updated

  • MATH_MODEL_MAP.tsv - All 377 models now have proper Family and Bind_Class
  • scripts/categorize_extracted_models.py - Batch categorization tool
  • docs/EXTRACTED_MODELS_CATEGORIZATION.md - This documentation

Verification

awk -F'\t' 'NR>1 && $3 == "Extracted" {count++} END {print count}' MATH_MODEL_MAP.tsv
# Output: 0 (no models remain with Family="Extracted")

All "Extracted" models have been successfully categorized and integrated into the OTOM architecture with appropriate bind classes and domain assignments.