acsets-relational-thinking（关系思维）

SKILL: ACSets Relational Thinking

Version: 2.0.0 Trit: 0 (ERGODIC) Domain: database, category-theory, rewriting Source: Topos Institute RelationalThinking Course + AlgebraicJulia

Overview

ACSets (Attributed C-Sets) are functors X: C → Set where C is a small category (schema). This skill integrates:

1. RelationalThinking Course - Topos Institute's pedagogical approach
2. DPO Rewriting - Double Pushout graph transformation
3. Self-Play Loop - Query → Execute → Evaluate → Refine
4. GF(3) Conservation - Triadic skill composition

Core Concept: C-Set as Functor

Schema (Category C)          Instance (Functor X: C → Set)
┌─────────────────┐          ┌─────────────────────────────┐
│  E ──src──→ V   │    X     │  X(E) = {e1, e2, e3}        │
│    ←──tgt──     │   ───→   │  X(V) = {v1, v2}            │
└─────────────────┘          │  X(src): e1↦v1, e2↦v1, e3↦v2│
                             │  X(tgt): e1↦v2, e2↦v2, e3↦v1│
                             └─────────────────────────────┘

Schema Definition

Basic Graph Schema

using Catlab.CategoricalAlgebra

@present SchGraph(FreeSchema) begin
  V::Ob                    # Vertices (object)
  E::Ob                    # Edges (object)
  src::Hom(E, V)           # Source morphism
  tgt::Hom(E, V)           # Target morphism
end

@acset_type Graph(SchGraph, index=[:src, :tgt])

Entity-Subtype Hierarchy (from RelationalThinking Ch8)

@present SchKitchen(FreeSchema) begin
  Entity::Ob

  Food::Ob
  food_in_on::Hom(Food, Entity)
  food_is_entity::Hom(Food, Entity)

  Kitchenware::Ob
  ware_in_on::Hom(Kitchenware, Entity)
  ware_is_entity::Hom(Kitchenware, Entity)

  BreadLoaf::Ob
  bread_loaf_is_food::Hom(BreadLoaf, Food)
  Knife::Ob
  knife_is_ware::Hom(Knife, Kitchenware)
end

@acset_type Kitchen(SchKitchen)

DPO Rewriting (Double Pushout)

The Pattern: L ← K → R

     L ←──l── K ──r──→ R
     │        │        │
match│        │        │
     ↓        ↓        ↓
     G ←───── D ──────→ H
       pushout      pushout
       complement

• L = Find (what to match)
• K = Keep (overlap preserved)
• R = Replace (new structure)
• G = Host graph
• H = Result graph

Rule Definition

using AlgebraicRewriting

# Slice bread rule: adds BreadSlice when knife + loaf present
slice_bread = @migration(SchKitchen, begin
  L => @join begin
    loaf::BreadLoaf
    knife::Knife
  end
  R => @join begin
    loaf::BreadLoaf
    slice::BreadSlice
    food_in_on(bread_slice_is_food(slice)) == food_in_on(bread_loaf_is_food(loaf))
    knife::Knife
  end
  K => @join begin
    loaf::BreadLoaf
    knife::Knife
  end
end)

rule = make_rule(slice_bread, yKitchen)

Apply Rewrite

matches = get_matches(rule, state)
new_state = rewrite_match(rule, matches[1])

Self-Play Loop

The ACSet self-refinement monad:

Query₀ → Execute → Evaluate → Mine Patterns → Refine → Query₁ → ...

Implementation

struct SelfRefinementLoop
  schema::Presentation
  state::ACSet
  patterns::Vector{Pattern}
  generation::Int
end

function step!(loop::SelfRefinementLoop, rule::Rule)
  # Find matches
  matches = get_matches(rule, loop.state)

  # Evaluate each match
  evaluations = [evaluate_match(m, loop.patterns) for m in matches]

  # Mine new patterns from successful evaluations
  new_patterns = mine_patterns(evaluations)
  append!(loop.patterns, new_patterns)

  # Apply best match
  best = argmax(e -> e.score, evaluations)
  loop.state = rewrite_match(rule, matches[best.index])
  loop.generation += 1

  loop
end

Convergence Criterion

function converged(loop::SelfRefinementLoop; threshold=0.95)
  recent = loop.patterns[end-10:end]
  stability = std([p.score for p in recent])
  stability < (1 - threshold)
end

GF(3) Integration

Trit Assignment

function acset_to_trits(g::Graph, seed::UInt64)
  rng = SplitMix64(seed)
  trits = Int[]
  for e in parts(g, :E)
    h = next_u64!(rng)
    hue = (h >> 16 & 0xffff) / 65535.0 * 360
    trit = hue < 60 || hue >= 300 ? 1 :
           hue < 180 ? 0 : -1
    push!(trits, trit)
  end
  trits
end

# Conservation check
gf3_conserved(trits) = sum(trits) % 3 == 0

Synergistic Triads Containing ACSets

clj-kondo-3color (-1) ⊗ acsets (0) ⊗ rama-gay-clojure (+1) = 0 ✓
three-match (-1) ⊗ acsets (0) ⊗ gay-mcp (+1) = 0 ✓
slime-lisp (-1) ⊗ acsets (0) ⊗ cider-clojure (+1) = 0 ✓
hatchery-papers (-1) ⊗ acsets (0) ⊗ frontend-design (+1) = 0 ✓

Orthogonal Bundles

ACSets participates in the STRUCTURAL bundle:

Visual Conventions (from RelationalThinking)

Commands

# Schema operations
just acset-schema FILE       # Display schema diagram
just acset-instance FILE     # Display instance elements
just acset-morphism F G      # Show homomorphism F → G

# DPO rewriting
just acset-rule RULE STATE   # Apply rewrite rule
just acset-matches RULE STATE # Find all matches
just acset-chain RULES STATE # Chain multiple rewrites

# Self-play
just acset-selfplay SCHEMA   # Run self-refinement loop
just acset-patterns STATE    # Mine patterns from state
just acset-converge SCHEMA   # Run until convergence

# GF(3) operations
just acset-trits STATE SEED  # Color state with seed
just acset-gf3 STATE         # Check GF(3) conservation
just acset-triads            # Show synergistic triads

References

Topos Institute

• RelationalThinking Book
• RelationalThinking Code

AlgebraicJulia

• Catlab.jl
• ACSets.jl
• AlgebraicRewriting.jl

Papers

• Patterson et al. "Categorical data structures for technical computing" (Compositionality 2022)
• Aguinaldo et al. "Categorical Representation Language for Knowledge-Based Planning" (AAAI 2023)

Related Skills

• rama-gay-clojure (+1) - Scalable backends with color tracing
• clj-kondo-3color (-1) - Schema validation/linting
• glass-bead-game (0) - World hopping across schemas
• unworld (0) - Derivational chains for state evolution
• discohy-streams (0) - DisCoPy categorical color streams

Skill Name: acsets-relational-thinking Type: Category-Theoretic Database / DPO Rewriting Trit: 0 (ERGODIC) GF(3): Conserved via triadic composition