Category-theoretic graph structures in Rust, anchored to the source papers: a strict implementation of Fong & Spivak, Hypergraph Categories (2019), with applied-CT, magnitude, Wolfram-physics, and categorical-deep-learning extensions.
Status: the five proven crates (core / applied / magnitude / physics / dl) have landed on a thin DeepCausality algebraic substrate (
deep_causality_num/deep_causality_haft), withnalgebrakept optional and numeric-only. Versioning is workspace-wide (tags v0.1.0 → v0.3.0) and work is tracked as GitHub issues. Phase 6 (catgraph-syntax, the Arrow presentation frontend, #5): the S1–S5 milestone surface is complete (printer, parser + presentation files, interpreter, Frobenius layer, Traced typed builder); post-milestone follow-ups are tracked on #5 (#79/#80/#81).
| Crate | Paper anchor |
|---|---|
catgraph |
Fong & Spivak 2019 — Hypergraph Categories |
catgraph-applied |
Fong & Spivak 2018 — Seven Sketches in Compositionality |
catgraph-magnitude |
Bradley–Vigneaux 2025; Leinster 2008/2013/2017 |
catgraph-physics |
Wolfram-physics extensions (DPO rewriting, multiway, branchial) |
catgraph-dl |
Gavranović et al., ICML 2024 — Categorical Deep Learning |
catgraph-syntax |
F&S 2018 Ch. 5 (props/presentations) + F&S 2019 (Frobenius layer); term language over catgraph-applied's NF engine |
cargo build --workspace
cargo test --workspace
cargo clippy --workspace --all-targets -- -D warnings # the CI gate (default lints)
cargo clippy --workspace --all-targets -- -W clippy::pedantic # advisory local pass (non-gating)MIT — see LICENSE.