Goals as Constraints: Writing miniKanren Constraints in miniKanren

on YouTube ↗45m

" summary

Adds an extension to miniKanren that runs arbitrary miniKanren goals as constraints, implemented as two cooperating interpreters that treat constraints as first‑order goal objects. Introduces three forms: constraint, P-constraint, and natto (negation), reuses miniKanren for constraint code, and generalizes disequality handling to arbitrary goals. Prototype in Scheme exists on GitHub; microbenchmarks vs fasterMK show consistent speedups on disequality-heavy tests and mixed results for absent/present constraints, and the system successfully expresses presento/absento and finite‑domain style checks.

AI-written summary · about

" chapters (22)

1. 0:00miniKanren constraints intro
2. 1:04Talk overview and motivation
3. 2:07Limitations of traditional constraints
4. 3:31Goals as constraints in miniKanren
5. 4:56Constraint forms in miniKanren
6. 7:23Recursive structures and matching
7. 8:26Negation in constraints
8. 10:11Primitive constraints with P
9. 11:58Type constraints implementation
10. 12:40Natto: Negation of goals
11. 14:25Q&A on constraint interface
12. 17:16Constraint solver overview
13. 19:02Unification and state management
14. 25:38Q&A on interpreter structure
15. 26:02Natto and disequality generalization
16. 31:00Disjunction handling in constraints
17. 35:56Attributed variables and dependencies
18. 37:37Fresh variables and matching
19. 38:19Performance benchmarks
20. 41:53Performance results analysis
21. 42:13GitHub repo and future work
22. 42:55Q&A on satisfiability