https://github.com/RobScheidegger/microsat
A tiny DPLL SAT-solver written in Rust.
https://github.com/RobScheidegger/microsat
Last synced: 4 months ago
JSON representation
A tiny DPLL SAT-solver written in Rust.
- Host: GitHub
- URL: https://github.com/RobScheidegger/microsat
- Owner: RobScheidegger
- License: gpl-3.0
- Created: 2024-11-19T00:05:30.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2024-11-19T02:34:22.000Z (about 1 year ago)
- Last Synced: 2025-07-05T23:49:00.954Z (5 months ago)
- Language: Rust
- Size: 77.1 KB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
- awesome-rust-formalized-reasoning - microsat - tiny DPLL SAT-solver. (Projects / Provers and Solvers)
README
# :microscope: microsat
A tiny (_microscopic_) DPLL SAT-solver written in Rust. This is not meant to be:
1. A particularly _fast_ solver
2. A particularly _extensible_ solver
3. A particularly _useful_ solver
But instead serves as a proof-of-concept for what a small, readable, and understandable [DPLL](https://en.wikipedia.org/wiki/DPLL_algorithm) SAT Solver could look like in Rust.
This project originated as a project for Brown's [CSCI2951-O Foundations of Prescriptive Analysis](https://cs.brown.edu/courses/csci2951-o/).
Authors:
- [Rob Scheidegger ](https://github.com/RobScheidegger)
- [Hammad Izhar](https://github.com/Hammad-Izhar)
## Benchmarks
Although `microsat` isn't intended to be used as a fast SAT-solver, I felt it appropriate to compare it at a basic level to the project, [`minisat`](https://github.com/niklasso/minisat) (a small [CDCL](https://en.wikipedia.org/wiki/Conflict-driven_clause_learning) SAT solver that disrupted the SAT solver scene many years back). Times were for release-compiled variants of `microsat` and `minisat` on the same computer, for all of the examples in `examples/cnf`:
|| `microsat` | `minisat` |
|---|---|---|
|Time to solve example suite| 44.158s | 41.432s |
|Lines of code| 791 | 3517 |
As you can see, `microsat` does pretty remarkably well in this benchmark, despite being _much_ smaller than the already small `minisat`. Further, it is important to note that for any reasonably large instance (eg. larger than the `1040` variable, `3668` clause file in `examples/cnf`, which is the largest in this benchmark), so in a way, this benchmark is clearly cheating (but fascinating regardless).