New paths through mathematics.
FunSearch combined language models with program evaluation to discover mathematical constructions.
- Maturity
- Demonstrated, stage 3 of 4
- Support
- 1 source · paper
- Evidence detail
- How we know ↓
Some combinatorial problems have vast search spaces.
Researchers develop and evaluate candidate constructions.
Generated programs were tested, selected, and iteratively improved.
What was shown, and what wasn’t
Shown
The study found new constructions for the cap set problem and useful packing heuristics.
Not shown · limits
- The approach depends on an evaluator; it is not a general proof engine.
Bright editorial interpretation
Researchers gained an additional experimental tool.
Still open
Apply verifiable search to other suitable problems.
How we know1 source · checked 2026-09-05 · no corrections
Original sources
Institutions: Google DeepMind
- Maturity
- Demonstrated
- Source published
- 2023-12-14
- Last source review
- 2026-09-05
- Editorial method
- Original source check
- Place / relevance
- London, United Kingdom · institution-location
Bright compared this account with the linked original and supporting sources and kept reported, budgeted, projected, and observed claims distinct. Bright did not independently audit the underlying records.
Maturity describes the tested or operational setting. Confidence describes support for the particular claim; one does not determine the other.
Revision & correction history
No corrections recorded.
Keep exploring
Explore the shared question in another setting. These connections do not imply replication.