Algorithmic Discovery & Superoptimization
Parent: AI Agent Ecosystems · Topic entry · 10 branches · skill ai-agents-orchestration/references/algorithmic-discovery-superoptimization.md
A published reference is not available for this topic yet.
Children
- Superoptimization & complexity reduction (STOKE, Souper, Massalin origins) (frontier)
- Program-synthesis & evolutionary substrate (sketching, CEGIS, SyGuS, GenProg) (frontier)
- AlphaTensor — TensorGame for matrix multiplication (frontier)
- AlphaDev — AssemblyGame for sorting/hashing (frontier)
- FunSearch — LLM proposer + systematic evaluator (frontier)
- AlphaEvolve — the evolutionary coding-agent keystone (frontier)
- Open descendants & sample-efficiency frontier (ShinkaEvolve, OpenEvolve) (frontier)
- LLM + formal-verifier hybrids for classical optimization (LPO) (frontier)
- ML-guided compiler optimization & autotuning (MLGO, OpenTuner, Halide) (frontier)
- Provable vs measured trust (frontier)
Frontier under this node: AlphaDev — AssemblyGame for sorting/hashing, AlphaEvolve — the evolutionary coding-agent keystone, AlphaTensor — TensorGame for matrix multiplication, FunSearch — LLM proposer + systematic evaluator, LLM + formal-verifier hybrids for classical optimization (LPO), ML-guided compiler optimization & autotuning (MLGO, OpenTuner, Halide), Open descendants & sample-efficiency frontier (ShinkaEvolve, OpenEvolve), Program-synthesis & evolutionary substrate (sketching, CEGIS, SyGuS, GenProg), Provable vs measured trust, Superoptimization & complexity reduction (STOKE, Souper, Massalin origins)