External pipelines · RISE project catalogue
AlphaEvolve (Google DeepMind)
A Gemini-powered evolutionary coding agent that combines LLM generative capabilities with automated evaluators in an iterative propose-test-refine loop. Targets the *algorithmic and scientific discovery* arc of the RISE pipeline: given a well-defined evaluator, AlphaEvolve searches the space of programs that improve the metric the evaluator scores. Demonstrated on data-center scheduling, hardware-circuit design, kernel optimization, and — most relevant to RISE — open mathematical problems.
Where it sits
left: what it builds on · right: what builds on it · pale: exampleContributed by Google DeepMind
How studies reach it
No published study reaches it yet.
Disciplines it reaches
No study reaches it yet.
Solid: published studies. Light: examples.
Computed from the records on this site: what each study, template and specialist names as used, which study extends which, and who contributed what. 0 studies in total.
What it does
An evolutionary-search harness over LLM-generated code that decouples *generation* (Gemini Pro / Flash) from *evaluation* (a domain-specific automated scoring function). In Georgiev, Gómez-Serrano, Tao & Wagner (2025) the system rediscovers best-known constructions for 56 of 67 mathematical problems and improves on the best-known result in several, occasionally generalizing finite-case results into closed-form expressions valid for all inputs. Distinctive among RISE-landscape projects because it does not write papers — it discovers algorithmic / mathematical *artifacts*, leaving the scholarly framing to humans.
- Focus
- end-to-end
- Inputs
- problem-specification, automated-evaluator
- Outputs
- algorithmic-artifacts, improved-constructions, closed-form-expressions
- Architecture
- evolutionary-search, tool-use, iterative-loop, automated-evaluation
- Maintained by
- Google DeepMind
- Started
- 2025
Description
Data model- Discipline
- Mathematics
- Method family
- not specified
- Design
- not specified
- Research stage
- HypothesesCode generationData analysis
- Contributors
- Google DeepMind
- Usage
- not used in published research yet
- Source
- RISE project catalogue · projects/landscape · @4c17bae
- Record
- pipeline:alphaevolve · JSON
Solid tags are declared by the source or mapped from its terms; dashed tags are inferred by a published rule. Hover a tag for its provenance.
Bring it into the standard
A pipeline built outside E2ER can meet the standard by describing its steps as a template, attaching the floor of checks and publishing evaluation records. Its authors keep ownership and credit.