Scaling functional multi-agent computations (...)
summary
Claims that a streaming execution model yields fine-grained concurrency for multi-agent evolutionary systems and implements it with Akka Streams on Scala. Key mechanisms are a looping graph with feedback buffers and a reservoir-style shuffle that supports random / best-fitness / hybrid buffers. Experiments on benchmark optimization show the hybrid (annealing) buffer gives better exploration then convergence and that a barrier buffer reproduces sequential generations while the random buffer matches actor-like behavior.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 0:24Evolutionary multiagent systems
- 1:12Example
- 3:03Initial implementations
- 4:04Design pattern
- 7:16Reactive streams
- 7:57Dynamic backpressure
- 8:42Scala stages
- 11:11Reactive strings
- 11:47Step stage
- 13:22Reservoir sampling
- 14:22Random shuffling buffer
- 17:52Future work
- 18:27Questions
related talks
51mComposing high performance process workflows with Akka Streams
56mAkka in Heterogenous Environments
36mStream All The Things!
26mFrom Chaos to Choreography: Multi-Agent Orchestration Patterns That Actually Work
42mSpeck flavored streaming micro services with Akka
24mUsing an actor framework for scientific computing
43mIntroduction to Akka Actors: Concurrency and Distributed Systems for the JVM
05mVladimir Bacvanski Interview