search

guicho271828 / common-lisp-project-ideas

Discuss future project ideas
12 stars 0 forks source link

Generalized Genetic Algorithms/Evolutionary Computing? #8

Open olewhalehunter opened 7 years ago

olewhalehunter commented 7 years ago

There are many Lisp libraries for domain-specific genetic algorithms but no good generalized ones (genome definition, problem space mappings, performance profiling, fitness functions)

guicho271828 commented 7 years ago

https://jorgetavares.com/2017/05/03/gp-code-on-github/ is this yours?

olewhalehunter commented 7 years ago

"core-gp" looks promising, but it could do with better documentation, enough to where a beginner-level lisp programmer could get started with genetic programming.

guicho271828 commented 7 years ago

Thinking about the "next big thing", having a good GP framework sounds increasingly beneficial. Cf. Neuroevolution, https://arxiv.org/abs/1703.01041

guicho271828 commented 7 years ago

Related: Automated parameter tuning (e.g. ParamILS http://www.cs.ubc.ca/labs/beta/Projects/ParamILS/)

olewhalehunter commented 7 years ago

this is the literature I'm currently reading: http://www.informit.com/store/genetic-algorithms-for-vlsi-design-layout-and-test-9780130115669

some other considerations for a general framework are parallelism and distributed processing

guicho271828 commented 7 years ago

parallelism and distributed processing

If you are talking about parallel GA then yeah, but is something that should be done later.

In terms of "parallel computation" in general, we already have several frameworks: For embarassingly parallel computation, lparallel and its distributed variant do the enough jobs. For HPC tasks requiring much better coordination I recommend CL-MPI. Personally I do not feel the need of new libraries in that field.

guicho271828 commented 6 years ago

Do you think this is a good place to go? https://github.com/guicho271828/dynotune

Note: The old dynotune is renamed to dynotune-failed. I abandoned the idea and reuse the name for this new project.

olewhalehunter commented 6 years ago

this is some great stuff

generalized parameter optimization is a bit esoteric to me, and likely the average programmer, any resources you would recommend to understand this library's semantics better? perhaps a concise review of optimization explained using expressions from dynotune would help in the readme

guicho271828 commented 6 years ago

oh it just tries to find the local/global minima of the function