Skip to main content

Asynchronous Differential Evolution, with efficient multiprocessing.

Project description

Performs the Differential Evolution (DE) algorithm asynchronously. With a multiprocess evaluation function running on a multicore CPU or cluster, ade can get the DE processing done several times faster than standard single-threaded DE. It does this without departing in any way from the numeric operations performed by the classic Storn and Price algorithm. You can use either a randomly chosen candidate or the best available candidate.

You get a substantial multiprocessing speed-up and the well-understood, time-tested behavior of the classic DE/rand/1/bin or DE/best/1/bin algorithm. (You can pick which one to use, or, thanks to a special ade feature, pick a probabilistic third version that effectively operates at a selected midpoint between the extremes of "random" and "best.") The underlying numeric recipe is not altered at all, but everything runs a lot faster.

The ade package also does simple and smart population initialization, informative progress reporting, adaptation of the vector differential scaling factor F based on how much each generation is improving, and automatic termination after a reasonable level of convergence to the best solution.

Comes with a couple of small and informative example files, which you can install to an ade-examples subdirectory of your home directory by typing ade-examples as a shell command.

For a tutorial and more usage examples, see the project page at edsuom.com.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

ade-1.3.3.tar.gz (92.0 kB view hashes)

Uploaded Source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page