POAP: Plumbing for Optimization with Asynchronous Parallelism

POAP provides an event-driven framework for building and combining asynchronous optimization strategies. A typical optimization code written with POAP might look like:

from poap.strategy import FixedSampleStrategy
from poap.strategy import CheckWorkStrategy
from poap.controller import ThreadController
from poap.controller import BasicWorkerThread

# samples = list of sample points ...

controller = ThreadController()
sampler = FixedSampleStrategy(samples)
controller.strategy = CheckWorkerStrategy(controller, sampler)

for i in range(NUM_WORKERS):
    t = BasicWorkerThread(controller, objective)
    controller.launch_worker(t)

result = controller.run()
print 'Best result: {0} at {1}'.format(result.value, result.params)

The basic ingredients are a controller capable of asking workers to run function evaluations and a strategy for choosing where to sample. The strategies send the controller proposed actions, which the controller then accepts or rejects; the controller, in turn, informs the strategies of relevant events through callback functions.

Most users will probably want to provide their own strategies, controllers, or both.

Developers

Build Status: