Skip to content

Parallel Sampling #70

Open
Open
@mileslucas

Description

@mileslucas

Although sampling in NS is not as easily parallelizable as MCMC, there are still opportunities for multi-threading/multi-processing during the sample proposal step. Following the formalism of https://www.wesleyhenderson.me/pdfs/Parallelized_Nested_Sampling_Henderson_Goggans.pdf when we propose a new point during the sample steps, the proposals can be done in parallel, accepting the first new point that fulfills the likelihood constraints. This is implemented in dynesty using Python's multiprocessing.pool.

If we overload AbstractMCMC.sample methods using the MCMCDistributed and MCMCThreads structs, we should be able to do something similar, although I don't have a clear idea of the implementation figured out.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions