Published:
As you might know from my previous posts, during the summer of 2023 I’ve been working on MDAnalysis’s project during Google Summer of Code. Here I’ll summarize what I’ve done, how others can use it, and what changes will follow that in the MDAnalysis codebase in the near future.
Published:
In the last blogpost, I explained how I decided to lift the parallelization class up, and how AnalysisBase.run() method changed after introducing this.
ParallelExecutor class Published:
In the previous blogpost, I briefly explained how decomposition works – we split all _single_frame() runs into independent groups that get executed in parallel. For this, we have _compute method that executes the frames group, and run method that orchestrates the _compute execution.
_setup_bslices Published:
In the previous blogpost, I briefly explained how I’m planning to decompose the AnalysisBase.run() method, so that its subclasses won’t notice the changes in the protocol, but at the same time will be able to perform computations in a parallel manner.
AnalysisBase.run() Published:
In the first GSOC-devoted blogpost I explained the idea behind my proposal and the approximate timeline for its execution.
Published:
This year I applied for the Google Summer of Code program, and luckily was selected as a contributor for MDAnalysis with my proposal for a parallel analysis project, extending the idea suggested by the developers.