README.md

Experiments on pairwise distances aggregations

See benchmark results

tl;dr

git clone git@github.com:scikit-learn-inria-fondation/pdist_aggregation.git
cd pdist_aggregation
make

Setup

Simply:

git clone git@github.com:scikit-learn-inria-fondation/pdist_aggregation.git
cd pdist_aggregation
conda env create -f environment.yml
conda activate pdist_aggregation

See:

make help

Running benchmarks on GNU/Linux

You can adapt the benchmarks' configuration editing benchmarks/config.yml.

And then simply run the benchmark script:

make benchmark-parallel

PDF reports will be written in a subfolder in results.

The implementation can be capped to the sequential execution using:

make benchmark-sequential

⚠ Currently this make target has been written for GNU/Linux as it makes uses of taskset(1) but you can adapt it easily using environment variables.