Ramble is an open source experimentation framework written in Python, focused on capturing provenance information and improving the reproducibility of computational experiments. It allows users to rapidly create, execute, analyze, and manage large sets of experiments across HPC, AI, and ML environments.

The framework supports a wide range of use cases, including test suites, parameter studies, performance tuning explorations, and acceptance testing. Ramble can also automate several stages of an experiment’s lifecycle, including downloading input files, creating software stacks through package managers such as Spack, executing experiments, extracting figures of merit, and archiving results.