Companion for the article "Profiling a task-based molecular dynamics application with a data science approach"

This repository contains the datasets, analysis scripts, and visualization outputs accompanying the paper "Profiling a Task-Based Molecular Dynamics Application with a Data Science Approach", submitted to CARLA 2025. It includes processed performance traces of LeanMD simulations using Charm++, along with the full analysis pipeline implemented using the CharmVZ tool (sources provided as a snapshot of https://github.com/caschb/charmvz). The code is organized into modular components for trace parsing, CSV and Parquet generation, and plot creation using data science libraries. One computational notebook is provided to reproduce the figures from the paper. This release promotes transparency and reproducibility, and supports the broader adoption of open data and software practices in HPC performance analysis.