CMS Open Data HH → 4b Multijet & Mixed Datasets

Reconstructed collision, hemisphere-mixed synthetic data, and simulations in Parquet format for di-Higgs multijet background modelling.

This dataset contains flat tabular (Apache Parquet) collision data and simulated physics processes released by the CMS Collaboration on the CERN Open Data Portal (Record cms-93965) and hosted on Hugging Face (alefisico/LHC_HH4b_mixeddata).

It is prepared for machine-learning-based background modelling, density ratio estimation, and domain-adaptation reweighting in multijet searches for di-Higgs production in the four-bottom-quark final state ($pp \to HH \to (b\bar{b})(b\bar{b})$).

Key Features

  • Format: Flat tabular Apache Parquet files partitioned by process and tag categories.
  • Collision Data: CMS Run 2 ($13\text{ TeV}$) multijet collision data in the 3b source domain.
  • Synthetic Mixed Data: Signal-free 4b pseudo-data generated via hemisphere mixing for authentic domain-adaptation benchmarks and unblinded evaluation.
  • Simulations: Detailed Monte Carlo samples for $t\bar{t}$ and resonant backgrounds ($ZH$, $ZZ$, and $HH$ signal).
  • Interactive Quickstart: End-to-end tutorial notebook runnable directly in Google Colab without downloading the full dataset locally.

Links & DOI