Feasibility studies for HH->bbtautau at the FCC-hh using graph-neural networks

A feasibility study of the di-Higgs production channel at the Future Circular Collider proton-proton collider, FCC-hh, has been conducted focusing on the $ hh \to b\bar{b} \tau^+ \tau^- $ final state. Both the leptonic-hadronic (lep-had) and fully hadronic (had-had) decays of the taus are  considered, leveraging advanced Graph Neural Network (GNN) techniques for event classification capabilities.  The analysis employs Graph Attention Networks (GATs), with each event represented as a fully connected graph, where nodes correspond to reconstructed physics objects, and edges capture their relational properties. Additional complex features -- such as invariant masses of $b$-jet and tau leptons, angular separations, and transverse kinematic variables -- are integrated to enhance event classification. This innovative application of GNNs significantly improve the ability to extract signal events, offering a powerful technique for precision Higgs self-coupling measurements at the FCC-hh. Considering a centre-of-mass energy of 84~TeV, a statistical significance of $Z\sim 26$ ($Z\sim 92$) is expected for the lep-had (had-had) channel, under the no-systematic assumption and for a specific value of the signal score threshold. Considering systematic uncertainties at the  2\% on the SM background, the expected significances is around 30 for the had-had channel alone, demonstrating how the Higgs self-coupling could be constrained at the percentage level at the FCC-hh.