Quantum Machine Learning Analysis of LHC data on IBM Quantum Computers: a case of potential quantum advantage

by Dr Panagiotis Barkoutsos (IBM Research Zurich)


In this work, we investigate the use of quantum machine learning techniques for the classification of high energy collision events as measured at the Large Hadron Collider (LHC).
In particular, we focus on the use of the support vector machine with quantum kernel estimation (SVM-QKE) for the analysis of Higgs boson production in association with a top quark pair ttH. Exploiting the potential exponential advantage associated with the use of the qubit Hilbert space for the definition of a highly dimensional feature space and corresponding scalar product we demonstrate the great potential of quntum kernel estimator for support vector machines in the analysis of high energy physics (HEP) data. In the particular case of the ttH process, we obtain both in simulation and by means of experiments using IBM quantum computers a classification performance that is equivalent, if not superior, to the performance of the currently best classical classifiers used in the HEP community. This preliminary study limited to a maximum of 15 qubits and 10 features (i.e., cross section parameters) provides nonetheless a clear hint for quantum advantage in LHC data classification. In this talk I will give an introduction to quantum computing for ML and discuss the details pertinent to quantum SVM and how they can be used for HEP simulations. I will also talk about the experiments and simulations we performed in order to demonstrate the use of near term quantum devices for classification purposes.


Videoconference via