Seminar

Speaker: M. Bureš

One of the key stages of data processing from particle physics experiments is the reconstruction of trajectories (tracks) of interacting particles from measurement data. In the SPD experiment planned at the NICA collider, a special difficulty will be caused by the extremely high frequency of interactions (3 MHz) due to the high luminosity of the particle beams, leading to overlap of events during their acquisition in the time-slice mode, as well as by the strong contamination of data by false hits due to the peculiarities of the SPD track detectors. This makes track reconstruction (tracking) algorithms very complicated. This study investigates methods based on the Hopfield neural network for tracking simulated events of the SPD experiment. Optimization of parameters for constructing the neural network energy function is proposed, which allows to improve tracking results taking into account the specifics of the experiment. The applicability of quantum algorithms to the SPD tracking problem is also investigated. In this setting, the tracking problem is formulated as quadratic unconstrained binary optimization (QUBO) and solved by simulated annealing or quantum annealing. In a number of recent publications, it has been shown that combinatorial optimization problems can be successfully solved in this way and efficient track recovery is possible, indicating that it can be used for fast processing of SPD data or other high-luminosity experiments. We should also point out an interesting possibility of applying the quantum Harrow-Hassadim-Lloyd (HHL) algorithm, which, when applied to the optimization of the Hopfield network, can serve to further accelerate the search for the global minimum of the proposed matrix representing the network energy function.

Information on the seminar and the link to connect via Webinar are available at Indico