Researcher for the Florence Section of INFN and member of the LHCb Collaboration since 2009, I coordinate the effort to adopt Machine Learning techniques in INFN research as National Representative for the ML INFN initiative (CSN5). I also lead the Simulation Techniques Work-Package of the LHCb Simulation Project, focussing on technologies for faster simulations. Since January 2022, I am National Representative (Responsabile Nazionale del Calcolo) for the LHCb Computing resources.
In order to increase the rate of collision events that can be studied with large scale experiments to accumulate a larger number of candidates for new states and decays, I am investigating new technologies to conceive novel simulation techniques and tracking detectors.
The detector simulation is computationally very expensive and not sustainable for the future Runs of the LHCb experiment, requiring larger and larger simulated samples to control the statistical and systematic uncertainties. Parts of the detector simulation, however, can be replaced with machine-learning algorithms trained on the detailed detector simulation or on real data, instead of simulating the quantum interactions for each particle of each event, reducing the computational cost of the simulated samples by several orders of magnitude.
While traditional tracking detectors measure the position at which a charged particle traverses subsequent layers of sensors, timing tracking detectors enable associating a crossing-time at each position, opening to four-dimensional tracking. Higher particle rate may also result into premature aging of the sensors due to radiation damage. To reduce this effect, semiconductors with atomic numbers smaller than silicon are studied. I contribute to the development of diamond sensors with time resolution of the order of few picoseconds to empower timing tracking detectors, within the TimeSPOT and IGNITE initiatives