Speaker
Description
The upcoming next-generation neutrino experiments, notably the Deep Underground Neutrino Experiment (DUNE), and the joint analysis of existing experiments, such as that of T2K and NOvA, bring long-baseline neutrino experiments into the precision era. Performing $5\sigma$ measurements of neutrino properties using Bayesian analysis typically requires hundreds of billions of Markov-Chain Monte Carlo (MCMC) steps, straining existing computational resources. Efficient sampling techniques, such as adaptive MCMC, can meaningfully reduce these requirements and hence enable high-significance measurements to be made within reasonable resource constraints. I present the application of adaptive MCMC to DUNE and report a reduction in the number of MCMC steps by a factor of 50, cutting person-power and computation time from months to days.