Marija Vucelja

Marija Vucelja is a statistical physicist interested in soft condensed matter and computational physics.

Lately, Marija has been focused on anomalous thermal relaxations of physical systems, phase transitions, and efficient randomized algorithms (such as those utilizing non-reversible Markov Chains) designed for evaluating marginals and inference and learning, use of neural networks and deep learning to enhance MCMC, and inference and learning.

Over the years, Marija studied the mixing and clustering of particles in flows, problems relevant to understanding the formation of rain droplets and planetesimals, clumping of pollutants on water surfaces, and industrial applications. She derived the compressibility of surface flows and described the aggregation-disorder transition of particles in flows. Using “chaotic mixing,” she substantially accelerated specific Monte Carlo algorithms (the main numerical tools for studying complex systems). Next, Marija investigated the emergence of clones in populations. Drawing analogies between glassy systems and population dynamics, she calculated the coalescence rate (the probability of two individuals belonging to the same clone).

• I. Klich, O. Raz, O. Hirschberg, and M. Vucelja, The Mpemba index and anomalous relaxation, Phys. Rev. X., 9, 2, 021060 (2019)

• K. S. Turitsyn, M. Chertkov, and M. Vucelja, Irreversible Monte Carlo Algorithms for Efficient Sampling, Physica D 240, 410 - 414 (2011)