Sun Woo Kim

Research

I am broadly interested in statistical physics, with a focus on dynamics in non-equilibrium systems, in both classical and quantum settings.

My most recent work was on Bayesian inference, specifically on hidden Markov models. In a 'filtering' task, the posterior distribution of the current state is inferred from the history of noisy measurements. For a simple model of inferring a random walk from its noisy images, we showed that the evolution of posterior distribution formally maps to the Boltzmann probability of the directed polymer, a well-studied problem, but with an extra delta potential at the signal location.

Recently, there has also been lots of attention on monitored quantum circuits, where unitary evolution is combined with measurements. This process is reminiscent of evolving the posterior through a Markov kernel then updating it with a measurement model/likelihood in Bayesian inference. I am currently working on the connection between the two problems.

My previous work has been on efficiently simulating dynamics of many-body localised systems by exploiting Poincaré-Linstedt perturbation theory to access very long times. In this work we recovered some of the hallmarks of MBL such as logarithmically slowly spreading of out-of-time correlators.