QLunch: Emilo Onorati

Speaker: Emilio Onorati

Title: Quantum generators: robust learning and sharp approximation

Abstract: Quantum dynamics are governed by a large and complex family of generators that encode fundamental properties of quantum systems. Their structure allows us to predict future evolutions and to characterise with precision the noise affecting experimental implementations.

 In the first part of this talk, we will consider the task of learning the Lindblad generator of an open quantum system. We will discuss how to retrieve the full description of a local Lindbladian from just a few tomographic snapshots, formulating the problem as a semidefinite programme constrained by the exact conditions of a Markovian process.

In the second part, we will turn to the Magnus expansion – the exponential representation of continuous-time Hamiltonian drives. To handle its increasingly complex series, we will introduce a novel, sharp error bound for the truncation tail. We will show how this result elegantly emerges from a natural correspondence with the framework of binary trees, yielding a sub-quadratic improvement over standard bounds.

References:

[1] E. Onorati, T. Kohler, and T. Cubitt, ‘Fitting quantum noise models to tomographic data’, https://arxiv.org/abs/2103.17243

[2] Y. Liu, J. R. Seddon, T. Kohler, E. Onorati, T. Cubitt, ‘Robust Lindbladian Estimation for Quantum Dynamics’, https://arxiv.org/abs/2507.07912

[3] H. Apel, T. Cubitt, and E. Onorati, ‘A sharper Magnus expansion bound woven in binary branches’, https://arxiv.org/abs/2509.18312