Talks

The field of artificial intelligence (AI) has experienced major developments over the last decade. Within AI, of particular interest is the paradigm of reinforcement learning (RL), where autonomous agents learn to accomplish a given task via feedback exchange with the world they are placed in, called an environment. Thanks to impressive advances in quantum technologies, the idea of using quantum physics to boost the performance of RL agents has been recently drawing the attention of many scientists. In my talk I will focus on the bridge between RL and quantum mechanics, and show how RL has proven amenable to quantum enhancements. I will provide an overview of the most recent results — for example, the development of agents deciding faster on their next move [1]— and I will then focus on how the learning time of an agent can be reduced using quantum physics. I will show that such a reduction can be achieved and quantified only if the agent and the environment can also interact quantumly, that is, if they can communicate via a quantum channel [2]. This idea has been implemented on a quantum platform that makes use of single photons as information carriers. The achieved speed-up in the agent’s learning time, compared to the fully classical picture, confirms the potential of quantum technologies for future RL applications. [1] Sriarunothai, T. et al. Quantum Science and Technology 4, 015014 (2018). [2] Saggio, V. et al. Nature 591, 229–233 (2021).

Dense hydrogen, the most abundant matter in the visible universe, exhibits a range of fascinating physical phenomena such as metallization and high-temperature superconductivity, with significant implications for planetary physics and nuclear fusion research. Accurate prediction of the equations of state and phase diagram of dense hydrogen has long been a challenge for computational methods. In this talk, we present a deep generative model-based variational free energy approach to tackle the problem of dense hydrogen, overcoming the limitations of traditional computational methods. Our approach employs a normalizing flow network to model the proton Boltzmann distribution and a fermionic neural network to model the electron wavefunction at given proton positions. The joint optimization of these two neural networks leads to a comparable variational free energy to previous coupled electron-ion Monte Carlo calculations. Our results suggest that hydrogen in planetary conditions is even denser than previously estimated using Monte Carlo and ab initio molecular dynamics methods. Having reliable computation of the equation of state for dense hydrogen, and in particular, direct access to its entropy and free energy, opens new opportunities in planetary modeling and high-pressure physics research.

The universe has turned out to be simpler than expected on both very small and very large scales. We propose a minimal, highly predictive framework connecting particle physics to cosmology. Instead of introducing an ``attractor” phase such as inflation we extrapolate the observed universe all the way back to the initial singularity where we impose a CPT symmetric boundary condition via a generalization of the method of images. If the hot plasma in the early universe is perfectly conformal, so is the singularity. The cosmos may then be analytically extended to a ``mirror image” universe prior to the bang. Using this new boundary condition we calculate the gravitational entropy for cosmologies with radiation, matter, Lambda and space curvature, finding it favours spatially flat, homogeneous and isotropic universes with a small positive cosmological constant in accord with observation. To maintain conformal symmetry, we include unusual Dim-0 (dimension zero) fields. They improve the SM’s coupling to gravity, cancelling the vacuum energy and two local “Weyl” anomalies, without introducing additional propagating modes. They also cancel pathologies introduced into the graviton propagator by loops of SM particles. Cancellation requires precisely 3 generations of SM fermions, each with a RH neutrino. It also requires a composite Higgs, presumably built with the Dim-0 fields. One of the RH neutrinos, if stable, is a viable candidate for the dark matter which will be tested soon. The Dim-0 fields source scale-invariant curvature perturbations in the early universe. Subject to two simple but crucial theoretical assumptions, the amplitude and spectral tilt match the observations with remarkable accuracy. (See arXiv:2302.00344 and references therein).

Zoom Link: https://pitp.zoom.us/j/95784600151?pwd=dkppa2s3ZDM4NG5yb0ZVV2w5SXErdz09

ZOOM: https://pitp.zoom.us/j/94595394881?pwd=OUZSSXpzYlhFcGlIRm81Y3VaYVpCQT09 We introduce a novel neural quantum state architecture for the accurate simulation of extended, strongly interacting fermions in continuous space. The variational state is parameterized via permutation equivariant message passing neural networks to transform single-particle coordinates to highly correlated quasi-particle coordinates. We show the versatility and accuracy of this Ansatz by simulating the ground-state of the 3D homogeneous electron gas at different densities and system sizes. Our model respects basic symmetries of the Hamiltonian, such as continuous translation symmetries. We compare our ground-state energies to results obtained by different state-of-the-art NQS Ansaetze for continuous space, as well as to different quantum chemistry methods. We obtain better or comparable ground-state energies, while using orders of magnitudes less variational parameters and optimization steps. We investigate its capability of identifying and representing different phases of matter without imposing any structural bias toward a given phase. We scale up to system sizes of N=128 particles, opening the door for future work on finite-size extrapolations to the thermodynamic limit.

ZOOM: https://pitp.zoom.us/j/94595394881?pwd=OUZSSXpzYlhFcGlIRm81Y3VaYVpCQT09 In this work, we use data-driven methods to reduce the dimensionality of the vertex function for the Hubbard model and spin liquid model. By employing a deep learning architecture based on the autoencoder, we show that the functional renormalization group (FRG) dynamics can be efficiently learned. Our approach is compared with other methods, including principal component analysis and dynamic mode decomposition. Our results demonstrate the effectiveness of our proposed approach for understanding the FRG flow in these models.