Home

Research

Statistical physics of natural and artificial systems. We work at AI × Physics: the physics of AI and AI for physics, and in condensed matter and quantum science.

Artwork of a crystal lattice of octahedra threaded by light.

AI

  • Physics of AI. Understanding deep learning with the tools of physics.
  • Generalization. Learning physical laws that generalize beyond the training data.
  • Science of scaling. Scaling laws, optimization, optimizers, and data.
  • Controlling AI systems. System-level control of LLMs, diffusion models, and world models.
  • AI for physics. Autoscience, data, and post-training domain models for physics.
Scaling of loss with model size and data size for sparse and dense features.
Asymmetric scaling laws from sparse features.

Physics

  • Energy transport. Exciton and polaron transport.
  • Quantum geometry in electron-phonon coupled materials. Building on bipolaronic high-temperature superconductivity.
  • Light-matter interaction. Light-induced superconductivity and dynamics in optically excited materials.
  • Quantum simulation of 2D systems. Efficient algorithms based on tensor networks.
  • Methods. Tensor networks and quantum Monte Carlo.
Superconducting transition temperature versus electron-phonon coupling for several phonon frequencies.
Bipolaronic high-temperature superconductivity.