Borjan Geshkovski
Papers
(Should you need a published version, please contact me.)

  10.     A mathematical perspective on Transformers
      with Cyril Letrouit, Yury Polyanskiy and Philippe Rigollet
      preprint, 2023
      arXiv, GitHub
  9.     The emergence of clusters in self-attention dynamics
      with Cyril Letrouit, Yury Polyanskiy and Philippe Rigollet
      NeurIPS, 2023
      arXiv, GitHub
  8.     Control of the Stefan problem in a periodic box
      with Debayan Maity
      Mathematical Models and Methods in Applied Sciences, 2023
      arXiv, GitHub
  7.     Turnpike in optimal control of PDEs, ResNets, and beyond
      with Enrique Zuazua
      Acta Numerica, 2022
      arXiv, GitHub
  6.     Optimal controller design via Brunovsky's normal form
      with Enrique Zuazua
      IEEE Transactions on Automatic Control, 2023
      arXiv, GitHub
  5.     Sparsity in long-time control of neural ODEs
      with Carlos Esteve
      Systems & Control Letters, 2023
      arXiv, GitHub
  4.     Turnpike in Lipschitz-nonlinear optimal control
      with Carlos Esteve, Dario Pighin and Enrique Zuazua
      Nonlinearity, 2022
      arXiv, GitHub
  3.     Large-time asymptotics in deep learning
      with Carlos Esteve, Dario Pighin and Enrique Zuazua
      preprint, 2021
      arXiv, GitHub
  2.     Controllability of one-dimensional viscous free boundary flows
      with Enrique Zuazua
      SIAM Journal on Control and Optimization, 2021
      arXiv
  1.     Null-controllability of perturbed porous medium gas flow
      ESAIM Control, Optimisation and Calculus of Variations, 2020
      arXiv

Theses

T1. Control in moving interfaces and deep learning
      Ph.D. Thesis
      Universidad Autónoma de Madrid, 2021
      (For some reason, the version of my thesis published on the UAM library webpage contains
      formatting errors that were not caused by me.)

T2. Obstacle problems: theory and applications
      M.Sc. Thesis
      Université de Bordeaux, 2018