BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:A roadmap for an atomistic machine learning software ecosystem
DTSTART;VALUE=DATE:20260119
DTSTAMP:20260506T193322Z
UID:cb60632699aa6118423b07cb5ad0193a16e177da190bc62ec884761b
CATEGORIES:Conferences - Seminars
DESCRIPTION:You can apply to participate and find all the relevant informa
 tion (speakers\, abstracts\, program\,...) on the event website: https://
 www.cecam.org/workshop-details/a-roadmap-for-an-atomistic-machine-learning
 -software-ecosystem-1376.\n\nRegistration is required to attend the full e
 vent\, take part in the social activities and present a poster at the post
 er session (if any).  However\, the EPFL community is welcome to atten
 d specific lectures without registration if the topic is of interest to 
 their research. Do not hesitate to contact the CECAM Event Manager if yo
 u have any question.\n\nDescription\n\nThe learning of accurate and effici
 ent atomic interaction potentials from quantum mechanical calculations has
  been one of the first applications of machine learning in the physical sc
 iences [1]\, and it is arguably one of the most successful. In the last 15
  years\, the development of machine-learned interaction potentials (MLIP) 
 has evolved significantly and many different models have been proposed and
  applied to a broad set of systems in material science\, chemistry\, and b
 iophysics [2-4]\, expanding also beyond the prediction of interatomic pote
 ntials to include any quantity accessible to electronic-structure calculat
 ions. The mathematical and conceptual framework underlying ML architecture
 s for atomistic simulations is now relatively well understood [5-7] and di
 fferent models share many common features. For instance\, equivariance is 
 a common theme as well as the use of message passing mechanisms between lo
 cal atomic environments - even though schemes to incorporate long-range ph
 ysics [8\,9] and unrestricted models that relax some of the symmetry const
 raints are also actively developed [10].\nDespite the many common ideas\, 
 the software ecosystem is currently very fragmented: each model usually co
 mes with its own monolithic implementation\, and there is little shared in
 frastructure\, even though lately some efforts have started to appear that
  aim to provide basic functionalities in a more general setting (e.g. e3nn
  [11]\, dscribe [12]\, sphericart [13]). This is due in part to the very f
 ast development of the field\, in part to the fact that rapid prototyping 
 is possible using general-purpose ML libraries such as Pytorch and Jax\, i
 n part to the difficulty in producing efficient domain-specific libraries 
 that can fully exploit accelerated hardware platforms. \nAnother open que
 stion is how to provide easy-to-use interfaces between the ML core and dif
 ferent types of traditional modelling software. Most machine-learning inte
 ratomic potentials provide a library interface to call them from a molecul
 ar dynamics software such as LAMMPS or OpenMM\, but this requires substant
 ial effort for each target code. When one considers the prediction of prop
 erties such as electronic densities or Hamiltonians that involve a interfa
 cing with quantum chemistry software\, the situation is even more problema
 tic\, as one has to handle quantities with a considerably more complicated
  structure\, and performing a machine learning task is more intimately cou
 pled to the technical choices in the host code\, such as the choice of bas
 is set used to discretize the electronic wavefunction.[14-16].\n\nReferenc
 es\n\n[1] J. Behler\, M. Parrinello\, Phys. Rev. Lett.\, 98\, 146401 (200
 7)\n[2] J. Smith\, O. Isayev\, A. Roitberg\, Chem. Sci.\, 8\, 3192-3203 (
 2017)\n[3] Y. Zhou\, W. Zhang\, E. Ma\, V. Deringer\, Nat. Electron.\, 6\
 , 746-754 (2023)\n[4] J. Abramson\, J. Adler\, J. Dunger\, R. Evans\, T. G
 reen\, A. Pritzel\, O. Ronneberger\, L. Willmore\, A. Ballard\, J. Bambric
 k\, S. Bodenstein\, D. Evans\, C. Hung\, M. O’Neill\, D. Reiman\, K. Tun
 yasuvunakool\, Z. Wu\, A. Žemgulytė\, E. Arvaniti\, C. Beattie\, O. Bert
 olli\, A. Bridgland\, A. Cherepanov\, M. Congreve\, A. Cowen-Rivers\, A. C
 owie\, M. Figurnov\, F. Fuchs\, H. Gladman\, R. Jain\, Y. Khan\, C. Low\, 
 K. Perlin\, A. Potapenko\, P. Savy\, S. Singh\, A. Stecula\, A. Thillaisun
 daram\, C. Tong\, S. Yakneen\, E. Zhong\, M. Zielinski\, A. Žídek\, V. B
 apst\, P. Kohli\, M. Jaderberg\, D. Hassabis\, J. Jumper\, Nature\, 630\,
  493-500 (2024)\n[5] F. Musil\, A. Grisafi\, A. Bartók\, C. Ortner\, G. C
 sányi\, M. Ceriotti\, Chem. Rev.\, 121\, 9759-9815 (2021)\n[6] J. Nigam\
 , S. Pozdnyakov\, G. Fraux\, M. Ceriotti\, The Journal of Chemical Physics
 \, 156\, (2022)\n[7] I. Batatia\, S. Batzner\, D. P. Kovács\, A. Musaeli
 an\, G. N. C. Simm\, R. Drautz\, C. Ortner\, B. Kozinsky\, and G. Csányi\
 , "The design space of E(3)-equivariant atom-centered interatomic potentia
 ls\," arxiv:2205.06643 (2022)\n[8] T. Ko\, J. Finkler\, S. Goedecker\, J. 
 Behler\, Nat. Commun.\, 12\, 398 (2021)\n[9] K. Huguenin-Dumittan\, P. Lo
 che\, N. Haoran\, M. Ceriotti\, J. Phys. Chem. Lett.\, 14\, 9612-9618 (20
 23)\n[10] S. Pozdnyakov and M. Ceriotti\, "Smooth\, exact rotational symme
 trization for deep learning on point clouds\," in Advances in Neural Infor
 mation Processing Systems (Curran Associates\, Inc.\, 2023)\, 36\, pp. 794
 69–79501.\n[11] M. Geiger and T. Smidt\, "e3nn: Euclidean neural network
 s\," arxiv:2207.09453 (2022).\n[12] L. Himanen\, M. Jäger\, E. Morooka\, 
 F. Federici Canova\, Y. Ranawat\, D. Gao\, P. Rinke\, A. Foster\, Computer
  Physics Communications\, 247\, 106949 (2020)\n[13] F. Bigi\, G. Fraux\, 
 N. Browning\, M. Ceriotti\, The Journal of Chemical Physics\, 159\, (2023
 )\n[14] J. VandeVondele\, M. Krack\, F. Mohamed\, M. Parrinello\, T. Chass
 aing\, J. Hutter\, Computer Physics Communications\, 167\, 103-128 (2005)
 \n[15] P. Giannozzi\, S. Baroni\, N. Bonini\, M. Calandra\, R. Car\, C. Ca
 vazzoni\, D. Ceresoli\, G. Chiarotti\, M. Cococcioni\, I. Dabo\, A. Dal Co
 rso\, S. de Gironcoli\, S. Fabris\, G. Fratesi\, R. Gebauer\, U. Gerstmann
 \, C. Gougoussis\, A. Kokalj\, M. Lazzeri\, L. Martin-Samos\, N. Marzari\,
  F. Mauri\, R. Mazzarello\, S. Paolini\, A. Pasquarello\, L. Paulatto\, C.
  Sbraccia\, S. Scandolo\, G. Sclauzero\, A. Seitsonen\, A. Smogunov\, P. U
 mari\, R. Wentzcovitch\, J. Phys.: Condens. Matter\, 21\, 395502 (2009)\n
 [16] V. Blum\, R. Gehrke\, F. Hanke\, P. Havu\, V. Havu\, X. Ren\, K. Reut
 er\, M. Scheffler\, Computer Physics Communications\, 180\, 2175-2196 (20
 09)
LOCATION:BCH 2103 https://plan.epfl.ch/?room==BCH%202103
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR