BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Neural density functional theory of liquid-gas phase coexistence a
 nd related phenomena. - Robert Evans (H.H. Wills Physics Laboratory\, Univ
 ersity of Bristol)
DTSTART:20250304T130000Z
DTEND:20250304T140000Z
UID:TALK225691@talks.cam.ac.uk
CONTACT:Sarah Loos
DESCRIPTION:Using supervised machine learning together with the rigorous c
 oncepts of classical density functional theory (DFT) we investigate the pa
 ir structure and thermodynamic properties\, including bulk liquid-gas coex
 istence and associated interfacial phenomena\, in many-body systems.\nLoca
 l learning of the one-body direct correlation functional is based on Monte
  Carlo simulations of inhomogeneous systems with randomized thermodynamic 
 conditions\, randomized planar shapes of the external potential\, and rand
 omized box sizes. Focusing on the prototypical Lennard-Jones system\, we t
 est predictions of the resulting neural DFT across a broad spectrum of phy
 sical behaviour. Specifically\, we analyse the bulk radial distribution fu
 nction g(r) obtained from automatic differentiation and the Ornstein-Zerni
 ke equation and determine : i) the Fisher-Widom line\, i.e.\, the crossove
 r of asymptotic (large distance r) decay of g(r) from monotonic to oscilla
 tory\, ii) the (Widom) line of maximal true correlation length\, iii) the 
 line of maximal isothermal compressibility and iv) the spinodal by calcula
 ting the poles of the structure factor in the complex plane. The bulk bino
 dal and the density profile of the free liquid-gas interface are obtained 
 from DFT minimization and the corresponding surface tension from functiona
 l line integration. Our neural DFT improves significantly upon standard me
 an-field treatments of interparticle attraction. It also describes accurat
 ely the phenomena of drying at a hard wall and capillary evaporation of a 
 liquid confined in a slit pore.\nComparison with independent simulation re
 sults demonstrates a consistent picture of phase separation even when rest
 ricting the training to supercritical states only. We argue that phase coe
 xistence and its associated signatures can be discovered as emerging pheno
 mena via functional mappings and educated extrapolation [1].\nExtending th
 e neural DFT to include functional dependence on the pair potential enable
 s new and powerful inversion of structural data for liquids to discover th
 e underlying interaction potential\, important in soft matter inverse desi
 gn [2].\n\n[1] F. Sammueller\, M. Schmidt & R. Evans PRX 15\, 011013-1-23 
 (2025).\n\n[2] S.F. Kampa\, F. Sammueller\, M. Schmidt & R. Evans PRL (to 
 appear).
LOCATION:Center for Mathematical Sciences\, Lecture room MR4
END:VEVENT
END:VCALENDAR