Partenaires

CNRS
UPS



Rechercher

Sur ce site

Sur le Web du CNRS


Accueil du site > Séminaires > Séminaires 2019 > Adventures in electrolytes : from nonergodicity in emergent ones
to rejection-free Monte Carlo in real ones

Lundi 16 décembre 2019 - 14:00

Adventures in electrolytes : from nonergodicity in emergent ones
to rejection-free Monte Carlo in real ones

Michael Faulkner (Université de Bristol) ¡ Attn. créneau inhabituel !

par Revaz Ramazashvili - 16 décembre 2019

Electrolytes appear in many contexts in cutting-edge physics. From real electrolytes in soft matter, such as proteins in biological cells and ionic fluids in batteries and carbon nanotubes, to emergent electrolytes in hard condensed matter, such as the frustrated three-dimensional magnet spin ice, and thin-film and layered magnets, superfluids and superconductors.

But despite their myriad applications and many years of study, we still don’t fully understand electrolytes, and usually turn to the saviour of condensed-matter physics – numerical simulation – to analyse them. In this talk, I will cover two modern methods of electrolyte simulation : electric-field Monte Carlo of the two-dimensional lattice electrolyte and rejection-free, all-atom Monte Carlo of three-dimensional liquid electrolytes.

For the two-dimensional case, I will present a lattice electric-field representation of a two-dimensional condensate-phase model of thin-film and layered superfluids and superconductors [1]. Simulation of this emergent electrolyte led to our theory of nonergodicity in the low-temperature Berezinskii-Kosterlitz-Thouless phase [2], which was subsequently used by experimentalists to explain nonergodic, long-timescale correlations in the electrical resistance of underdoped lanthanum strontium copper oxide [3], a layered superconducting material.

Following this, I will present our new rejection-free Monte Carlo algorithm [4] and accompanying Python application – JeLLyFysh – for atomistic liquid-electrolyte simulation [5]. This package achieves machine precision with a computational complexity that we predict to be at least as favourable as that of molecular dynamics. This is because our event-chain Monte Carlo algorithm is based on deterministic Markov processes, which have resulted in much faster mixing times for both one-dimensional test cases [6] and its current applications [7,8], and which advance a single particle with O(1) computations, even for systems with long-range interactions [4].

[1] Faulkner, Bramwell and Holdsworth, J. Phys. : Condens. Matter 29, 085402 (2017)

[2] Faulkner, Bramwell and Holdsworth, Phys. Rev. B 91, 155412 (2015)

[3] Shi, Shi and Popovic, Phys. Rev. B 94, 134503 (2016)

[4] Faulkner, Qin, Maggs, Krauth, J. Chem. Phys. 149, 064113 (2018)

[5] Hoellmer, Qin, Faulkner, Maggs, Krauth, arXiv:1907.12502 (2019)

[6] Lei and Krauth, Europhys. Lett. 124, 20003 (2018)

[7] Bernard and Krauth, Phys. Rev. Lett. 107, 155704 (2011)

[8] Michel, Mayer and Krauth, Europhys. Lett. 112, 20003 (2015)

Post-scriptum :

contact : M. Manghi, N. Destainville