Transport
in Physics, Biology and Urban traffic
CIRM, Marseille, France
July 18 - August 26, 2022
Research projects
List of the projects
Modèles cinétiques pour les plasmas de bords
Supervisors: Mehdi Badsi, Anaïs Crestetto, Nicolas Crouseilles, Michel Mehrenberger
Team: Valentin Ayot, Averil Prost, Christian Tayou Fotso
L'interaction d'un plasma avec un bord matériel est un problème très étudié en physique. Lorsqu'un plasma
interagit avec une paroi, une couche mince (appelée "gaine de Debye") se forme près du bord due au déplacement
rapide des électrons (plus léger que les ions) qui sont absorbés par la paroi, créant un déséquilibre de charge près
du bord et ainsi la formation d'une couche limite pour le potentiel électrique solution de l'équation de Poisson.
La modélisation de ces phénomènes requiert la prise en compte d'équations dites cinétiques pour chaque espèce
(électrons et ions). Dans ce cadre, une théorie des états d'équilibre a été établie récemment dans le cas collisionnel
[1] et une exploration numérique du cas dynamique a été effectuée dans [2]. L'objectif de ce projet est d'étendre
ces travaux récents en prenant en compte les effets collisionnels ou de type source volumique au modèle double
espèce (ions et électrons) de Vlasov-Poisson, ce qui s'avère nécessaire pour modéliser les cas pertinents d'un point
de vue de la physique.
Le coeur du projet sera de construire et d'analyser des méthodes numériques adaptées pour étudier la dynamique
de la formation de gaine en prenant en compte les effets collisionnels.
[1] M. Badsi, M. Campos-Pinto, B. Després, A minimization formulation of a bi-kinetic sheath, 2016, Kinetic
Related Models.
[2] M. Badsi, M. Merhenberger, L. Navoret, Numerical stability of plasma sheath, 2018, ESAIM Proc.
Modelling, analysis, and simulation of traffic jam
in colonies of Myxococcus xanthus
Supervisors: Vincent Calvez (ICJ, Lyon), Sylvain Faure (LMO, Orsay), Loïc Gouarin (CMAP, Palaiseau),
Aline Lefebvre (CMAP, Palaiseau), Tâm Mignot (LCB, Marseille)
Team: Hélène Bloch, Michèle Romanos, Jean-Baptiste Saulnier, Benoît Gaudeul
Myxococcus xanthus is a social bacterium that experiences a fascinating life cycle, with a transi-
tion from uni-cellular to multi-cellular stages. One step of this developmental transition has raised
great modelling efforts, the so-called rippling phase, when the population of bacteria synchronizes
in accordion waves with high activity, but very limited net motion of each individual (see, e.g.
[2, 3] and more references therein). This fascinating pattern is results from repeated reversions of
individual bacteria embedded in a dense population. The reversion events are known to be finely
tuned by biochemical processes at the intra-cellular level [1]. However, the onset of collective
motion is still the subject of current investigation.
The goal of the project is to revisit these previous modelling studies based on high-resolution
data of collective motion of bacteria [4]. More precisely, we aim to investigate the relationship
between modulation of cell reversion and local congestion among the population, following [5]
with active particles.
[1] M. Guzzo et al, A gated relaxation oscillator mediated by FrzX controls morphogenetic movements in Myxococcus xanthus. Nature Microbiology, 2018.
[2] O. Igoshin et al, Waves and aggregation patterns in myxobacteria. PNAS, 2004.
[3] A. Manhart, Counter-propagating wave patterns in a swarm model with memory. J. Math. Biol., 2019.
[4] S. Panigrahi et al, Misic, a general deep learning-based method for the high-throughput cell
segmentation of complex bacterial communities. eLife, 2021.
[5] SA. Seguin et al, Clustering and flow around a sphere moving into a grain cloud. Eur. Phys. J. E, 2016.
Collective motion of biofilms
Supervisors: Vincent Calvez (ICJ, Lyon), Florence Hubert (I2M, Marseille), Julien Olivier (I2M, Marseille)
and Magali Tournus (I2M, Marseille)
Team: Maxime Estavoyer, Adil El Abdouni, Ignacio Madrid
Collective organisms in nature are capable of a wide range of motility patterns to find food,
reproduce or avoid predation. Remarkably, multicellular systems can also display collective move-
ment to achieve similar goals. Myxococcus xanthus, a soil predatory bacterium, assembles multi-
cellular biofilms to collectively predate on other microorganisms. During the hunting phase, the
bacterial biofilm is composed with isolated bacteria and largers clusters of various sizes. In her
PhD thesis, S. Rombouts conducted experiments and analysis about various collective be-
haviours occurring during predation. One of the main outcome of her study is the evidence of
synergistic effects when various motility apparatus are expressed within the cell population, with
the consequence of increasing the rate of predation.
The goal of the project is to explore the fascinating biological issues raised in [4] from a
mathematical point of view.
Hybrid semi-Lagrangian and skew-symmetric
discretization of the gyrokinetic equations
Supervisors : Martin Campos Pinto, Michel Mehrenberger, Eric Sonnendrücker.
Team: Dominik Bell, Xue Hong, Davor Kumozec, Frederik Schnack
Méthode de tenseurs pour la résolution de modèles de jeux à champ moyen.
Supervisors : Virginie Erlacher, Luca Nenna.
Team: Laila Baroukh, Damien Prel, Léopold Trémant
Les modèles de jeux à champs moyen sont utilisés dans de nombreux contextes,
en particulier pour la simulation de mouvement de foules. Dans ce type de modèle,
l'objectif est de déterminer m(t,x) la densité d'agents en un point de l'espace
x à un instant de temps t. Cependant, la résolution numérique de ce type de modèles
est extrêmement coûteuse, voire impossible, d'un point de vue numérique car elle
nécessite de résoudre des problèmes globaux espace-temps. L'objectif de ce projet
est d'investiguer le potentiel des méthodes dites de faible rang (ou méthodes de tenseurs)
pour contourner ce problème dans le cas où le nombre de degrés de liberté du système
considéré est trop grand pour que des méthodes numériques classiques soient envisageables.
Les approches de faible rang ont été utilisées avec succès pour la résolution de nombreux
problèmes en grande dimension, mais leur applicabilité pour la résolution de modèles de jeux
à champs moyens n'a à ce jour pas été explorée. Ces dernières se basent sur le principe de
la séparation de variables et consistent à rechercher une approximation de la fonction
m(t,x) sous la forme \sum_{k=1}^r r_k(t) s_k(x) où le nombre de termes r (appelé rang)
est aussi petit que possible, et où les fonctions r_k et s_k ne dépendent que de la
variable temporelle ou de la variable spatiale respectivement.
Réduction ordre en vitesse et apprentissage pour les équations de Vlasov-Maxwell
Supervisors : Emmanuel Franck, Laurent Navoret
Team: Ibtissem Lannabi, Youssouf Nasseri, Giuseppe Parasiliti Rantone, Guillaume Steimer
Etude de l’apport de schémas volumes finis composites d’ordre élevé prenant
en compte les termes sources.
Supervisors: Emmanuel Franck, Philippe Hoch, Clément Lasuen.
Team: Mohamed Boujoudar, Yoann Le Hénaff, Paul Paragot
Schémas cinétiques : aspects théoriques et optimisations informatiques
Supervisors: Philippe Helluy
Team: Clément Flint, Kévin Guillon, Romane Hélie
Modeling Compartmentalization within Intracellular Signaling Pathway
Supervisors: Erwan Hingant, Béatrie Laroche, Juan Calvo Yagüe, Romain Yvinec
Team: Claire Alamichel, Nathan Quiblier, Saoussen Latrach
Recent biological imaging have shown that membrane receptors involved in intracellular
signaling pathways are capable of inducing cascades of biochemical reactions from extremely
dynamic pools of intracellular compartments. The subcellular trafficking of receptors generates
spatio-temporal heterogeneous cellular compartments with a critical role on physiological
functions, and profound consequences on the search for new therapeutic strategies.
In this project, we aim to develop new modeling formalism, combining biochemical reaction
networks, coagulation-fragmentation and advection-diffusion dynamics to fully represent the
complexity of signaling pathways. We will design efficient numerical schemes to explore the
role of spatio-temporal heterogeneity of intracellular compartments in the response of cells
to extracellular signals.
Numerical methods for mean field optimal transport
Supervisors : Mathieu Lauriere
Team: Sebastian Baudelet, Brieuc Fresnais, Amal Machtalay
Mean field games correspond to the limit of finite population games when
the number of players grows to infinity. In the setting introduced by
Lasry and Lions [1], and Caines, Huang and Malhammé [2] around 2006,
one generally looks for a Nash equilibrium. Alternatively, one can look
for a social optimum, which corresponds to the situation where the players
cooperate in order to optimize a joint objective [3,4]. When the objective
is formulated with a terminal constraint on the population distribution instead
of a terminal cost, the problem can be viewed as a generalization of standard
optimal transport, in which the dynamics and the running cost involve mean
field interactions. An example is the so-called mean field Shrödinger problem [5].
The goal of this project is to develop efficient numerical methods for such
problems, based either on traditional tools such as finite difference schemes,
or on machine learning tools such as deep neural networks.
[1] Lasry, J. M., & Lions, P. L. (2007). Mean field games. Japanese journal of mathematics, 2(1), 229-260.
[2] Huang, M., Malhamé, R. P., & Caines, P. E. (2006). Large population stochastic dynamic games: closed-loop McKean-Vlasov systems and the Nash certainty equivalence principle. Communications in Information & Systems, 6(3), 221-252.
[3] Bensoussan, A., Frehse, J., & Yam, P. (2013). Mean field games and mean field type control theory (Vol. 101). New York: Springer.
[4] Carmona, R., Delarue, F., & Lachapelle, A. (2013). Control of McKean–Vlasov dynamics versus mean field games. Mathematics and Financial Economics, 7(2), 131-166.
[5] Backhoff, J., Conforti, G., Gentil, I., & Léonard, C. (2020). The mean field Schrödinger problem: ergodic behavior, entropy estimates and functional inequalities. Probability Theory and Related Fields, 178(1), 475-530.
Advanced numerical schemes for the simulation of complex ecosystems
Supervisors: Bastien Polizzi, Sébastian Minjeaud, Olivier Bernard, Thierry Goudon
Team: Mickaël Bestard, Léo Meyer, Florent Noisette
The goal of this project is to implement new schemes with unknowns stored
on staggered grids, for the simulation of mixture flows arising in the
modeling of biofilms formation. These models involve PDE coupled to constraints.
In order to handle these constraints with accuracy, the construction of the numerical scheme should
be consistent with the various equivalent formulations of the constraint.
The staggered framework turns out to be well adapted to this purpose.
This will allow us to test the impact of several constitutive laws on the
behavior of the solutions, and to discuss their biological interpretation.
Planning of the presentations
Wednesday 24 August
14h: Advanced numerical schemes for the simulation of complex ecosystems, Mickaël Bestard, Léo Meyer, Florent Noisette
14h40: Modelling, analysis, and simulation of traffic jam in colonies of Myxococcus xanthus, Hélène Bloch, Michèle Romanos, Jean-Baptiste Saulnier, Benoît Gaudeul
15h20: break
15h40: Kinetic schemes: theoretical aspects and optimization, Clément Flint, Kévin Guillon, Romane Hélie
16h20: Modeling Compartmentalization within Intracellular Signaling Pathway, Claire Alamichel, Nathan Quiblier, Saoussen Latrach
Thursday 25 August
9h: Collective motion of biofilms, Maxime Estavoyer, Adil El Abdouni, Ignacio Madrid
9h40: A reduced model for the Vlasov-Poisson Fokker-Planck model, Ibtissem Lannabi, Youssouf Nasseri, Giuseppe Parasiliti Rantone, Guillaume Steimer
10h20: break
10h40: Hybrid semi-Lagrangian and skew-symmetric discretization of the gyrokinetic equations, Dominik Bell, Xue Hong, Davor Kumozec, Frederik Schnack
11h20: Tensor method for solving mean field game model, Laila Baroukh, Damien Prel, Léopold Trémant
14h: Numerical methods for mean field optimal transport, Sebastian Baudelet, Brieuc Fresnais, Amal Machtalay
14h40: Study of the contribution of high order composite finite volume schemes taking into account the source terms, Mohamed Boujoudar, Yoann Le Hénaff, Paul Paragot
15h20: Kinetic methods for edge plasma, Valentin Ayot, Averil Prost, Christian Tayou Fotso