Session régionale du séminaire des doctorantes

vendredi 28 nov. 2025, 08:00 → 19:00 Europe/Paris

Les séminaires des docorant.e.s des instituts de mathématiques de Montpellier et Toulouse, avec le soutien de la fédération occitane de Mathématiques, s'associent pour une session commune. Cette session aura lieu le 28 novembre après-midi à l'institut Montpelliérain Alexander Grothendieck (IMAG).

 

---

The doctoral student seminars of the Montpellier and Toulouse mathematics institutes, with the support of the federation OcciMath, are joining forces for a joint session. This session will take place on the afternoon of November 28 at the institut Montpelliérain Alexander Grothendieck (IMAG).

 

Comité d'organisation / Organizing committee

• Alexandre Capel (IMAG)
• Jan-Luka Fatras (IMT),
• Luca Froger (IMAG)
• Florian Gossard (IMT),
• Adrien Kachkachi (IMT),
• Fabien Lespagnol (IMAG),
• Sofian Tur-Dorvault (IMAG).

avec le soutien de / with  Matthieu Hillairet (OcciMath)

 

 

 

13:15 → 13:30 Accueil

13:30 → 14:00 Femmes et Mathématiques : une comparaison transnationale des carrières universitaires en France et au Kazakhstan

Cette communication propose une analyse comparative des carrières universitaires en mathématiques en France et au Kazakhstan sous l’angle du genre. Dans les deux pays, les jeunes filles réussissent mieux que les garçons dans les disciplines scientifiques et poursuivent des études plus longues. Pourtant, elles restent minoritaires dans les filières universitaires de mathématiques et voient leur progression ralentir dès l’obtention d’un poste permanent. Ce paradoxe formation-carrière, observé dans des contextes nationaux très différents, invite à interroger les mécanismes sociaux et genrés qui structurent les trajectoires académiques.

À partir d’un corpus combinant données statistiques, bases de thèses, reconstitution de parcours académiques et entretiens biographiques, l’étude met en évidence trois modèles de trajectoires : l’excellence républicaine, où l’État joue un rôle moteur mais reproduit les rapports sociaux de sexe ; l’héritage scientifique, où les réseaux familiaux et professionnels favorisent la reproduction sociale et une évolution partielle du système de genre ; et l’affirmation féministe, où l’agentivité individuelle et collective des femmes contribue à transformer les rapports sociaux de genre.

Cette comparaison transnationale permet de comprendre comment se tissent égalité et inégalités dans des environnements contrastés, et d’interroger l’impact des résistances individuelles sur l’évolution du système global de genre.

Orateur: Zhanna Karimova (Centre Max Weber)

14:30 → 15:05 Caractérisation algébrique des horomorphismes

TBA

Orateur: Patrick Salhany

14:30 → 15:05 Følner sequences without group actions

Amenable groups have historically been introduced as a response to the Banach-Tarski paradox, which states that one can separate a ball in a disjoint union of subsets, rotate said subsets around and obtain two copies of the original ball. I will present this notion in this classical framework, and more precisely a common strategy to prove amenability: finding Følner sequences. This will lead us to a result of Alex Eskin, David Fisher and Kevin Whyte using Følner sequences in a slightly different context. This result is a good example on how to use Følner sequences without information on any group in order to study behaviors of areas and volumes.

Orateur: Luca Froger

15:05 → 15:40 Concentration in selection-mutation models: error estimates and asymptotic expansions

TBA

Orateur: Caroline Guinet

15:05 → 15:40 The extremal process of two-speed branching random walk

We consider a two-speed branching random walk, which consists of two macroscopic stages with different reproduction laws. We prove that the centered maximum converges in law to a Gumbel variable with a random shift and the extremal process converges in law to a randomly shifted decorated Poisson point process, which can be viewed as a discrete analog for the corresponding results for the two-speed branching Brownian motion, previously established by Bovier and Hartung(2014)

Orateur: Lianghui Luo

16:00 → 16:35 Motivation to K-theory of C* algebras via Atiyah Singer theorem and Fredholm operators

This talk will be structured as follows: the first will be quite historical, talking about operators on a Hilbert space, Fredholm operators, indices.. In the second one, we introduce the C* algebraic K-theory framework and we relate it to the problem of Fredholm indices computation. If we have enough time, we look at a specific example illustrating which kind of link is arised by Atiyah Singer theorem.

Orateur: Florian Thiry

16:00 → 16:35 Quantification of limit theorem for nearly unstable Hawkes process

Orateur: Benjamin Massat

16:35 → 17:10 Image-based multi-scale plant species monitoring: evaluating the impact of land management changes on biodiversity

Thanks to modern data infrastructures, large amounts of images describing vegetation are available at multiple scales: at the individual scale (e.g., Pl@ntNet images), at the landscape scale (e.g., LUCAS images), and at the ecosystem scale (e.g., Sentinel 2 data). Usually, these different types of visual content are processed separately. Images of individual plants are typically used to train species identification models. Images of communities are rather used for ecotope characterization. Remote sensing images are used for land monitoring, or more recently for species distribution modeling. The combination of these three scales, however, has not yet been leveraged, or only in very specific scenarios. We propose to combine those multi-scale vegetation images to derive biodiversity-relevant features using deep contrastive learning models, and incorporate these multi-scale multi-modal features in species distribution models to more accurately address species mapping at EU scale and derive biodiversity indicators at local scales for better land management. We first train our model on a pair-matching pretext task to align GPS coordinates with images in a common representation space; before evaluating it on a downstream species prediction task. We train and evaluate our method over the CBN-Med region of France, by using Sentinel-2A satellite imagery, landscape images form the EU LUCAS dataset, Pl@ntNet images and citizen science plant observations from the GeoLifeCLEF2024 dataset. The code, model and parts of the dataset are publicly available on GitHub through the Malpolon framework. We show that enriching our classification model with a contrastive pre-training task gives similar or better performances on a species prediction task compared to training a BCE loss from scratch. This opens more possibilities to training species prediction models in areas with few or no observations (few-shot learning).

Orateur: Théo Larcher

16:35 → 17:10 Univariate Stochastic Modeling of Rainfall Extremes in Montpellier

Flood risk is particularly high in urban areas due to soil impermeability, which prevents water absorption. Flooding can occur after periods of intense rainfall or during prolonged episodes of moderate rain. This is especially true in Montpellier, where heavy precipitation events frequently result in urban flooding. Modeling heavy rainfall and dry periods is essential to prevent urban flooding. In this study, we use rainfall data recorded at a one-minute resolution, which provides a much finer temporal scale than typically available in similar studies. While most approaches rely on continuous distributions, our study focuses on discrete data, motivating the use of a discrete distribution. For now, we assume temporal independence of the rainfall observations. The objective of this work is to develop a stochastic rainfall generator capable of modeling precipitation at a high temporal resolution.

Orateur: Anne Bernard

17:10 → 18:30 Session Poster + Goûter