~~NOCACHE~~

{{page>.:indexheader}}

\\ ==== Next talks ====
[[en:seminaires:SemDoc:index|Doctoral Seminar]]\\
Thursday December 4, 2025, 5:30PM, Sophie Germain - Salle 1013 (1er étage)\\
**Roland Sogan + Francesca Cottini**  //Scalable Graphon inference: A Fast and Consistent Approach for Multi-Network Data (R. Sogan) + Directed polymer in critical correlated environment (F. Cottini)//
<button type="link" size="xs" collapse="b10-933">{{icon>info-circle?lg&color=#777}}</button>\\
<collapse id="b10-933" collapsed="true">
<well size="sm">
Modern network datasets increasingly consist of multiple heterogeneous graphs with varying node sets and sizes, posing significant challenges for statistical estimation. This paper addresses the fundamental problem of graphon estimation from such collections, a core task in nonparametric modeling of exchangeable random graphs. Existing graphon estimators face a critical trade-off, since they are either computationally efficient but statistically inconsistent, or statistically consistent but prohibitively slow for practical applications. We break this trade-off by introducing a fast histogram-based estimator that leverages joint node alignment across all available networks. Our method achieves both computational efficiency and statistical consistency by synchronously organizing node information rather than processing graphs sequentially. Theoretical guarantees establish consistency under mild regularity conditions. Extensive numerical experiments demonstrate clear advantages such as superior accuracy with small, variable-sized networks, orders-of-magnitude speed improvements over competing consistent methods, and improved performance in downstream graph neural network applications thanks to more effective data augmentation.

Directed polymers in random environments describe a perturbation of the simple random walk given by a random disorder (environment). The partition functions of this model have been thoroughly investigated in recent years, also motivated by their link with the solution of the Stochastic Heat Equation. While classical results focus on space-time independent disorder, we consider a Gaussian environment with (critical) spatial correlations decaying as $|x|^{-2}$ times a slowly varying function. We show that a phase transition, analogous to that in the space-time independent case, still occurs: in the high temperature regime the log-partition function satisfies a central limit theorem, while it vanishes in law in the low temperature regime. Remarkably, the inverse temperature needs to be tuned differently from the independent case, where the scaling constant $\hat{\beta}$ emerges from a nontrivial multi-scale dependence in the second moment computation.
</well>
</collapse>
Organisation : Sacha Quayle, Thomas Le Guerch, Nina Drobac, Pierre Faugere, Maxime Guellil, Eyal Vayness

[[en:seminaires:SemDoc:index|Doctoral Seminar]]\\
Thursday December 18, 2025, 5:30PM, Jussieu - Salle Paul Lévy (16-26 209)\\
**Chloé Hashimoto-Cullen + Emmanuel Gnabeyeu**  //Non encore annoncé + Non encore annoncé//
\\
Organisation : Sacha Quayle, Thomas Le Guerch, Nina Drobac, Pierre Faugere, Maxime Guellil, Eyal Vayness

{{page>.:info}}

\\ ==== Previous talks ====
\\ === Year 2025 ===

{{page>.:semdoc2025}}

\\ === Year 2024 ===

{{page>.:semdoc2024}}

\\ === Year 2023 ===

{{page>.:semdoc2023}}

\\ === Year 2022 ===

{{page>.:semdoc2022}}

\\ === Year 2021 ===

{{page>.:semdoc2021}}

\\ === Year 2020 ===

{{page>.:semdoc2020}}

\\ === Year 2019 ===

{{page>.:semdoc2019}}

\\ === Year 2018 ===

{{page>.:semdoc2018}}