Guerre en Iran: le sismographe des anticipations énergétiques


La guerre en Iran ne se déroule pas uniquement au Moyen‑Orient. Elle résonne dans les entrepôts d’Europe, dans les factures de gaz des ménages, dans les coûts des engrais qui nourrissent les champs, et jusque dans le prix du panier de courses. Ce qui se joue n’est pas seulement une rupture des approvisionnements en pétrole et en gaz, mais une fracture plus discrète et plus profonde: celle des anticipations. La crise iranienne agit comme un sismographe qui capte, bien avant la secousse physique, les tremblements de l’offre future d’énergie.

Avec Stéphane Auray pour Telos EU, le 09/04/2026

La guerre en Iran : un choc d’anticipations qui recompose la carte mondiale de l’énergie


Choc des prix ou choc pétrolier ? Les responsables politiques hésitent à utiliser un terme anxiogène qui renvoie aux deux crises des années 1970. Nous vivons en tous cas déjà une “crise des anticipations” qui entraîne les prix à la hausse et qui aura de puissants effets sur nos économies à peine remises des précédents chocs de la guerre contre l’Ukraine et des guerres commerciales voulues par le Président américain.

Par Stéphane Auray, le 01/04/2026.

March 8: Celebrating the Work of Women Researchers an PhD students at CREST


INTERNATIONAL_WOMEN_DAY_2026

From climate economics to machine learning and public policy, women researchers and PhD students at CREST contribute every day to advancing knowledge and addressing major societal challenges.

Every year on March 8, International Women’s Rights Day highlights the achievements of women across society and offers an opportunity to reflect on progress toward gender equality. In academia, women researchers play a crucial role in advancing knowledge, even though they remain underrepresented in several disciplines and at senior levels.

At CREST, women researchers and PhD students contribute actively to research in economics, sociology, finance and statistics. Through their work, they help strengthen the laboratory’s international visibility while addressing some of the major economic and social challenges of our time.

Research shedding light on societal transformations

Many women researchers at CREST conduct research that directly informs debates on public policy and social change.

For example, Marion Leroutier has studied differences in carbon footprints between men and women. Her research shows that women emit on average around 26% less CO₂ than men, a gap largely explained by differences in consumption patterns, particularly in transportation and food. This work contributes to discussions on environmental policy and highlights how gendered behaviors can influence the distributional effects of climate policies.

Research conducted at CREST also addresses broader demographic and economic transformations. Studies on the decline in fertility in France and other developed countries, by Pauline Rossi in her recent book, examine the economic and social implications of demographic change, a topic that has become increasingly central to public debate.

Recognition of women researchers at CREST

The work of women researchers at CREST is regularly recognized by the international academic community.

In 2026, Yuki Tamura received the Young Female Researchers Award from the Japanese Economic Association, highlighting the growing international visibility of her research.

In statistics and machine learning, Anna Korba was awarded a prestigious ERC Starting Grant for her project OptInfinite, which aims to develop new optimization and machine learning methods in infinite-dimensional settings.

These distinctions illustrate the diversity and excellence of research conducted by women scientists at CREST across multiple disciplines.

Women contributing to economic policy and public debate

Women researchers at CREST are also actively involved in shaping economic debate and public policy.

For instance, Emmanuelle Taugourdeau and Pauline Rossi were recently appointed members of the Conseil d’analyse économique, an advisory body that provides economic expertise to the French government. Their work contributes to bringing rigorous economic analysis into policy discussions.

CREST researchers also regularly participate in public debates through media appearances and collaborations with institutions, helping to disseminate research findings beyond academia.

The role of doctoral researchers

Doctoral students are another essential component of the laboratory’s research ecosystem. Many PhD students at CREST contribute to innovative research projects in areas such as labor markets, inequality, discrimination, data science and financial economics.

In the 2025-2026 cohort, nearly half of the newly recruited PhD students are women (12 women and 15 men), reflecting CREST’s ability to attract talented young female researchers. This dynamic is an important aspect of the laboratory’s commitment to promoting diversity in academic careers, not only at the doctoral level but throughout all stages of recruitment.

Looking ahead

International Women’s Rights Day is both a moment to celebrate achievements and a reminder of the importance of continuing to promote diversity in research.

At CREST, the work of women researchers and doctoral students contributes every day to advancing knowledge and to improving our understanding of economic and social transformations.

On this March 8, we celebrate their contributions to the laboratory’s scientific life and to the broader academic community.


Scientific excellence depends on the diversity of perspectives within the research community. Supporting the careers of women researchers and doctoral students is therefore not only a matter of equality, but also a key condition for the vitality and impact of research. At CREST, we are committed to fostering an academic environment where talent can thrive at every stage of the scientific career.

2025 CREST Highlights


As 2025 draws to a close, CREST reflects on a year marked by outstanding research achievements, prestigious recognitions, and impactful initiatives. Below is an overview of the laboratory’s key highlights.

This year was notably marked by CREST’s evaluation by HCERES, which recognised the laboratory as an excellent multidisciplinary research centre, particularly for the quality of its publications and its strong international visibility.

In 2025, CREST also organised its first retreat, a collective event dedicated to reflecting on the future of the laboratory and initiating concrete measures to improve laboratory life, internal communication, and the consideration of key issues such as research organisation, environmental policy, and international inclusion.

Key figures

  • Researchers | 112
  • PhD candidates | 100 PhD currently enrolled at CREST-Institut Polytechnique de Paris
  • ERC Grants | 9 ERC grants in Economics, Sociology, and Statistics

Doctoral training

In 2025, CREST organised a series of doctoral courses delivered by professors from leading universities (MIT, Oxford, University of Tokyo, among others), offering advanced training to ENSAE Paris students and, in particular, to CREST PhD candidates. These courses enabled doctoral researchers to deepen their theoretical and methodological skills while engaging with cutting-edge research developed in academic environments beyond CREST.

20 PhD candidates graduated from CREST-IP Paris in 2025, pursuing a wide range of high-level career paths across academia, the public sector, and industry. Several alumni secured academic positions as Assistant Professors at institutions such as LMU Munich, ETH Zurich, and Hitotsubashi University. Other joined key public institutions, including a position as Head of the Families Study Section at INSEE, or transitioned to the private sector as Applied Scientist at Amazon and Economist at Malt. These placements reflect CREST’s strong commitment to doctoral training and its ability to prepare PhD graduates for impactful careers across sectors.

Research Breakthroughs: 204 Publications

In 2025, CREST published 204 scientific contributions, including conference papers presented at major international venues. Nearly 80% of these publications appeared in top Q1 journals, reflecting the breadth and depth of research conducted across the laboratory’s clusters.

Selected highlights include:

The Negligible Effect of Free Contraception on Fertility: Experimental Evidence from Burkina Faso, Pascaline Dupas, Seema Jayachandran, Adriana Lleras-Muney, Pauline Rossi, American Economic Review

From Public Labs to Private Firms: Magnitude and Channels of Local R&D Spillovers, Antonin Bergeaud, Arthur Guillouzouic, Emeric Henry, Clément Malgouyres, Quarterly Journal of Economics

The Biodiversity Premium, Guillaume Coqueret, Thomas Giroux, Olivier-David Zerbib, Ecological Economics

Propagation of a Carbon Price in a Credit Portfolio through Macroeconomic Factors, Géraldine Bouveret, Jean-François Chassagneux, Smail Ibbou, Antoine J. Jacquier, Lionel Sopgoui, SIAM Journal on Financial Mathematics

Machine Bias: How Do Generative Language Models Answer Opinion Polls?, Julien Boelaert, Samuel Coavoux, Etienne Ollion, Ivaylo Petev, Patrick Präg, Sociological Methods & Research

Beyond Indices: Profiles of Social Vulnerability Gap in Disaster Risk Perception, Eric Tate, Samuel Rufat, Md Asif Rahman, Shelley Hoover, International Journal of Disaster Risk Reduction

Mapping Cells trough Time and Space with MOSCOT, Dominik Klein et al., Nature

Asymptotic Equivalence of Locally Stationary Processes and Bivariate Gaussian White Noise, Cristina Butucea, Alexander Meister, Angelika Rohde, Annals of Statistics

Discover more CREST publications on our HAL webpage.

Impactful Events and Conferences

Chairs and Research Structures

In 2025, CREST welcomed two new chairs:

  • CARE | Assurabilité des Risques Emergents, led by Olivier Lopez, focusing on climate risk modelling, risk coverage and distribution mechanisms, and prevention strategies, with Allianz IARD, Institut Louis Bachelier, Fondation du Risque. More information here.
  • Hi! Paris Chair, led by Etienne Ollion, developing the Textual Politics, project on the transformation of political analysis through natural language processing. More information here.

Awards and Recognitions

CREST researchers received numerous distinctions in 2025, including:

Several CREST researchers were also appointed to key institutional roles, including the Conseil d’Analyse Economique, for Emmanuelle Taugourdeau and Pauline Rossi, Patricia Crifo was named a Senior Advisor at la Cour des Comptes. Julien Prat was named Head of the Department of Economics-Sociology at Institut Polytechnique de Paris.

Books, Projects, and Initiatives

In 2025, CREST researchers authored and contributed to major collective works and projects:

  • Handbook of Quantitative Finance, edited by Peter Tankov and Ruixing Zhang, addressing sustainability, climate risk, regulation, and sustainable financial instruments.
  • Google has again supported CREST researchers:
    • Anna Korba for her project “Optimizing Diffusion Models via Generative Bilevel Learning
    • Vianney Perchet for his project “Design, Incentivization, Optimization and (Reinforcement-)Learning of Multi-Layered Market
  • Guillaume Hollard launched OrienteExpress, a research project with Ecole polytechnique and Docaposte (Index Education) on AI for career guidance and equal opportunities.
  • Observatory of Equal Opportunities, jointly launched by IP Paris, the Ecoles Normales Supérieures, and the Institut des Politiques Publiques, with CREST researchers contributing to the analysis of inequalities in access to elite higher education.

This year also marked the third season of the Beyond the PhD video series, featuring Benoît Schmutz-Bloch, Caroline Hillairet, Paola Tubaro, and Nicolas Chopin, who shared their thoughts on the PhD journey, its impact on their careers, and the broader role of research in society.

Media and Outreach

In 2025, CREST researchers were featured in:

  • More than 100 media outlets, including, Le Monde, Les Echos, France Culture, Libération, and La Tribune.
  • 24+ op-eds and expert articles contributing to public debate.

Featured interview: Béatrice Cherrier discusses the 2025 Nobel Prize in Economics. Listen here.

International Scientific Exchange

In 2025, CREST organised 228 research seminars across macroeconomics, applied microeconomics, sociology, finance and financial econometrics, quantitative sustainable economics and finance, statistics, actuarial science, mathematical finance, and AI for social sciences. Researchers from 20 countries were invited, representing institutions across North America, Europe, and Asia, including UCLA, MIT, Yale, Columbia, Northwestern, Oxford, LSE, ETH Zurich, Bocconi, HEC Lausanne, KU Leuven, LMU, Toronto, Waseda, and Osaka University. This extensive programme further reinforced CREST’s role as a central hub for international research dialogue.

CREST celebrated a year of remarkable achievements and meaningful contributions to research, society, and global debates. From groundbreaking publications to prestigious awards and high-level scientific exchanges, CREST community continues to push boundaries and foster innovation.

Looking ahead, CREST remains committed to advancing interdisciplinary research, addressing major societal challenges, and nurturing a collaborative and inclusive environment for researchers and students alike.

Marion Goussé : comprendre et mesurer la discrimination à l’embauche liée au handicap


Chercheuse au CREST et professeure à l’ENSAI, Marion Goussé consacre ses travaux aux inégalités sur le marché du travail, en particulier aux discriminations à l’embauche.
Avec Naomie Mahmoudi, elle a récemment mené pour APF France handicap une étude inédite sur les effets du handicap dans le processus de recrutement.
Basée sur une opération de testing à grande échelle, cette recherche met en évidence l’ampleur persistante des biais de sélection auxquels font face les candidats handicapés.
Les résultats, présentés dans plusieurs médias nationaux ont contribué à nourrir le débat public sur l’inclusion et l’égalité des chances.

Une étude rigoureuse, des constats sans appel

Pour évaluer la réalité de ces discriminations, les chercheuses ont envoyé 2000 candidatures fictives en réponse à des offres d’emploi réelles, ne différant que par la mention d’un handicap moteur ou auditif. Le handicap est parfois révélé dans une lettre de motivation et parfois révélé dans un CV Vidéo.
Résultat : à compétences égales, les personnes handicapées sont rappelées significativement moins souvent que les autres. Le handicap moteur diminue de 23% le taux de rappel, tandis que le cumul d’un handicap moteur et d’une déficience auditive le diminue de 36%.

L’analyse des réponses aux candidatures avec CV Vidéo est édifiante. Sur la vidéo, le handicap de la candidate est apparent. Impossible de passer à côté. Les chercheuses constatent que lorsque les vidéos sont vues, le taux de rappel est trois fois plus faible si la candidate est en fauteuil roulant (25% de taux de rappel) que si elle ne l’est pas (75% de taux de rappel).

L’écart se creuse particulièrement pour les métiers en contact avec le public, comme celui de secrétaire réceptionniste, alors qu’il reste plus faible pour les postes administratifs.
La mention d’une RQTH n’annule pas ces écarts, montrant que la discrimination ne tient pas à la forme de la candidature mais bien à la perception du handicap.

« La discrimination à l’embauche reste forte, même quand le handicap n’a aucun lien avec les compétences professionnelles », Marion Goussé sur France Inter.

Des leviers d’action à renforcer

Les résultats invitent à agir sur plusieurs fronts :
• Mieux former les recruteurs à la diversité et aux stéréotypes ;
• Rendre les environnements de travail plus accessibles, ce qui semble réduire les biais ;
• Encourager la transparence et le suivi statistique des pratiques de recrutement.

Ces conclusions alimentent une réflexion plus large sur la manière de garantir une égalité réelle d’accès à l’emploi.
Elles rappellent qu’au-delà des politiques publiques, l’inclusion repose aussi sur des pratiques concrètes dans les entreprises et sur une évolution des mentalités.

Pour aller plus loin

19 CREST Papers Accepted at NeurIPS 2025


This year again, CREST research team have made a strong mark at one of the world’s most prestigious conferences in artificial intelligence and machine learning: NeurIPS 2025.

A total of 19 papers by CREST members have been accepted, reflecting the lab’s growing influence in the global research community and the vitality of its teams in statistics, finance-insurance, and data science.

The accepted papers showcase the breadth and depth of the work carried out within the CREST.

Several contributions advance the fast-evolving field of diffusion and generative models, while others deepen the theoretical understanding of optimal transport and Wasserstein-based methods, approaches that connect CREST’s expertise in statistics and quantitative finance.

Other works explore kernel and inference methods, bandit algorithms, or privacy-preserving learning, illustrating the CREST’s commitment to developing both fundamental and responsible AI.

Beyond their diversity, these projects share a common spirit: a blend of mathematical rigor, computational innovation, and interdisciplinary collaboration. Together, they embody CREST’s ongoing mission to push the boundaries of what data-driven research can achieve, from theory to real-world impact.

Authors: Louis Allain, Sébastien Da Veiga, Brian Staber

Abstract: Conformal Prediction (CP) is a popular framework for constructing prediction bands with valid coverage in finite samples, while being free of any distributional assumption. A well-known limitation of conformal prediction is the lack of adaptivity, although several works introduced practically efficient alternate procedures. In this work, we build upon recent ideas that rely on recasting the CP problem as a statistical learning problem, directly targeting coverage and adaptivity. This statistical learning problem is based on reproducible kernel Hilbert spaces (RKHS) and kernel sum-of-squares (SoS) methods. First, we extend previous results with a general representer theorem and exhibit the dual formulation of the learning problem. Crucially, such dual formulation can be solved efficiently by accelerated gradient methods with several hundreds or thousands of samples, unlike previous strategies based on off-the-shelf semidefinite programming algorithms. Second, we introduce a new hyperparameter tuning strategy tailored specifically to target adaptivity through bounds on test-conditional coverage. This strategy, based on the Hilbert-Schmidt Independence Criterion (HSIC), is introduced here to tune kernel lengthscales in our framework, but has broader applicability since it could be used in any CP algorithm where the score function is learned. Finally, extensive experiments are conducted to show how our method compares to related work. All figures can be reproduced with the accompanying code.

Link: https://doi.org/10.48550/arXiv.2505.21039

Authors: Marguerite Petit–Talamon, Marc Lambert, anna Korba

Abstract:Variational inference (VI) is a popular approach in Bayesian inference, that looks for the best approximation of the posterior distribution within a parametric family, minimizing a loss that is typically the (reverse) Kullback-Leibler (KL) divergence. In this paper, we focus on the following parametric family: mixtures of isotropic Gaussians (i.e., with diagonal covariance matrices proportional to the identity) and uniform weights. We develop a variational framework and provide efficient algorithms suited for this family. In contrast with mixtures of Gaussian with generic covariance matrices, this choice presents a balance between accurate approximations of multimodal Bayesian posteriors, while being memory and computationally efficient. Our algorithms implement gradient descent on the location of the mixture components (the modes of the Gaussians), and either (an entropic) Mirror or Bures descent on their variance parameters. We illustrate the performance of our algorithms on numerical experiments.

Link: https://doi.org/10.48550/arXiv.2506.13613

Authors: Touqeer Ahmad, Mohammadreza Mousavi Kalan, François Portier, Gilles Stupfler

Abstract: Oversampling synthetic minority examples using and its variants is a leading strategy for addressing imbalanced classification problems. Despite the success of this approach in practice, its theoretical foundations remain underexplored. We develop a theoretical framework to analyze the behavior of and related methods when classifiers are trained on synthetic data. First, we establish an exponential inequality that characterizes the gap between the empirical risk computed on synthetic samples and the true population risk on the minority class. Second, we show that a kernel-based classification rule trained on synthetic data can achieve the minimax rate of convergence. This leads to practical guidelines for better parameter tuning of both and the downstream learning algorithm. Numerical experiments are provided to illustrate and support the theoretical findings.

Link: https://neurips.cc/virtual/2025/poster/117246

Authors: Vahan Arsenyan, Elen Vardanyan, Arnak Dalalyan

Abstract: Generative modeling aims to produce new random examples from an unknown target distribution, given access to a finite collection of examples. Among the leading approaches, denoising diffusion probabilistic models (DDPMs) construct such examples by mapping a Brownian motion via a diffusion process driven by an estimated score function. In this work, we first provide empirical evidence that DDPMs are robust to constant-variance noise in the score evaluations. We then establish finite-sample guarantees in Wasserstein-2 distance that exhibit two key features: (i) they characterize and quantify the robustness of DDPMs to noisy score estimates, and (ii) they achieve faster convergence rates than previously known results. Furthermore, we observe that the obtained rates match those known in the Gaussian case, implying their optimality.

Link: https://doi.org/10.48550/arXiv.2506.09681

Authors: Marius Potfer, Vianney Perchet

Abstract: Repeated multi-unit auctions, where a seller allocates multiple identical items over many rounds, are common mechanisms in electricity markets and treasury auctions. We compare the two predominant formats: uniform-price and discriminatory auctions, focusing on the perspective of a single bidder learning to bid against stochastic adversaries. We characterize the learning difficulty in each format, showing that the regret scales similarly for both auction formats under both full-information and bandit feedback, as and , respectively. However, analysis beyond worst-case regret reveals structural differences: uniform-price auctions may admit faster learning rates, with regret scaling as   in settings where discriminatory auctions remain at . Finally, we provide a specific analysis for auctions in which the other participants are symmetric and have unit-demand, and show that in these instances a similar regret rate separation appears.

Link: https://neurips.cc/virtual/2025/poster/115489

Authors: Georgios Gavrilopoulos, Guillaume Lecué, Zong Shang

Abstract: We obtain upper bounds for the estimation error of Kernel Ridge Regression (KRR) for all non-negative regularization parameters, offering a geometric perspective on various phenomena in KRR. As applications: 1. We address the multiple descent problem, unifying the proofs of arXiv:1908.10292 and arXiv:1904.12191 for polynomial kernels and we establish multiple descent for the upper bound of estimation error of KRR under sub-Gaussian design and non-asymptotic regimes. 2. For a sub-Gaussian design vector and for non-asymptotic scenario, we prove a one-sided isomorphic version of the Gaussian Equivalent Conjecture. 3. We offer a novel perspective on the linearization of kernel matrices of non-linear kernel, extending it to the power regime for polynomial kernels. 4. Our theory is applicable to data-dependent kernels, providing a convenient and accurate tool for the feature learning regime in deep learning theory. 5. Our theory extends the results in arXiv:2009.14286 under weak moment assumption.
Our proof is based on three mathematical tools developed in this paper that can be of independent interest: 1. Dvoretzky-Milman theorem for ellipsoids under (very) weak moment assumptions. 2. Restricted Isomorphic Property in Reproducing Kernel Hilbert Spaces with embedding index conditions. 3. A concentration inequality for finite-degree polynomial kernel functions.

Link: https://doi.org/10.48550/arXiv.2404.07709

Authors: Yvann Le Fay, Nicolas Chopin, Simon Barthelmé

Abstract: Variational inference consists in finding the best approximation of a target distribution within a certain family, where `best’ means (typically) smallest Kullback-Leiber divergence. We show that, when the approximation family is exponential, the best approximation is the solution of a fixed-point equation. We introduce LSVI (Least-Squares Variational Inference), a Monte Carlo variant of the corresponding fixed-point recursion, where each iteration boils down to ordinary least squares regression and does not require computing gradients. We show that LSVI is equivalent to stochastic mirror descent; we use this insight to derive convergence guarantees. We introduce various ideas to improve LSVI further when the approximation family is Gaussian, leading to a O(d³) complexity in the dimension d of the target in the full-covariance case, and a O(d)complexity in the mean-field case. We show that LSVI outperforms state-of-the-art methods in a range of examples, while remaining gradient-free, that is, it does not require computing gradients.

Link: https://doi.org/10.48550/arXiv.2502.18475

Authors: Luca Arnaboldi, Bruno Loureiro, Ludovic Stephan, Florent Krzakala, Lenka Zdeborova

Abstract: We study the dynamics of stochastic gradient descent (SGD) for a class of sequence models termed Sequence Single-Index (SSI) models, where the target depends on a single direction in input space applied to a sequence of tokens. This setting generalizes classical single-index models to the sequential domain, encompassing simplified one-layer attention architectures. We derive a closed-form expression for the population loss in terms of a pair of sufficient statistics capturing semantic and positional alignment, and characterize the induced high-dimensional SGD dynamics for these coordinates. Our analysis reveals two distinct training phases: escape from uninformative initialization and alignment with the target subspace, and demonstrates how the sequence length and positional encoding influence convergence speed and learning trajectories. These results provide a rigorous and interpretable foundation for understanding how sequential structure in data can be beneficial for learning with attention-based models.

Link: https://doi.org/10.48550/arXiv.2506.02651

Authors: Oussama Zekri, Nicolas Boullé

Abstract: Discrete diffusion models have recently gained significant attention due to their ability to process complex discrete structures for language modeling. However, fine-tuning these models with policy gradient methods, as is commonly done in Reinforcement Learning from Human Feedback (RLHF), remains a challenging task. We propose an efficient, broadly applicable, and theoretically justified policy gradient algorithm, called Score Entropy Policy Optimization (SEPO), for fine-tuning discrete diffusion models over non-differentiable rewards. Our numerical experiments across several discrete generative tasks demonstrate the scalability and efficiency of our method. Our code is available at this https URL.

Link: https://doi.org/10.48550/arXiv.2502.01384

Authors: Stanislas Strasman, Sobihan Surendran, Claire Boyer, Sylvain Le Corff, Vincent Lemaire, Antonio Ocello

Abstract: Score-based Generative Models (SGMs) have achieved impressive performance in data generation across a wide range of applications and benefit from strong theoretical guarantees. Recently, methods inspired by statistical mechanics, in particular Hamiltonian dynamics, have introduced Critically-damped Langevin Diffusions (CLD), which define diffusion processes in extended spaces by coupling the data with auxiliary variables. These approaches, along with their associated score-matching and sampling procedures, have been shown to outperform standard diffusion-based samplers numerically. In this paper, we propose an upper bound on the sampling error for CLD-based generative models in the Wasserstein metric. To better exploit the extended space, we also propose a modified dynamic that introduces an additional hyperparameter controlling the noise applied to the data coordinates. This hyperparameter influences the smoothness of sample paths, and our discretization error analysis offers practical guidance for tuning, leading to improved sampling performance.

Link: https://neurips.cc/virtual/2025/poster/117301

Authors: Nina Vesseron, Louis Béthune, Marco Cuturi

Abstract: The canonical approach in generative modeling is to split model fitting into two blocks: define first how to sample noise (e.g. Gaussian) and choose next what to do with it (e.g. using a single map or flows). We explore in this work an alternative route that ties sampling and mapping. We find inspiration in moment measures, a result that states that for any measure ρ, there exists a unique convex potential u such that ρ=ueu. While this does seem to tie effectively sampling (from log-concave distribution eu) and action (pushing particles through u), we observe on simple examples (e.g., Gaussians or 1D distributions) that this choice is ill-suited for practical tasks. We study an alternative factorization, where ρ is factorized as wew, where w is the convex conjugate of a convex potential w. We call this approach conjugate moment measures, and show far more intuitive results on these examples. Because w is the Monge map between the log-concave distribution ew and ρ, we rely on optimal transport solvers to propose an algorithm to recover w from samples of ρ, and parameterize w as an input-convex neural network. We also address the common sampling scenario in which the density of ρ is known only up to a normalizing constant, and propose an algorithm to learn w in this setting.

Link: https://doi.org/10.48550/arXiv.2503.10576

Author: Imad Aouali

Abstract: Efficient exploration is a key challenge in contextual bandits due to the large size of their action space, where uninformed exploration can result in computational and statistical inefficiencies. Fortunately, the rewards of actions are often correlated and this can be leveraged to explore them efficiently. In this work, we capture such correlations using pre-trained diffusion models; upon which we design diffusion Thompson sampling (dTS). Both theoretical and algorithmic foundations are developed for dTS, and empirical evaluation also shows its favorable performance.

Link: https://doi.org/10.48550/arXiv.2402.10028

Authors: Jung-Hun Kim, Milan Vojnovic, Min-hwan Oh

Abstract: We study the combinatorial semi-bandit problem where an agent selects a subset of base arms and receives individual feedback. While this generalizes the classical multi-armed bandit and has broad applicability, its scalability is limited by the high cost of combinatorial optimization, requiring oracle queries at *every* round. To tackle this, we propose oracle-efficient frameworks that significantly reduce oracle calls while maintaining tight regret guarantees. For worst-case linear rewards, our algorithms achieve regret using only oracle queries. We also propose covariance-adaptive algorithms that leverage noise structure for improved regret, and extend our approach to general (non-linear) rewards. Overall, our methods reduce oracle usage from linear to (doubly) logarithmic in time, with strong theoretical guarantees.

Link: https://neurips.cc/virtual/2025/poster/119751

Authors: Rémi Castera, Felipe Garrido, Patrick Loiseau, Simon Mauras, Mathieu Molina, Vianney Perchet

Abstract: We consider matroid allocation problems under opportunity fairness constraints: resources need to be allocated to a set of agents under matroid constraints (which includes classical problems such as bipartite matching). Agents are divided into C groups according to a sensitive attribute, and an allocation is opportunity-fair if each group receives the same share proportional to the maximum feasible allocation it could achieve in isolation. We study the Price of Fairness (PoF), i.e., the ratio between maximum size allocations and maximum size opportunity-fair allocations. We first provide a characterization of the PoF leveraging the underlying polymatroid structure of the allocation problem. Based on this characterization, we prove bounds on the PoF in various settings from fully adversarial (wort-case) to fully random. Notably, one of our main results considers an arbitrary matroid structure with agents randomly divided into groups. In this setting, we prove a PoF bound as a function of the size of the largest group. Our result implies that, as long as there is no dominant group (i.e., the largest group is not too large), opportunity fairness constraints do not induce any loss of social welfare (defined as the allocation size). Overall, our results give insights into which aspects of the problem’s structure affect the trade-off between opportunity fairness and social welfare.

Link: https://arxiv.org/pdf/2403.00397v2

Authors: Shiyun Lin, Simon Mauras, Nadav Merlis, Vianney Perchet

Abstract: We study the problem of matching markets with ties, where one side of the market does not necessarily have strict preferences over members at its other side. For example, workers do not always have strict preferences over jobs, students can give the same ranking for different schools and more. In particular, assume w.l.o.g. that workers’ preferences are determined by their utility from being matched to each job, which might admit ties. Notably, in contrast to classical two-sided markets with strict preferences, there is no longer a single stable matching that simultaneously maximizes the utility for all workers.
We aim to guarantee each worker the largest possible share from the utility in her best possible stable matching. We call the ratio between the worker’s best possible stable utility and its assigned utility the \emph{Optimal Stable Share} (OSS)-ratio. We first prove that distributions over stable matchings cannot guarantee an OSS-ratio that is sublinear in the number of workers. Instead, randomizing over possibly non-stable matchings, we show how to achieve a tight logarithmic OSS-ratio. Then, we analyze the case where the real utility is not necessarily known and can only be approximated. In particular, we provide an algorithm that guarantees a similar fraction of the utility compared to the best possible utility. Finally, we move to a bandit setting, where we select a matching at each round and only observe the utilities for matches we perform. We show how to utilize our results for approximate utilities to gracefully interpolate between problems without ties and problems with statistical ties (small suboptimality gaps).

Link: https://doi.org/10.48550/arXiv.2411.03270

Authors: Ziyad Benomar, Romain Cosson, Alexander Lindermayr, Jens Schöter

Abstract: In non-clairvoyant scheduling, the goal is to minimize the total job completion time without prior knowledge of individual job processing times. This classical online optimization problem has recently gained attention through the framework of learning-augmented algorithms. We introduce a natural setting in which the scheduler receives continuous feedback in the form of progress bars: estimates of the fraction of each job completed over time. We design new algorithms for both adversarial and stochastic progress bars and prove strong competitive bounds. Our results in the adversarial case surprisingly induce improved guarantees for learning-augmented scheduling with job size predictions. We also introduce a general method for combining scheduling algorithms, yielding further insights in scheduling with predictions. Finally, we propose a stochastic model of progress bars as a more optimistic alternative to conventional worst-case models, and present an asymptotically optimal scheduling algorithm in this setting.

Link: https://doi.org/10.48550/arXiv.2509.19662

Authors: Achraf Azize, Yulian Wu, Junya Honda, Francesco Orabona, Shinji Ito, Debabrota Basu

Abstract: As sequential learning algorithms are increasingly applied to real life, ensuring data privacy while maintaining their utilities emerges as a timely question. In this context, regret minimisation in stochastic bandits under ϵ-global Differential Privacy (DP) has been widely studied. Unlike bandits without DP, there is a significant gap between the best-known regret lower and upper bound in this setting, though they “match” in order. Thus, we revisit the regret lower and upper bounds of ϵ-global DP algorithms for Bernoulli bandits and improve both. First, we prove a tighter regret lower bound involving a novel information-theoretic quantity characterising the hardness of ϵ-global DP in stochastic bandits. Our lower bound strictly improves on the existing ones across all ϵ values. Then, we choose two asymptotically optimal bandit algorithms, i.e. DP-KLUCB and DP-IMED, and propose their DP versions using a unified blueprint, i.e., (a) running in arm-dependent phases, and (b) adding Laplace noise to achieve privacy. For Bernoulli bandits, we analyse the regrets of these algorithms and show that their regrets asymptotically match our lower bound up to a constant arbitrary close to 1. This refutes the conjecture that forgetting past rewards is necessary to design optimal bandit algorithms under global DP. At the core of our algorithms lies a new concentration inequality for sums of Bernoulli variables under Laplace mechanism, which is a new DP version of the Chernoff bound. This result is universally useful as the DP literature commonly treats the concentrations of Laplace noise and random variables separately, while we couple them to yield a tighter bound.

Link: https://doi.org/10.48550/arXiv.2505.05613

Authors: Marc Jourdan, Achraf Azize

Abstract: Best Arm Identification (BAI) algorithms are deployed in data-sensitive applications, such as adaptive clinical trials or user studies. Driven by the privacy concerns of these applications, we study the problem of fixed-confidence BAI under global Differential Privacy (DP) for Bernoulli distributions. While numerous asymptotically optimal BAI algorithms exist in the non-private setting, a significant gap remains between the best lower and upper bounds in the global DP setting. This work reduces this gap to a small multiplicative constant, for any privacy budget . First, we provide a tighter lower bound on the expected sample complexity of any δ-correct and -global DP strategy. Our lower bound replaces the Kullback–Leibler (KL) divergence in the transportation cost used by the non-private characteristic time with a new information-theoretic quantity that optimally trades off between the KL divergence and the Total Variation distance scaled by . Second, we introduce a stopping rule based on these transportation costs and a private estimator of the means computed using an arm-dependent geometric batching. En route to proving the correctness of our stopping rule, we derive concentration results of independent interest for the Laplace distribution and for the sum of Bernoulli and Laplace distributions. Third, we propose a Top Two sampling rule based on these transportation costs. For any budget , we show an asymptotic upper bound on its expected sample complexity that matches our lower bound to a multiplicative constant smaller than 8. Our algorithm outperforms existing δ-correct and -global DP BAI algorithms for different values of .

Link: https://openreview.net/forum?id=IFso8G8gwJ

Authors: Adrien Vacher, Omar Chehab, Anna Korba

Abstract: Sampling from multi-modal distributions is challenging, even in low dimensions. We provide the first sampling algorithm for a broad class of distributions — including all Gaussian mixtures — with a query complexity that is polynomial in the parameters governing multi-modality, assuming fixed dimension. Our sampling algorithm simulates a time-reversed diffusion process, using a specific Monte Carlo estimator of the intermediate score functions. Unlike previous works, it avoids metastability, requires no prior knowledge of the mode locations, and does not rely on restrictive smoothness assumptions that exclude general Gaussian mixtures. We illustrate this result on a low-dimensional but challenging multi-modal sampling task, where our algorithm significantly outperforms existing approaches.

Link: https://neurips.cc/virtual/2025/poster/119085

A New Academic Year, New Faces at CREST


Every year, CREST welcomes a new cohort of researchers who bring fresh perspectives, curiosity, and energy to our community. This fall, we are pleased to host new researchers, postdoctoral fellow, PhD students, and visiting researchers from around the world.

Their arrival marks an important moment in the life of our lab: the opportunity to grow, to renew dialogues across disciplines, and to continue building a vibrant and diverse research environment. Whether they join us for a few months or several years, these newcomers contribute to the intellectual richness and collaborative spirit that define CREST.

We are pleased to introduce the researchers who are joining CREST this year, each bringing their unique background and expertise to our four research areas: economics, sociology, statistics, and finance-insurance.

Researchers

This year, several new researchers are joining CREST, strengthening our core faculty and bringing new perspectives across ou disciplines.

Philippe Coulangeon is a CNRS Research Director, affiliated with the Centre for Research on Social Inequalities (CRIS) at SciencesPo Paris.

His work is set at the intersection of cultural sociology and social inequalities. He explores the stratification of cultural practices and tastes, the dynamics of mass versus elite culture, artistic professions, and the democratization of culture. More recently, he has integrated environmental concerns into his analysis.

More information here.

Fanny Landaud is a CNRS researcher from CY Cergy Paris University and she is an IZA Research Fellow since 2019.

Her research covers applied microeconomics – particularly labor, education, family, and health economics – with a focus on the determinants and consequences of socioeconomic and gender inequalities in education and the labor market.

More information here.

Antonio Ocello was a Postdoctoral Researcher in Statistics and Machine Learning at École polytechnique (CMAP) since 2023.

His research focuses on mean-field games and mean-field control problems, branching diffusion processes, and stochastic optimal control, combining advanced probabilistic methods with applications in machine learning. He has contributed to work on convergence bounds for trust-region policy optimization in mean-field games and is developing stochastic target frameworks for branching processes.

More information here.

Enrico Ruolino was a Senior Researcher at the University of Lausanne.

His work focuses on public and labor economics, with a particular emphasis on behavioral responses to tax policy, gender inequality, intergenerational mobility, and educational outcomes.

More information here.

Clémentine Van Effenterre was previously an Assistant Professor of Economics at the University of Toronto.

Her research lies at the intersection of labor economics, applied microeconomics, and political economy. Notably, she investigates how norms, institutions, and policies shape labor market outcomes – examining issues like maternal labor supply, gender gaps in science and tech, and how having daughters influences political attitudes. She also hosts the “InequaliTalks” podcast on economics and inequality.

More information here.

Postdoctoral Fellows

Our new postdoctoral researchers are contributing to ongoing projects and launching their own work in close collaboration with CREST teams.

Stéphane Lhaut, in the Finance-Insurance team, with Caroline Hillairet and Olivier Lopez.

Angelica Martinez-Leyva, in the Economics cluster at CREST-ENSAI, with Marion Goussé

Andrea Pandolfi, in Statistics, with Nicolas Chopin

Arthur Stephanovitch, in Statistics, with Austin Stromme

Lishu Zhang, in the Finance-Insurance team, with Olivier-David Zerbib

PhD Students

This year, we welcome a new cohort of PhD students who are beginning their research journeys within our vibrant academic community.

30 new PhD students are joining our laboratory in our different research clusters.

9 PhD students will join our Economic team, 7 in the Finance-Insurance cluster, 12 new PhD students in Statistics, and 2 in our Sociology team.

4 of them will join our lab on the CREST-ENSAI campus in Bruz while the remaining PhD students will join our CREST-ENSAE Paris campus in Palaiseau.

This new team will also be joined by 3 PhD students in ATER contracts in the CREST-ENSAI campus.

Visiting Researchers

CREST is also hosting several visiting researchers this year, joining us for research stays that foster international collaboration.

Yang Chen (Lausanne University, Switzerland) is joining our Economic team to work with Julien Combe on Microeconomic theory and matching theory.

Matias Ortiz (Universidad de Chile, Chile) will work with Vianney Perchet on Learning-Augmented Online Algorithms for Matroid-Constrained problems under Partial in our Statistical team.

Simon Luck (Universita di Bologna, Italy) is joining Etienne Ollion’s team in Sociology to work on using natural language processing to study the relevance of the news media for political representation and decision-making processes.

Jules Verin (ENS Lyon, France) will work with Bertrand Garbinti in Economics on Wealth Accumulation, Life Cycle, and Taxation.

Väinö Yrjänäinen (Uppsala University, Sweden) is joining the Sociology team with Etienne Ollion to work on using work embeddings and transformer models in computational social sciences.

Les Mouettes Savantes : mathématiques et informatique au service de l’écologie


Du 16 au 20 juin 2025, la deuxième édition des Mouettes Savantes a réuni 30 élèves de seconde issues de lycées de Brest, Rennes et du sud de Paris.  

Au programme de ce projet scientifique conçu et organisé par cinq chercheuses et enseignantes-chercheuses, dont Marie Etienne (ENSAI-CREST) : découverte de la recherche en mathématique au service des transitions environnementales, à la station biologique de l’Université de Rennes, en forêt de Paimpont.

Bilan carbone de l’IA générative, alimentation, évolution des températures, moustique tigre et pêche durable : les matinées ont été consacrées à des ateliers scientifiques mobilisant des compétences en mathématiques et informatique, appliquées à des enjeux environnementaux actuels. L’après-midi, place aux activités sportives, ludiques et culturelles au cœur de la forêt de Brocéliande.

Conclusion ? Un franc succès pour ce séjour dont l’objectif est de renforcer la présence des jeunes filles dans les filières et carrières scientifiques.

Marie Etienne a présenté le projet Les Mouettes Savantes lors des Journées Parité de la communauté mathématique 2025 qui se se sont tenues les 23 et 24 juin à l’Université de Rennes, campus Beaulieu.

Ouest France a couvert la dernière édition des Mouettes Savantes dans l’édition du 21 juin 2025.

En savoir plus sur le projet Les Mouettes Savantes