Sign in

Gabriel Peyré

@gabrielpeyre.bsky.social
2.8K followers 181 following 105 posts

CNRS researcher at ENS

PostsRepliesMedia
Gabriel Peyré @gabrielpeyre.bsky.social · 03/10/2026
Inverse optimization seeks to recover the parameters of an optimization problem from observed solutions. When the objective depends linearly on these parameters, the gap loss is convex, yielding a convex formulation. Inverse optimal transport is one example. www.gpeyre.com/blog/2026/10...
0236
Gabriel Peyré @gabrielpeyre.bsky.social · 15/09/2026
Why I love Gaussian-preserving flows www.gpeyre.com/blog/2026/09...
0173
Gabriel Peyré @gabrielpeyre.bsky.social · 15/09/2026
Clarifications sur mon intervention à France Inter au sujet de l’IA en mathématiques www.gpeyre.com/blog/2026/09... Merci à @jkobject.com de m'avoir poussé à l'écrire.
gpeyre.com
IA pour les maths : clarifications - Homepage of Gabriel Peyré
Une clarification de mon intervention sur France Inter, en français et en anglais.
2202
Gabriel Peyré @gabrielpeyre.bsky.social · 14/08/2026
My favorite P-L function www.gpeyre.com/mathematical...
1142
Gabriel Peyré @gabrielpeyre.bsky.social · 07/08/2026
The Mathematical Nexus brings together 800 animated vignettes and 140 accompanying Python notebooks, encompassing most of the mathematical content I have shared on social media. www.gpeyre.com/mathematical...
26718
Gabriel Peyré @gabrielpeyre.bsky.social · 30/07/2026
Top: Markov chains contract the probability simplex toward the unique stationary positive eigenvector (Perron-Frobenius). Bottom: Sinkhorn contracts the simplex non-linearly to an approximate solution of optimal transport (nonlinear Perron-Frobenius).
2284
Gabriel Peyré @gabrielpeyre.bsky.social · 13/07/2026
The KL barycenter of Gaussians is Gaussian: argmin_μ ∑ᵢ λᵢ KL(μ | 𝒩(mᵢ, Σᵢ)) = 𝒩(m, Σ), with Σ⁻¹ = ∑ᵢ λᵢ Σᵢ⁻¹ m = ∑ᵢ λᵢ Σ Σᵢ⁻¹ mᵢ
1130
Gabriel Peyré @gabrielpeyre.bsky.social · 10/07/2026
Duhamel (discrete) formula for the difference of matrix products: A1 ... An − B1 ... Bn = sum_{k=1}^n A1 ... A_{k-1} (Ak − Bk) B_{k+1} ... Bn
481
Gabriel Peyré @gabrielpeyre.bsky.social · 08/07/2026
A new (?) concept: the interactiv logo: www.gpeyre.com/ot4ml/
0191
Gabriel Peyré @gabrielpeyre.bsky.social · 07/07/2026
Denoting f⁎ the Legendre transform of f, the map {f convex} ↦ −log ∫ exp(−f⁎) is convex. arxiv.org/abs/1304.0630
arxiv.org
Moment Measures
With any convex function F on a finite-dimensional linear space X such that F goes to infinity at infinity, we associate a Borel measure on the dual space X*. This measure is obtained by pushing forwa...
1174
Gabriel Peyré @gabrielpeyre.bsky.social · 07/07/2026
If θ: ℝ₊ → ℝ is concave and L: ℝᵈ → ℝ is convex, then (a,m) ↦ θ(a)L(m/θ(a)) is convex. The special case θ = id is the perspective transformation (see arxiv.org/pdf/1610.01552). A large class of exotic optimal transport extensions relies on this construction.
arxiv.org
1122
Gabriel Peyré @gabrielpeyre.bsky.social · 04/07/2026
Doubly positive matrices: D_n = { X = Xᵀ in R_+^{n×n}, eig(X) ≥ 0 }. Totally positive matrices: T_n = { X = AAᵀ : ∃p, A in R_+^{n×p} }. One has T_n ⊂ D_n. For n ≤ 4: T_n = D_n. For n ≥ 5: T_n ≠ D_n.
optimization-online.org
The Difference Between 5×5 Doubly Nonnegative and Completely Positive Matrices – Optimization Online
191
Gabriel Peyré @gabrielpeyre.bsky.social · 02/07/2026
Slides for an introductory talk on IA for maths. speakerdeck.com/gpeyre/ia-fo...
speakerdeck.com
IA for theory
Slide for an introduction talk on IA for maths.
1329
Gabriel Peyré @gabrielpeyre.bsky.social · 16/06/2026
The alpha version of my new book "Optimal Transport for Machine Learners" is out, with in particular an online version with interactive figures www.gpeyre.com/ot4ml/
09843
Gabriel Peyré @gabrielpeyre.bsky.social · 23/05/2026
Drifting vs kernel (un-normalized) flows (yes, I know I love Gaussians). Stochastic matrices (and Sinkhorn divergences) are back in business :)
2264
Gabriel Peyré @gabrielpeyre.bsky.social · 23/05/2026
Thanks to LLMs, I think the exponent for the (refutal) of Erdos unit distance conjecture will be more famous than omega (the exponent for matrix multiplication)!
1121
Reposted by Gabriel Peyré
David Monniaux @monniauxd.bsky.social · 15/05/2026
L'IA générative face au concours d'entrée à l'École normale supérieure blogs.mediapart.fr/david-monnia...
blogs.mediapart.fr
L'IA générative face au concours d'entrée à l'École normale supérieure
Dans un précédent billet, j'avais essayé un outil d'IA générative sur un commentaire historique. Passons maintenant à un énoncé de concours particulièrement difficile.
196227
Gabriel Peyré @gabrielpeyre.bsky.social · 06/05/2026
Fun fact: a norm one covariances lifts to a norm on matrices if and only if it is monotone (e.g., Shatten norms).
181
Gabriel Peyré @gabrielpeyre.bsky.social · 16/04/2026
While it is intuitively clear that straighter trajectories should reduce discretization error when integrating an ODE (for instance, in flow matching), I could not find a precise bound. I therefore rewrote the proof of Cauchy-Lipschitz to make this explicit. github.com/gpeyre/Discr...
1286
Gabriel Peyré @gabrielpeyre.bsky.social · 09/04/2026
I wrote a short mathematical companion tutorial to my notebook on discrete diffusion models. It gives an informal derivation of the connection between maximum likelihood estimation of the backward transition kernel and denoising score matching. github.com/gpeyre/Discr...
0405
Gabriel Peyré @gabrielpeyre.bsky.social · 08/04/2026
It seems that analysts are not lagging behind ... www.scottnarmstrong.com/2026/04/form...
scottnarmstrong.com
Formalizing De Giorgi-Nash-Moser theory in Lean - Scott Armstrong
Julia Kempe and I have just completed a Lean 4 formalization of De Giorgi--Nash--Moser theory. We formalized the full slate of interior regularity statements for weak solutions of divergence-form elli...
062
Gabriel Peyré @gabrielpeyre.bsky.social · 07/04/2026
For those interested in normalized gradient methods and optimal transport: I introduce a new class of "spectral" Wasserstein distances for which spectrally normalized gradient descent (Muon but without momentum and small step size ...) is a spectral-W gradient flow: arxiv.org/abs/2604.04891
arxiv.org
Muon Dynamics as a Spectral Wasserstein Flow
Gradient normalization is central in deep-learning optimization because it stabilizes training and reduces sensitivity to scale. For deep architectures, parameters are naturally grouped into matrices ...
12410
Gabriel Peyré @gabrielpeyre.bsky.social · 21/03/2026
Pinsker inequality (which should be your favorite inequality ever, right?!) extends verbatim to matrices!
1306
Gabriel Peyré @gabrielpeyre.bsky.social · 01/03/2026
I have added a new tutorial on discrete diffusion models: github.com/gpeyre/ot4ml
05717
Gabriel Peyré @gabrielpeyre.bsky.social · 14/01/2026
Sinkhorn is back in business! arxiv.org/abs/2512.24880
3130
Gabriel Peyré @gabrielpeyre.bsky.social · 02/01/2026
Cadeau de Noel pour moi même! Chaudement recommandé. Bravo @roger-mansuy.bsky.social
1181
Gabriel Peyré @gabrielpeyre.bsky.social · 14/11/2025
I have updated my "Optimal Transport for Machine Learners" repository with a Pytorch illustration of Wasserstein gradient flows on pairwise interaction functionals (MMD distances) github.com/gpeyre/ot4ml
1414
Gabriel Peyré @gabrielpeyre.bsky.social · 29/10/2025
What is the set of "means" one can approximate using only arithmetic and harmonic means ? For instance the geometric mean belongs to this closure, but can one approximate any mean sandwitched between the two?
140
Gabriel Peyré @gabrielpeyre.bsky.social · 27/09/2025
Thought of the day: It is somewhat mysterious why Gaussians remain stable under the particle-minimizing flow (i.e. the Wasserstein gradient flow) for so many widely used energies: entropy, Fisher information, quadratic interaction potentials, functionals depending only on mean and covariance,
1182
Reposted by Gabriel Peyré
CNRS @cnrs.fr · 11/09/2025
#Communiqué 🗞️ La médaille d'or 2025 du CNRS est décernée à Stéphane Mallat, mondialement reconnu pour ses travaux autour des mathématiques appliquées au traitement du signal et à l’intelligence artificielle. 👏 👉 cnrs.fr/fr/presse/en... #TalentsCNRS 🏅
39844
Gabriel Peyré @gabrielpeyre.bsky.social · 09/08/2025
To meditate while resting on the beach...
1110
Gabriel Peyré @gabrielpeyre.bsky.social · 02/07/2025
Fun (...) fact: the only linear operators on matrices that preserves the rank are X->AXB, where A and B are invertible (with X->X^T in the square case). This was apparently first proved (?) in 1959 by Marcus and Moyls.
1342
Reposted by Gabriel Peyré
Jeremie Kalfon 👨‍💻🧬🤖🚀 @jkobject.com · 17/06/2025
scPRINT is now finally on the Chan Zuckerberg Institute's Model Hub! 🎉 🧬 🌈 It is one more way you can use this cell foundation model to embed, denoise, predict cell type, get gene networks from your data from scratch, or fine-tune it on your own application / usecase: virtualcellmodels.cz...
virtualcellmodels.cziscience.com
scPRINT | v1.0 | Virtual Cells Platform
scPRINT is a cell foundation model, also called a Large Cell Model (LCM), trained on single-cell RNA sequence (scRNAseq) data from more than 50M human and mouse cells available through CZ CELLxGENE. Based on the transformer architecture, the model is fully open source and reproducible, with multiple checkpoint sizes available from 2M to 100M parameters. scPRINT demonstrated high performance for genome-wide cell-specific gene network inference when benchmarked against state-of-the-art models (e.g., scGPT, Geneformer v2, GENIE3). In addition, scPRINT has various zero-shot capabilities, including cell embedding, cell label prediction (e.g., cell type, sex, disease), and gene expression imputation, highlighting its potential as a versatile tool for single-cell analysis.
251
Gabriel Peyré @gabrielpeyre.bsky.social · 04/06/2025
Le prochain Data Science Colloquium à l'ENS, jeudi 12 juin, sera donné par David Louapre d'Ubisoft, "What modern AI and neuroscience can bring to non-playing characters in video games". David c'est bien sûr également le vulgarisateur scientifique de www.youtube.com/scienceetonn...
2194
Gabriel Peyré @gabrielpeyre.bsky.social · 31/05/2025
If one of the two distributions is an isotropic Gaussian, then flow matching is equivalent to a diffusion model. This is known as Tweedie's formula. In particular, the vector field is a gradient vector, as in optimal transport. speakerdeck.com/gpeyre/compu...
1474
Gabriel Peyré @gabrielpeyre.bsky.social · 25/05/2025
Lectures note for the course of Cyril Letrouit at Collège de France on the quantitative stability of optimal transport. www.imo.universite-paris-saclay.fr/~cyril.letro...
imo.universite-paris-saclay.fr
1145
Gabriel Peyré @gabrielpeyre.bsky.social · 25/05/2025
I have cleaned up the notebooks for my course on Optimal Transport for Machine Learners and added links to the slides and lecture notes. github.com/gpeyre/ot4ml
16010
Gabriel Peyré @gabrielpeyre.bsky.social · 20/05/2025
I have updated my slides on the maths of AI by an optimal pairing between AI and maths researchers ... speakerdeck.com/gpeyre/the-m...
3253
Gabriel Peyré @gabrielpeyre.bsky.social · 16/05/2025
Mon article sur les maths de l’IA est paru dans la gazette de la smf smf.emath.fr/publications... La version en anglais est sur arxiv arxiv.org/abs/2501.10465
0268
Gabriel Peyré @gabrielpeyre.bsky.social · 13/05/2025
I have cleaned a bit my lecture notes on Optimal Transport for Machine Learners arxiv.org/abs/2505.06589
arxiv.org
Optimal Transport for Machine Learners
Optimal Transport is a foundational mathematical theory that connects optimization, partial differential equations, and probability. It offers a powerful framework for comparing probability distributi...
011930
Reposted by Gabriel Peyré
Lénaïc Chizat @lenaicchizat.bsky.social · 14/04/2025
Announcing : The 2nd International Summer School on Mathematical Aspects of Data Science mathsdata2025.github.io EPFL, Sept 1–5, 2025 Speakers: Bach @bachfrancis.bsky.social Bandeira Mallat Montanari Peyré @gabrielpeyre.bsky.social For PhD students & early-career researchers Apply before May 15!
mathsdata2025.github.io
Mathematical Aspects of Data Science
Graduate Summer School - EPFL - Sept. 1-5, 2025
14424
Reposted by Gabriel Peyré
NAVER LABS Europe @naverlabseurope.bsky.social · 11/04/2025
Applications are 📣OPEN📣 for #PAISS2025 THE AI summer school in #Grenoble 1-5 Sept! Speakers so far @yann-lecun.bsky.social @dimadamen.bsky.social @arthurgretton.bsky.social @gabrielpeyre.bsky.social @science4all.org A. Cristia J. Revaud M. Caron J. Carpentier M. Vladimirova ➡️ paiss.inria.fr
0309
Gabriel Peyré @gabrielpeyre.bsky.social · 01/04/2025
The AI for Science summer school, coorganized by CNRS and U of Chicago will be in Paris, June 30th to july 4th, register asap if you want attend! datascience.uchicago.edu/events/ai-sc...
datascience.uchicago.edu
AI+Science Summer School 2025 – DSI arrow-right-large arrow-left-smallarrow-right-large-greyarrow-right-large-yellowarrow-right-largearrow-right-long-yellowarrow-right-smallclosefacet-arrow-down...
02210
Gabriel Peyré @gabrielpeyre.bsky.social · 28/03/2025
Futur best seller!
2376
Reposted by Gabriel Peyré
Francis Bach @bachfrancis.bsky.social · 24/03/2025
Characterizing finely the decay of eigenvalues of kernel matrices: many people need it, but explicit references are hard to find. This blog post reviews amazing asymptotic results from Harold Widom (1963!) and proposes new non-asymptotic bounds. francisbach.com/spectrum-ker...
0517
Reposted by Gabriel Peyré
Joséphine Raugel @josephine-raugel.bsky.social · 26/02/2025
⚡️Check out our workshop tomorrow at @lpiparis.bsky.social, great speakers (@gabrielpeyre.bsky.social, @sdascoli.bsky.social, @samillingworth.com & many more) will cover Theory and Applications of Generative AI + Connexions with neuroscience 🧠 And there's food 🍰 ➡️ genai-conference-website.vercel.app
073
Reposted by Gabriel Peyré
kyunghyuncho.bsky.social @kyunghyuncho.bsky.social · 07/02/2025
kyunghyuncho.me/softmax-fore...
kyunghyuncho.me
Softmax forever, or why I like softmax – Kyunghyun Cho
35010
Gabriel Peyré @gabrielpeyre.bsky.social · 07/02/2025
@vickykalogeiton.bsky.social and @davidpicard.bsky.social updating live there slides to quote my talk just before ... next level presentation!
0352
Reposted by Gabriel Peyré
Peyman Milanfar @docmilanfar.bsky.social · 01/02/2025
This is due to a beautiful and little-known covariance identity by Hoeffding (1940): Cov(x,y) = ∫∫[F(x, y) - F(x)*F(y)] dxdy So the difference between independence and uncorrelated-ness comes down to point-wise equality vs. a (weaker) integral equality between the CDFs. 2/2
3556
Reposted by Gabriel Peyré
Rémi Flamary @rflamary.bsky.social · 17/01/2025
Titouan Vayer and I are organizing a one day workshop on optimal transport and machine learning in ENS Lyon on Feb. 17. Registration is free but mandatory. The incredible keynote speakers are Laetitia Chapel, Filippo Santambrogio and @brunolevy01.bsky.social. gdr-iasis.cnrs.fr/reunions/tra...
gdr-iasis.cnrs.fr
Transport optimal et ses applications en machine learning et analyse de données - GdR IASIS
Pour les journées scientifiques organisées en 2025 : les inscriptions seront ouvertes à partir de mi-janvier et les demandes de prise en charge de missions seront traitées à partir du 27 janvier. En r...
21812