Portail HAL du Collège de France
Not a member yet
22122 research outputs found
Sort by
When Less is More: Approximating the Quantum Geometric Tensor with Block Structures
The natural gradient is central in neural quantum states optimizations but it is limited by the cost of computing and inverting the quantum geometric tensor, the quantum analogue of the Fisher information matrix. We introduce a block-diagonal quantum geometric tensor that partitions the metric by network layers, analogous to block-structured Fisher methods such as K-FAC. This layer-wise approximation preserves essential curvature while removing noisy cross-layer correlations, improving conditioning and scalability. Experiments on Heisenberg and frustrated - models show faster convergence, lower energy, and improved stability
Quand les mots disent les choses: Une archéologie linguistique de la dyade mère-enfant
International audienceThis paper tests the hypothesis that certain sound–meaning patterns associated with “mother” and “breast” may reflect a very ancient codiffusion, inherited from the earliest migrations of Homo sapiens and from the centrality of the mother–infant dyad. Within a framework of “linguistic archaeology,” four phonotactic traits are examined across 2,959 languages (“mother”) and 7,322 languages (“breast”) from the Lexibank and ASJP databases: [n]/[ŋ] and [na]/[ŋa] in initial position for “mother,” and [mu] and [amu] for “breast.” Their distribution is assessed through spatial analyses (Moran’s I, binomial z-scores on 2°×2° grids, random permutations, great-circle distances). The results reveal a non-random structuring for [n]/[ŋ] and [mu], with hotspots in Africa, South Asia, Island Southeast Asia, Melanesia, and Australia, regions that align with the southern routes of early Homo sapiens dispersal. The forms [na] and [amu], more geographically restricted, appear as regional archaisms. The strong geographic co-occurrence between [n]/[na] and [mu], contrasting with the limited spread of [amu], suggests an ancient lexical core linking “mother” and “breast,” not reducible to articulatory biases alone. Without positing a single protolanguage, the study shows that linguistic areology can reveal fossil traces of an early cultural structuring around the mother–infant dyad, providing partial support for the hypothesis of an initial cultural unity disseminated during the first out-of-Africa dispersals of Homo sapiens.Cet article teste l’hypothèse selon laquelle certains patrons son/sens associés à « mère » et « sein » pourraient refléter une codiffusion très ancienne, héritée des premières migrations d’Homo sapiens et de la centralité de la dyade mère-enfant. Dans une perspective d’« archéologie linguistique », quatre traits phonotactiques sont examinés dans 2 959 langues (« mère ») et 7 322 langues (« sein ») issues des bases Lexibank et ASJP : [n]/[ŋ] et [na]/[ŋa] en initiale de « mère » ; [mu] et [amu] pour « sein ». Leur distribution est évaluée via des analyses spatiales (Moran I, z-scores sur grilles 2°×2°, permutations aléatoires, distances orthodromiques). Les résultats révèlent une structuration non aléatoire pour [n]/[ŋ] et [mu], avec des foyers en Afrique, Asie du Sud, Insulinde, Mélanésie et Australie, soit des zones concordant avec les routes australes des premières dispersions d’homo sapiens. Les formes [na] et [amu], plus localisées, apparaissent comme des archaïsmes régionaux. La forte cooccurrence géographique entre [n]/[na] et [mu], contrastant avec la diffusion restreinte de [amu], suggère un noyau lexical ancien liant « mère » et « sein », non réductible aux seuls biais articulatoires. Sans postuler de protolangage, l’étude montre que l’aréologie linguistique permet d’identifier des traces fossiles d’une structuration culturelle précoce autour de la dyade mère-enfant, offrant un appui partiel à l’hypothèse d’une unité culturelle initiale diffusée avec les premières sorties d’Afrique d’Homo sapiens
BV-BRST Noether theorem
International audienceThe BRST Noether theorem, or ``Noether's 1.5 theorem'', asserts the triviality of the BRST Noether current. We provide two proofs of this theorem that are both valid without restriction on the structure of the gauge theory, extending thereby previous proofs holding in the case of gauge theories for which the solution of the master equation is linear in the antifields. We also relate explicitly the BRST Noether current to the BRST master current appearing in the master equation
Droit, culture et société de la Rome antique: [résumé des cours et travaux : 2021-2022]
International audienc
Innovation technologique Liliane Bettencourt (chaire annuelle 2021-2022): [résumé des cours et travaux : 2021-2022]
International audienc
Champs, cordes et gravité: [résumé des cours et travaux : 2021-2022]
International audienc
Keep Singing the Gospel
International audienceGospel is a strongly typed, contract based specification language for OCaml. Gospel aims to be a lightweight surface syntax for Separation Logic formulae that can be used to specify arbitrary values. It is meant to be general enough that it may be used in conjunction with whatever verification environment (or environments) the programmer prefers. Although Gospel has already been presented in previous publications, we have made improvements to its semantics, type systems and tooling ecosystem that are worth reporting on. We present these through a series of examples and their translation into GospelSL, an intermediate Separation Logic that we use to define the semantics of Gospel. From GospelSL, we show how we can automatically create a Rocq file that enables verification using CFML or Iris
Actualités 2025 en neurosciences fondamentales : présentée par Armelle Rancillac (Chargée de Recherche à l’Inserm, Collège deFrance, Paris).
In this presentation at the Strasbourg Sleep Congress, Armelle Rancillac, a researcher at Inserm and the Collège de France in Paris, presented a summary of the most significant work published in 2025 on sleep in fundamental neuroscience. The selected studies were presented around four themes: (i) the molecular and cellular mechanisms of sleep pressure, (ii) the genetic variability of sleep needs, (iii) brain cleansing, and finally (iv) the effects of contextual determinants such as stress and external temperature.Dans cette présentation au Congrès du Sommeil de Strasbourg, Armelle Rancillac, chargée de recherche à l’Inserm, au Collège de France à Paris, a exposé une synthèse des travaux marquants publiés en 2025 sur le sommeil en neurosciences fondamentales. Les études sélectionnées ont été présentées autour de quatre axes : (i) les mécanismes moléculaires et cellulaires de la pression de sommeil, (ii) la variabilité génétique des besoins de sommeil, (iii) le nettoyage du cerveau, et enfin (iv) les effets des déterminants contextuels tels que le stress et la température externe
Deep Search for Joint Sources of Gravitational Waves and High-Energy Neutrinos with IceCube During the Third Observing Run of LIGO and Virgo
International audienceThe discovery of joint sources of high-energy neutrinos and gravitational waves has been a primary target for the LIGO, Virgo, KAGRA, and IceCube observatories. The joint detection of high-energy neutrinos and gravitational waves would provide insight into cosmic processes, from the dynamics of compact object mergers and stellar collapses to the mechanisms driving relativistic outflows. The joint detection of multiple cosmic messengers can also elevate the significance of the common observation even when some or all of the constituent messengers are sub-threshold, i.e. not significant enough to declare their detection individually. Using data from the LIGO, Virgo, and IceCube observatories, including sub-threshold events, we searched for common sources of gravitational waves and high-energy neutrinos during the third observing run of Advanced LIGO and Advanced Virgo detectors. Our search did not identify significant joint sources. We derive constraints on the rate densities of joint sources. Our results constrain the isotropic neutrino emission from gravitational-wave sources for very high values of the total energy emitted in neutrinos (> erg)
Code Generation via Meta-programming in Dependently Typed Proof Assistants
Dependently typed proof assistants offer powerful meta-programming features, which allow users to implement proof automationor compile-time code generation. This paper surveys meta-programmingframeworks in Rocq, Agda, and Lean, with seven implementations of arunning example: deriving instances for the Functor typeclass. This example is fairly simple, but realistic enough to highlight recurring difficulties with meta-programming: conceptual limitations of frameworks suchas term representation – and in particular binder representation –, meta-language expressiveness, and verifiability, as well as current limitationssuch as API completeness, learning curve, and prover state management,which could in principle be remedied. We conclude with insights regarding features an ideal meta-programming framework should provide