Source author record

Sébastien Petit

Sébastien Petit appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Computation Machine Learning cond-mat.mtrl-sci cond-mat.str-el cond-mat.supr-con Methodology

Catalog footprint

What is connected

5works

6topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2026arXiv

Relaxed Gaussian process interpolation: a goal-oriented approach to Bayesian optimization

This work presents a new procedure for obtaining predictive distributions in the context of Gaussian process (GP) modeling, with a relaxation of the interpolation constraints outside ranges of interest: the mean of the predictive distributions no longer necessarily interpolates the observed values when they are outside ranges of interest, but are simply constrained to remain outside. This method called relaxed Gaussian process (reGP) interpolation provides better predictive distributions in ranges of interest, especially in cases where a stationarity assumption for the GP model is not appropriate. It can be viewed as a goal-oriented method and becomes particularly interesting in Bayesian optimization, for example, for the minimization of an objective function, where good predictive distributions for low function values are important. When the expected improvement criterion and reGP are used for sequentially choosing evaluation points, the convergence of the resulting optimization algorithm is theoretically guaranteed (provided that the function to be optimized lies in the reproducing kernel Hilbert space attached to the known covariance of the underlying Gaussian process). Experiments indicate that using reGP instead of stationary GP models in Bayesian optimization is beneficial.

preprint2024arXiv

Electron-phonon mediated superconductivity in La$_6$Ni$_5$O$_{12}$ nickel oxides

Nickel oxide superconductors offer an alternative playground for understanding the formation of Cooper pairs in correlated materials such as the famous cuprates. By studying the La$_{n+1}$Ni$_n$O$_{2n+2}$ phase diagram on the basis of hybrid and spin-polarized density functional theory simulations, we reveal the existence of charge and bond ordered (CBO) insulating phases that are quenched by doping effects, ultimately resulting in a metallic phase at the n=5 member. Nevertheless, the phonons associated with the CBO identified in the phase diagram remain sufficiently large to mediate Cooper pairs in La$_6$Ni$_5$O$_{12}$, yielding a computed critical temperature between T$_c$=11-19K consistent with the 13K observed experimentally in Nd$_6$Ni$_5$O$_{12}$. Thus, in order to identify the superconducting mechanism, extracting the relevant instabilities in the doping phase diagram of superconductors appears critical.

preprint2021arXiv

Numerical issues in maximum likelihood parameter estimation for Gaussian process interpolation

This article investigates the origin of numerical issues in maximum likelihood parameter estimation for Gaussian process (GP) interpolation and investigates simple but effective strategies for improving commonly used open-source software implementations. This work targets a basic problem but a host of studies, particularly in the literature of Bayesian optimization, rely on off-the-shelf GP implementations. For the conclusions of these studies to be reliable and reproducible, robust GP implementations are critical.

preprint2020arXiv

Towards new cross-validation-based estimators for Gaussian process regression: efficient adjoint computation of gradients

We consider the problem of estimating the parameters of the covariance function of a Gaussian process by cross-validation. We suggest using new cross-validation criteria derived from the literature of scoring rules. We also provide an efficient method for computing the gradient of a cross-validation criterion. To the best of our knowledge, our method is more efficient than what has been proposed in the literature so far. It makes it possible to lower the complexity of jointly evaluating leave-one-out criteria and their gradients.

preprint2012arXiv

First principles calculation of the phonons modes in the hexagonal $\rm YMnO_3$ ferroelectric and paraelectric phases

The lattice dynamics of the $\rm YMnO_3$ magneto-electric compound has been investigated using density functional calculations, both in the ferroelectric and the paraelectric phases. The coherence between the computed and experimental data is very good in the low temperature phase. Using group theory, modes continuity and our calculations we were able to show that the phonons modes observed by Raman scattering at 1200K are only compatible with the ferroelectric $P6_{3} cm$ space group, thus supporting the idea of a ferroelectric to paraelectric phase transition at higher temperature. Finally we proposed a candidate for the phonon part of the observed electro-magnon. This mode, inactive both in Raman scattering and in Infra-Red, was shown to strongly couple to the Mn-Mn magnetic interactions.