Your research scout briefing

Charles Meynard

We investigate two learning procedures and their relationship to the formation of equilibrium in potential mean field games (MFGs). The first one is a version of the fictitious play in which players observe directly the actions (controls) of other players and learn the distribution of the MFG through an update rules on controls.

Next action: Read the introduction and main theorem statements, focusing on mean field games, learning, mean.

Zhao Zhang, Chengjin Guo, Zhenglong Chen, Kai Zhang, Piyang Liu, Xia Yan

Surrogate modelling is important to enhance the computational efficiency of uncertainty quantification for physical processes involving random parameters. The advantage of Fourier neural operator (FNO) compared to conventional surrogates is its cross-resolution prediction capability.

Next action: Read the introduction and main theorem statements, focusing on neural operator, operator, sobolev.

Anis Bel Hadj Hassin, Matthias Ehrhardt

In this work, we study transparent boundary conditions for the time-fractional heat equation on a metric star graph. This equation is the standard continuum model for subdiffusive heat and mass transport in disordered and porous branched media.

Next action: Read the introduction and main theorem statements, focusing on metric graphs, fractional, metric.

Shilun Wei, Xiaoqiang Sun, Wei Li, Kejun Tang

Given incomplete measurements of a single physical field in a coupled system with unknown parameters, can we infer its full physical state and identify the underlying parameters? This problem is challenging because multiple coupled fields must be reconstructed simultaneously from limited observations of only one, while the system parameters are unknown.

Next action: Read the introduction and main theorem statements, focusing on neural operator, operators, neural.

Valentin Pesce

This paper introduces a robust strategy to establish large deviation principles for mean field interacting particle systems in T d with vanishing noise. Viewed as a Freidlin-Wentzell type problem, our approach relies on recent techniques for partial differential equations (in short PDE) on the space of probability measures [5].

Next action: Read the introduction and main theorem statements, focusing on systems, mean, field.

Léonard Ferdinand, Samaël Mackowiak, Aurélien Minguella

We construct Anderson-type operators in the full subcritical regime via a SPDE argument. Our work encompasses perturbations by a Gaussian noise of the fractional Laplacian on the torus or on the full space, and of the harmonic oscillator on the full space.

Next action: Read the introduction and main theorem statements, focusing on stochastic pde, operators, stochastic.

Charles Bertucci, Pierre-Louis Lions, Panagiotis E. Souganidis

We provide a complete study of the optimal control problem associated to the so-called Dyson equation, which describes the mean field evolution of the spectrum of large random matrices. We treat both the case of regular and singular terminal data.

Next action: Read the introduction and main theorem statements, focusing on optimal control, optimal, control.

Gi-Chan Bae, Hongxu Chen, Chanwoo Kim

We establish global energy compactness for rough three-dimensional axisymmetric Euler flow without swirl and use it to construct hydrodynamic limits of collisional kinetic equations beyond finite-time Euler regularity breakdown. For relative vorticity in $L^1\cap L^p$, regular approximate Euler flows are globally compact in $C_tL_x^2$ for $p\ge4/3$, without a sign condition or any spatial-moment assumption.

Next action: Read the introduction and main theorem statements, focusing on kinetic, regularity, landau equation.

Jomar Fajardo Rabajante

Multiple epidemic waves are often taken as evidence that simple compartmental models are structurally inadequate. Yet the key question is not whether a simple model can reproduce multiple peaks, but how much time-varying structure is minimally required to preserve observed epidemic-wave patterns and what can legitimately be inferred from such a representation.

Next action: Read the introduction and main theorem statements, focusing on epidemic, inference, inverse problems.

Solmaz S. Kia

The adjoint method is a standard approach for computing gradients in PDE-constrained optimization and central to parameter estimation, model calibration, and optimal control. This tutorial presents a discrete, implementation-oriented introduction to the method with an emphasis on conceptual understanding.

Next action: Read the introduction and main theorem statements, focusing on inverse problems, inverse, optimal control.

Ali Akbar Rezaei Lori, Piyush Grover

Mean-field games provide a continuum framework for modeling the dynamics of large, interacting populations of non-cooperative agents. This paper studies symmetry-breaking in a finite-horizon, two-choice min-LQG mean-field game in which identical agents with linear stochastic dynamics choose one of two equally desirable terminal destinations, while trading off control effort against social pressure to conform.

Next action: Read the introduction and main theorem statements, focusing on mean, field, game.

Xiaoyu Gu, Song Jiang, Qiangchang Ju, Yaobin Ou

In this paper, we rigorously justify the low Mach number limit of the full compressible Navier-Stokes equations in a three-dimensional smooth bounded domain, subject to the no-slip boundary condition for the velocity and well-prepared initial data. In the presence of heat conduction and large temperature variations, we establish the uniform estimates for strong solutions on a time interval independent of the Mach number by constructing an energy functional in a Sobolev framework.

Next action: Read the introduction and main theorem statements, focusing on navier-stokes, number, boundary.

Lorenzo Marinucci, Leonardo Di Nino, Gabriele D'Acunto, Paolo Di Lorenzo, Sergio Barbarossa

CoVariance Neural Networks and their extensions have emerged as effective tools for processing multivariate data, deriving graph shift operators directly from second-order statistics. These architectures, however, are designed for independent and identically distributed observations and do not fully capture the joint structure of temporal and cross-variable dependencies in multivariate time series.

Next action: Read the introduction and main theorem statements, focusing on inverse, neural, networks.

R{é}gis Duvigneau

Physics-informed machine learning as recently emerged as a new paradigm for the resolution of partial differential equations, including functionalities for parametric modeling and inverse design, using neural networks to represent the solution fields. However, several difficulties have been reported concerning the training of such networks, resulting in slow convergence, low accurate solutions and sensitivity to the choice of hyper-parameters.

Next action: Read the introduction and main theorem statements, focusing on learning, neural, inverse.

Laura Weigl, Anton Schiela

In PDE-constrained optimization a control is applied within a partial differential equation (PDE), e.g., as a force field, to achieve a desired configuration of the PDE solution. The objective is usually to find an optimal balance between control cost and deviation of the controlled state from the desired configuration.

Next action: Read the introduction and main theorem statements, focusing on optimal control, optimal, control.

Erhan Bayraktar, Haichuan Ding, Ibrahim Ekren

We study state-constrained stochastic control problems with a scalar surplus $R \geq 0$. Controls for which the drift and volatility of $R$ vanish at zero define a lower-dimensional Hamilton-Jacobi-Bellman equation.

Next action: Read the introduction and main theorem statements, focusing on control, stochastic, drift.

Robert Richardson

Observing a prediction error at one node can help correct predictions elsewhere, but the benefit depends on the residual dependence between nodes. A graph suggests where that dependence might occur, yet does not establish its sign or strength.

Next action: Read the introduction and main theorem statements, focusing on bayesian, uncertainty, when.

Viet-Anh Le, Rahul Mangharam

We propose a learning-to-optimize framework for solving parametric mixed-integer nonlinear programs arising in model predictive control and trajectory optimization. The proposed method combines neural network prediction with a sequential quadratic programming (SQP) refinement layer.

Next action: Read the introduction and main theorem statements, focusing on learning, neural, networks.

João Böger, Simon Driscoll, Niccolò Zagli, Valerio Lucarini, Francisco Camara Pereira

Machine-learning emulators of chaotic and stochastic systems are usually validated on forecast skill and long-run statistics. Neither certifies that an emulator responds correctly to forcing, the property that projection and attribution studies rely on.

Next action: Read the introduction and main theorem statements, focusing on stochastic, response, drift.

Sima Naraghi, Kourosh Parand, Amirhossein Sadr, Dara Rahmati

Isogeometric analysis (IGA) solves partial differential equations accurately on exact NURBS geometry, whereas neural solvers are mesh-free but often orders of magnitude less accurate and typically trained by non-convex optimization without error control. We propose IGA-KAN, which uses local Kolmogorov-Arnold networks, fitted in closed form, to improve the IGA solution instead of replacing it.

Next action: Read the introduction and main theorem statements, focusing on inverse, pdes, networks.

Mustafa Avci

For a prescribed classical profile and a fixed positive regularization parameter, we construct an Itô diffusion whose interior generator reproduces the regularized $p(x)$-Laplacian when applied to that profile. We determine the exact tangent spectra of the unregularized and regularized fluxes, characterize their behavior at zero gradient, and establish quantitative ellipticity bounds on bounded gradient ranges.

Next action: Read the introduction and main theorem statements, focusing on stochastic, drift, behavior.

José R. Berrendero, Antonio Coín, Antonio Cuevas

In the context of functional data analysis, we propose a Bayesian test to assess whether the mean function of a Gaussian population is identically zero. By leveraging the structure of the reproducing kernel Hilbert space (RKHS) associated with the covariance function of the underlying process, we approximate the mean by a finite linear combination of kernel sections.

Next action: Read the introduction and main theorem statements, focusing on bayesian, mean, data.

Arnab Ganguly, Padmanabhan Sundar

We establish large and moderate deviation principles for a broad class of nonlinear stochastic partial differential equations driven by multiplicative Poisson random measures in the small-noise regime. Our abstract framework is formulated for locally monotone evolution equations over a Hilbert triple and encompasses several important models arising in fluid dynamics and turbulence theory, including the two-dimensional Navier--Stokes equations, magnetohydrodynamics (MHD), the GOY shell model of turbulence, the...

Next action: Read the introduction and main theorem statements, focusing on stochastic, fluid, mhd.

Thomas Ryckeboer, Giacomo Meanti, Julien Mairal

Recovering the forward operator of an imaging system from unpaired data avoids both calibration hardware and the clean/degraded pairs that supervision requires-pairs that, for a real lens, are often impossible to acquire. Existing unpaired approaches that explicitly estimate the forward operator by distribution matching evaluate a candidate only through the measurements it generates, whose distribution should match that of the real ones.

Next action: Read the introduction and main theorem statements, focusing on inverse problems, inverse, data.