Your research scout briefing

Zecheng Zhang

Operator learning aims to learn mappings between function spaces and has been widely used for PDE-related problems. This raises two fundamental questions.

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

Rui Liu, Jie Shen

We study uniform approximation for operator learning on compact subsets of Banach spaces. We bound the recovery dimensions and the number of network parameters in terms of compact-set complexity, the target gauge, and recovery stability.

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

Jinniao Qiu

We study stochastic Hamilton--Jacobi--Bellman equations arising from stochastic optimal control problems with random coefficients and possible degenerateness. The viscosity solution is reformulated by using adapted smooth random test fields that touch on stopped space--time neighborhoods almost surely.

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

Kaïs Ammari, Taoufik Hmidi, Haocheng Yang

In this article, we investigate the boundary controllability of finite point-vortex systems in bounded multiply connected planar Lipschitz domains. The controls are applied locally on an arbitrary nonempty relatively open subset of the boundary.

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

Omar Lazar, Liutang Xue, Jiakun Yang

We study global regularity and the infinite Prandtl number limit for temperature patches of the two-dimensional nondiffusive Boussinesq system on $\mathbb R^2$, with velocity dissipation $Λ^{2α}$, $\frac12<α\leq1$. We establish global well-posedness for $θ_0\in L^1\cap L^\infty$ and divergence-free $u_0\in W^{1,p}$, $2\leq p<\infty$.

Next action: Read the introduction and main theorem statements, focusing on boussinesq, regularity, number.

Minxing Zheng, Shixiang Zhu

Generative conformal prediction builds uncertainty sets from samples of a conditional generator, which are efficient only when the samples represent the response distribution well. This can require many samples, each of which can be costly, as in large diffusion models and scientific simulators, so the sampling budget must be used efficiently.

Next action: Read the introduction and main theorem statements, focusing on conformal prediction, conformal, prediction.

Edward Lester, Dao Nguyen

We study the calibration of interaction-driven ordinary differential equation (ODE) models for traffic and pedestrian dynamics by means of optimal control and stochastic gradient descent. The interaction forces are either prescribed by classical physics-inspired laws, whose parameters are identified from trajectory data, or represented by feed-forward neural networks whose weights act as control variables.

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

Julian Chan, Javier Mora Jimenez

Do LLM scientific agents rely only on their initial task context, or do they adapt their decisions in response to experimental feedback? We study this question in neural operator adaptation, where a large language model (LLM) selects fine-tuning configurations under a limited trial budget.

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

Lingbing He, Jin-Cheng Jiang, Jong-In Kim, Donghyun Lee, Sungbin Park

We establish a conditional theory of genuinely non-perturbative global dynamics for the cutoff Boltzmann equation with hard potentials on the three-dimensional torus. We prove that any global mild solution satisfying \[ \sup_{t\ge0}\int_{\mathbb R^3}W(v)\sup_{x\in\mathbb T^3}F(t,x,v)\,dv<\infty, \] for a suitable stretched-exponential weight \(W\), enjoys uniform weighted \(L^\infty\) control, is unique in the dissipative mild class, and converges exponentially to the global Maxwellian.

Next action: Read the introduction and main theorem statements, focusing on kinetic, data, boltzmann.

Wenshuai Hu, Guanghua Ji, Xiao Li

Structure-preserving numerical methods for gradient flows have been extensively developed when the governing equations and free energies are known. However, accurately predicting dynamics from observational data while preserving intrinsic physical properties remains challenging.

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

Marko Tvrdic, Justin Richmond Domingo, Jae Ann Buenaluz, Jydell Ashley Palomo Penollar, Edmayelle Villavicencio Alforja, Jobi Fallaeria Subosa et al.

Preclinical models poorly predict human drug efficacy, particularly in neurological disorders. Neural activity offers a uniquely rich source of translational information because it captures high-dimensional variation in nervous-system function that can be measured in both animals and humans.

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

Hossein Mohammadi

Spatial modelling and inference arise across many areas of science and engineering, with Kriging and Gaussian process (GP) regression serving as two widely used predictive models. Although grounded in the same random field theory, these approaches are often treated in isolation.

Next action: Read the introduction and main theorem statements, focusing on processes, prediction, uncertainty quantification.

Bingjie Wu, Hao Kang, Chen Jia

Exponential ergodicity of stochastic reaction networks has attracted considerable attention in recent years [SIAM J. Appl.

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

Junming Duan, Mengni Li

In this paper, we establish a quantitative finite-time approximation of scattering fields for nonlinear Alfvén waves in three-dimensional ideal incompressible magnetohydrodynamics near a strong constant magnetic background. Using the global weighted estimates, we prove that truncating the scattering profiles at time $T$ introduces an error in both the $L^\infty$ norm and the $H^{N_*+1}$ norm of order $$ O\bigl(\varepsilon^2(R+T)^{-δ}\bigr), $$ where $\varepsilon\in (0,1)$ measures the size of the initial...

Next action: Read the introduction and main theorem statements, focusing on mhd, sobolev, regularity.

Xing Cheng, Zeyu Lyu, Chenmin Sun, Yunrui Zheng

We study the energy-subcritical and energy-critical complex Ginzburg-Landau equations on three-dimensional Riemannian manifolds. We establish the asymptotic behavior of solutions via a purely energy-based approach and subsequently prove a fast/slow dichotomy of the solutions.

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

Adonis Jamal, Samy Mekkaoui, Yadh Hafsi, Huyên Pham

We develop a model-free policy gradient method for discrete-time mean-field control (MFC). In MFC, the policy affects the objective both through the controlled dynamics and through the population distribution.

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

Aravinda Kanchana Ruwanpathirana, Hemant Tyagi, Sunny G. W. Wang

We consider the problem of learning structured linear dynamical systems over convex sets $\mathcal{K}$, where only a small subset of the observations are available at each time point. An estimator which minimizes a bias-corrected, potentially non-convex objective function is proposed.

Next action: Read the introduction and main theorem statements, focusing on learning, systems, applied.

Jack Jewson

The likelihood input to a Bayesian analysis almost never exactly represents how the data were generated, calling into question the validity of posterior inferences. We review a variational interpretation of the Bayesian posterior as an alternative justification for its use under model misspecification, and consider the resulting implications on uncertainty quantification and parameter estimation.

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

Sanguan Zhong, Jie Qi

The paper considers a system of unidirectional first-order hyperbolic partial differential equations (PDEs) coupled with ordinary differential equations (ODEs), modeling two-phase advective transport processes subject to both boundary and distributed input delays. To address the system defined over two spatial domains, we design a set of three backstepping transformations involving both Volterra- and Fredholm-type integral operators, which introduce eight kernel functions: five associated with the PDE states and...

Next action: Read the introduction and main theorem statements, focusing on systems, boundary, operators.

Anna Zimmel, Fleur Hendriks, Markus Holzleitner, Florian Sestak, Martin Weichselbaumer, Vlado Menkovski et al.

Bifurcations are ubiquitous in physical systems, from structural buckling to fluid and climate dynamics, yet they remain largely unexplored in deep learning. At a symmetry-breaking bifurcation, a single input admits multiple equally valid solutions, violating the one-to-one assumption underlying most learned physical surrogates.

Next action: Read the introduction and main theorem statements, focusing on modeling, systems, high.

Joel-Pascal Ntwali N'konzi, Feliks Nüske, Stefan Klus

Koopman operator theory provides a linear framework for analyzing nonlinear dynamical systems and has become a major tool for data-driven modeling. A central challenge, however, is that finite-dimensional approximations computed by methods such as extended dynamic mode decomposition (EDMD) require the dictionary to be specified a priori.

Next action: Read the introduction and main theorem statements, focusing on learning, data, operator.

L Baratchart, D P Hardin, C Villalobos-Guillén

Inverse source problems in divergence form consist in finding a vector field with prescribed support S, whose divergence is the Laplacian of some observed potential. In this paper, we assume the unknown vector field is a vector-valued measure, and we study the corresponding least square inversion problems, regularized by penalizing the total variation, without discretizing the criterion nor the unknown.

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

Abdelhafid Zenati, Dave D Muir, Kamal Youcef-Toumi

Computing local Stackelberg equilibria in nonlinear sequential games is challenging because gradient-based best-response methods may diverge, cycle, or converge to stationary points that do not satisfy the required equilibrium conditions. These failures can result from curvature asymmetry, unstable response mappings, and lack of invariance with respect to the follower's best-response manifold.

Next action: Read the introduction and main theorem statements, focusing on computing, high, games.

Mert Albaba, Jens Beißwenger, Anna Manasyan, Daniel Marta, Michael J. Black, Wieland Brendel et al.

Teaching a humanoid to follow instructions with its whole body runs into two obstacles. Its action space is large and tightly coupled: legs, arms, and fingers must move together while the robot keeps its balance, which makes joint-level actions hard to learn.

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