Results 101 to 110 of about 1,563 (194)
Local Polynomial Regression and Filtering for a Versatile Mesh‐Free PDE Solver
International Journal for Numerical Methods in Fluids, Volume 98, Issue 7, Page 804-839, July 2026.A high‐order, mesh‐free finite difference method for solving differential equations is presented. Both derivative approximation and scheme stabilisation is carried out by parametric or non‐parametric local polynomial regression, making the resulting numerical method accurate, simple and versatile. Numerous numerical benchmark tests are investigated for
Alberto M. Gambaruto
wiley +1 more source
Numerical Methods for Partial Differential Equations, Volume 42, Issue 4, July 2026.ABSTRACT While recent Anderson acceleration (AA) convergence theory [Pollock et al., IMA Num. An., 2021] requires that the AA optimization norm match the Hilbert space norm associated with the fixed point operator, in implementations the ℓ2$$ {\ell}^2 $$ norm is the most common choice. So far there is little research done regarding this discrepancy. To
Elizabeth Hawkins, Leo G. Rebholz
wiley +1 more source
Geophysical Research Letters, Volume 53, Issue 11, 16 June 2026.Abstract Fractured porous rocks host a complex interplay of heat, fluids, deformation, and chemical reactions, yet their combined influence on solute transport remains poorly resolved. Here we show that these processes are more tightly coupled than previously assumed.
Kai Wang +6 more
wiley +1 more source
Positional Embeddings for Solving PDEs with Evolutional Deep Neural Networks
This work extends the paradigm of evolutional deep neural networks (EDNNs) to solving parametric time -dependent partial differential equations (PDEs) on domains with geometric structure.
Kast, Mariella, Hesthaven, Jan S.
core +1 more source
Advanced Science, Volume 13, Issue 33, 15 June 2026.We introduce AutomataGPT, a generative pretrained transformer (GPT) trained on synthetic spatiotemporal data from 2D cellular automata to learn symbolic rules. Demonstrating strong performance on both forward and inverse tasks, AutomataGPT establishes a scalable, domain‐agnostic framework for interpretable modeling, paving the way for future ...
Jaime A. Berkovich +2 more
wiley +1 more source
Optimal energy growth lower bounds for a class of solutions to the vectorial Allen-Cahn equation
, 2014 We prove optimal lower bounds for the growth of the energy over balls of minimizers to the vectorial Allen-Cahn energy in two spatial dimensions, as the radius tends to infinity. In the case of radially symmetric solutions, we can prove a stronger result
Sourdis, Christos
core
AIChE Journal, Volume 72, Issue 6, June 2026.Abstract This article demonstrates the integration of in‐line mass spectrometry as a process analytical technology (PAT) tool with model‐based soft sensors in a continuous filtration‐drying carousel system for solid–liquid separation (SLS) of crystal slurries.
Inyoung Hur +3 more
wiley +1 more source
Advanced Intelligent Discovery, Volume 2, Issue 3, June 2026.We propose a residual‐based adversarial‐gradient moving sample (RAMS) method for scientific machine learning that treats samples as trainable variables and updates them to maximize the physics residual, thereby effectively concentrating samples in inadequately learned regions.
Weihang Ouyang +4 more
wiley +1 more source
A symmetric BEM approach to strain gradient elasticity for 2D static boundary-value problems
, 2014 The symmetric Galerkin Boundary Element Method is used to address a class of strain gradient elastic materials featured by a free energy function of the (classical) strain and of its (first) gradient.
Terravecchia, S +2 more
core
Biotechnology and Bioengineering, Volume 123, Issue 6, Page 1465-1480, June 2026.ABSTRACT The growing demand for biopharmaceutical products reflects their effectiveness in medical treatments. However, developing new biopharmaceuticals remains a major bottleneck, often taking up to a decade before market approval. Machine learning (ML) models have the potential to accelerate this process, but their success depends on access to large
Mohammad Golzarijalal +2 more
wiley +1 more source