CORTEXA
← Browse
openalexZenodo (CERN European Organization for Nuclear Research)2026-07-23Cited by 0

An Honest Physics-Informed Neural Network Atlas: Sub-Percent on Smooth Forward PDEs, Orders Worse on Inverse, High-Frequency and Real Data

Felipe Santibañez-Leal

Physics-informed neural networks (PINNs) are promoted as a general differential-equation solver, but the accuracy actually achieved varies by orders of magnitude across problem types, and that variation is rarely laid out in one place. This report is a method atlas: a runnable catalogue in which each of the state-of-the-art PINN families (hard constraints, residual-based adaptive sampling, Fourier features, domain decomposition, operator learning, physics-informed operators, conformal-prediction uncertainty, structure-preserving Hamiltonian learning, and inverse solving) is exercised on at least one differential-equation case, trained by a dedicated engine, validated against an analytic or numerical reference, and scored by relative L2 error. Across the seventeen cases with a committed reference-validated error, the honest picture is a clean stratification. PINNs reach sub-percent accuracy on smooth, well-posed forward PDEs (a one-dimensional wave at 5e-5, a reaction-diffusion Allen-Cahn field at 1.2e-3, several mining and transport reduced models below 1e-3), with ten of seventeen cases under 1% relative error. But they degrade by two to three orders of magnitude on the hard classes: inverse problems (4e-2), high-wavenumber Helmholtz (1.0e-1), the lid-driven cavity Navier-Stokes benchmark (1.7e-1), operator learning (5.6e-2), and real noisy field data (6.9e-2). The finding matches the field's own failure-mode literature and is the report's point: PINNs are an accurate and flexible instrument where the solution is smooth and the problem well-posed, and they are not a drop-in replacement for a good classical solver where it is not. Every case ships its equations, its method, an honest synthetic provenance label, and a reproducible error; every number and figure regenerates from the committed artifacts. Runnable catalogue and reproducible pipeline (MIT): https://github.com/fsantibanezleal/CAOS_PINNLAB ; live at https://pinnlab.fasl-work.com .

View free PDFSource page

Related papers

openalexZenodo (CERN European Organization for Nuclear Research)2026-07-23

Code for Long-Time KdV Soliton Propagation Using Co-Moving Conservation-Regularized Physics-Informed Neural Networks

Ahmed Fathi, Baraa Ahmed, A. Atteya

Google Colab and Python code for generating an ETDRK4 numerical reference, training matched laboratory-frame and co-moving-frame physics-informed neural networks, applying PDE-dominant and conservation-aware optimization stages, computing diagnostics, and reproducing the main and…

View free PDFSource page
openalexZenodo (CERN European Organization for Nuclear Research)2026-07-26

From Differential Equations to Deep Networks: A Unified Applied Mathematics and Computer Science Framework for Physics-Informed Computational Mechanics

Md.Nimur Rahman Durjoy

Here’s a line that’s been true for a while now but that we don’t talk about enough: the oldwalls between pure mathematical analysis, numerical computation, and mechanical modelingare quietly coming down, and modern scientific machine learning is basically the wreckingball. In thi…

View free PDFSource page
openalexZenodo (CERN European Organization for Nuclear Research)2026-07-26

Electrical resistivity tomography surveys, trained physics-informed neural network models and code for amortized ERT inversion along Route Regionale 707, Moroccan Middle Atlas

Rajae Ajana

This deposit contains the field data, synthetic training datasets, trained network weights and analysis code supporting the article "Physics-informed neural network inversion of electrical resistivity tomography data: amortized optimization with field validation in the Moroccan M…

View free PDFSource page
openalexZenodo (CERN European Organization for Nuclear Research)2026-07-24

Bayesian-Optimized Physics-Informed Neural Networks for the FitzHugh-Nagumo Model

Bogdan Miličević, N Filipovic

Physics-Informed Neural Networks (PINNs) offer a promising bridge between deep learning and biophysical modeling by embedding differential equations directly into the learning process. This paper explores an automated framework using Bayesian Optimization (BO) and PINNs in order…

View free PDFSource page
openalexZenodo (CERN European Organization for Nuclear Research)2026-07-25

# Artificial Intelligence in Metallurgical Engineering: A Comprehensive Review of Applications, Challenges, and Future Direction

Sudhakar Geruganti

## ALTERNATIVE TITLES ### Alternative Title 1 (Comprehensive)**"AI-Driven Transformation in Metallurgical Engineering: From Microstructure Analysis to Smart Manufacturing and Sustainable Production"** ### Alternative Title 2 (Process-Focused)**"Machine Learning and Deep Learning…

View free PDFSource page
openalexZenodo (CERN European Organization for Nuclear Research)2026-07-25

# Artificial Intelligence-Enabled Quantification of Cube and Goss Textures in Polycrystalline Materials: A Comprehensive Review of Machine Learning, Deep Learning, and EBSD-Based Characterization Approaches

Sudhakar Geruganti

## ALTERNATIVE TITLES ### Alternative Title 1 (Comprehensive)**"AI-Driven Analysis of Cube {100}<001> and Goss {110}<001> Textures: Machine Learning, Deep Learning, and Generative Models for Crystallographic Texture Quantification in Metallurgical Engineering"** ### Alternative T…

View free PDFSource page