Skip to content
Log in
Open access

Comparative Study of Physics-Informed Neural Networks (PINNs) and Optimizing a Discrete Loss (ODIL) for Forward and Inverse Problems in Unsteady Non-Linear Partial Differential Equations

Vasco L. Carvalho José M. C. Pereira
Jul 2026 · Computation · 0 citations · 40 references

TL;DR

This work systematically compares two state-of-the-art frameworks-Physics-Informed Neural Networks (PINNs) and Optimizing a Discrete Loss (ODIL) across benchmark elliptic, hyperbolic, and parabolic problems, culminating in a challenging inverse source reconstruction task.

Abstract

This work systematically compares two state-of-the-art frameworks-Physics-Informed Neural Networks (PINNs) and Optimizing a Discrete Loss (ODIL)-across benchmark elliptic (Poisson), hyperbolic (Wave), and parabolic (2D diffusion-reaction with unknown source term) problems, culminating in a challenging inverse source reconstruction task: inferring a space-time-varying heat source that enforces a prescribed temperature profile along a moving line. PINNs enforce physics via continuous residuals and automatic differentiation, while ODIL discretizes the PDE on a grid and optimizes discrete field values directly. For the tested configurations and benchmarks, ODIL achieves lower errors and substantially lower per-epoch and total computational cost compared to the baseline PINN implementations considered. Multi-scale Fourier Feature Networks (MsFFN) and Spatio-temporal Multi-scale Fourier Features Networks (STMsFFN) significantly improve PINNs performance on multi-scale problems, while ODIL with multigrid decomposition achieves comparable or superior accuracy at lower cost. The findings highlight the advantages of discrete optimization approaches over neural-network-based physics enforcement for many PDE problems, offering practical insights into hyper-parameter selection and optimization strategies. This work provides a rigorous head-to-head evaluation and guidance for choosing or combining these frameworks in computational science and engineering applications.

DOI Semantic Scholar
Read PDF

Similar papers

Open access Aug 2026

Data-Guided Physics-Informed Neural Network with Fourier Features Enhancement for Euler-Bernoulli Beam Analysis

The results demonstrate that PINN achieves more accurate and stable full-field vibration reconstructions than conventional PINNs, particularly under conditions involving high-frequency modes, and highlights the potential of hybrid data-physics neural frameworks as an efficient and reliable approach for solving complex PDE-governed dynamical systems.

Hailong Liu, S. Hedayatrasa, Yunpeng Zhu et al. · 0 citations
Preprint Aug 2026

Physics-Informed Stochastic Configuration Machine: A Backpropagation-Free Neural Network with Fast Training for Nonlinear Differential Equations

The Physics-Informed Stochastic Configuration Machine is proposed, a novel backpropagation-free framework for both forward and inverse problems in differential equations that achieves high-fidelity predictive accuracy and robust parameter identification while accelerating the training process by orders of magnitude compared to standard PINNs.

Yueze Song, Zhongzhe Chen, Lihui Cen et al. · 0 citations
Aug 2026

Physics-informed feature decomposition in residual dense block neural networks for incompressible viscous flow

The growing application of physics-informed neural networks (PINNs) for solving parametric partial differential equations (PDEs) in fluid dynamics has demonstrated their potential for modeling complex multiscale flows; however, conventional PINNs often exhibit spectral bias and slow, unstable convergence, limiting accuracy in boundary layers and wakes. This research presents novel physics-informed feature decomposition in residual dense block neural networks (PI-RDB-NN), which embeds physical constraints directly into the network architecture rather than relying solely on soft constraints. PI-RDB-NN uses hierarchical residual dense blocks for multi-scale feature extraction, allocates feature channels to velocity and pressure in a 2:1 ratio consistent with two-dimensional incompressible Navier–Stokes physics, and enforces mass conservation via a learnable divergence-aware projection applied at the feature level. The model is evaluated on National Advisory Committee for Aeronautics (NACA) 0012 airfoil flow at Reynolds numbers (Re)=5000 and Re=1000 using a hybrid loss combining PDE residuals, boundary conditions, and sparse computational fluid dynamics (CFD) data. PI-RDB-NN reduces PDE residual and divergence error by 91.2% and 71.7% vs traditional PINNs (Re=5000) and by 85.5% and 85.4% vs a physics-informed Deep Operator Network (DeepONet) baseline (Re=1000). These physics consistency gains improve aerodynamic force predictions and CFD agreement, confirmed by velocity, wake, and pressure coefficient (Cp) distributions. Consistent accuracy across both Reynolds regimes supports the framework's generality, with three-dimensional and unsteady extensions identified as future work.

Sarmad Iftikhar, Ishfaq Ahmad, Diltaj Ali et al. · 0 citations