Research map / Probabilistic and scientific modeling / Differential equations and PDEs
Self-Scalable Tanh (Stan): Multi-Scale Solutions for Physics-Informed Neural Networks
Raghav Gnanasambandam, Bo Shen, Jihoon Chung, Xubo Yue, Zhenyu Kong
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use Stan instead of tanh | PINN/MLP · Smooth function regression | test MSE: 1369.60 -> 185.46 | Improved |
| use Stan instead of N-LAAF | PINN/MLP · Smooth function regression | test MSE: 988.18 -> 185.46 | Improved |
| use Stan instead of tanh | PINN/MLP · Discontinuous function regression | test MSE: 47.77 -> 23.97 | Improved |
| use Stan instead of tanh | PINN · Forward ODE d2u/dx2+du/dx=6u | relative error: 0.8757 -> 0.0005 | Improved |
| use Stan instead of N-LAAF | PINN · Forward ODE d2u/dx2+du/dx=6u | relative error: 0.6449 -> 0.0005 | Improved |
| use Stan instead of tanh | PINN · Forward ODE du/dx=cos(0.01x) | relative error: 0.0088 -> 0.0000 | Improved |
| use Stan instead of tanh | PINN · Klein-Gordon | relative error: 0.7725 -> 0.3541 | Improved |
| use Stan instead of N-LAAF | PINN · Klein-Gordon | relative error: 0.5936 -> 0.3541 | Improved |
| use Stan instead of tanh | PINN · Inverse heat conduction | relative error: 0.3927 -> 0.0928 | Improved |
| use Stan instead of N-LAAF | PINN · Inverse heat conduction | test MSE: 564.08 -> 30.36 | Improved |
| vary initial beta over ten values from 0.25 to 1.15 (ablation) | Stan · Function regression | relative error | No clear effect |
| use N-LAAF instead of tanh | PINN · Forward ODE du/dx=cos(0.01x) | relative error: 0.0088 -> 0.0063 | Improved |
Closest papers by meaning
- PINNacle: A Comprehensive Benchmark of Physics-Informed Neural Networks for Solving PDEs (NeurIPS 2024 · similarity 0.88)
- Efficient Error Certification for Physics-Informed Neural Networks (ICML 2024 · similarity 0.87)
- Sub-Sequential Physics-Informed Learning with State Space Model (ICML 2025 · similarity 0.86)
- The Challenges of the Nonlinear Regime for Physics-Informed Neural Networks (NeurIPS 2024 · similarity 0.86)
- RoPINN: Region Optimized Physics-Informed Neural Networks (NeurIPS 2024 · similarity 0.86)
- Gradient Alignment in Physics-informed Neural Networks: A Second-Order Optimization Perspective (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
