Mitigating Propagation Failures in Physics-informed Neural Networks using Retain-Resample-Release (R3) Sampling
About
Despite the success of physics-informed neural networks (PINNs) in approximating partial differential equations (PDEs), PINNs can sometimes fail to converge to the correct solution in problems involving complicated PDEs. This is reflected in several recent studies on characterizing the "failure modes" of PINNs, although a thorough understanding of the connection between PINN failure modes and sampling strategies is missing. In this paper, we provide a novel perspective of failure modes of PINNs by hypothesizing that training PINNs relies on successful "propagation" of solution from initial and/or boundary condition points to interior points. We show that PINNs with poor sampling strategies can get stuck at trivial solutions if there are propagation failures, characterized by highly imbalanced PDE residual fields. To mitigate propagation failures, we propose a novel Retain-Resample-Release sampling (R3) algorithm that can incrementally accumulate collocation points in regions of high PDE residuals with little to no computational overhead. We provide an extension of R3 sampling to respect the principle of causality while solving time-dependent PDEs. We theoretically analyze the behavior of R3 sampling and empirically demonstrate its efficacy and efficiency in comparison with baselines on a variety of PDE problems.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| PDE solving | Klein-Gordon equation | Relative L2 Error1.5 | 31 | |
| Forward PDE solving | Helmholtz | Relative Error0.0267 | 26 | |
| Forward PDE solving | Forward benchmarks 5 PDEs | P(A<B)100 | 21 | |
| Forward PDE problem solving | Burgers | Relative L2 Error0.0046 | 19 | |
| Forward PDE solving | Allen–Cahn 10K-epoch | Relative L2 Error1.6 | 16 | |
| Forward PDE solving | Helmholtz 10K-epoch | Relative L2 Error1.4 | 16 | |
| Forward PDE solving | Klein-Gordon 10K-epoch | Relative L2 Error2.5 | 16 | |
| Forward PDE solving | Burgers 10K-epoch | Relative L2 Error4.7 | 16 | |
| Forward PDE solving | Conv-Diff 10K-epoch | Relative L2 Error3.55 | 16 | |
| Solving coupled thermoelastic PDEs | Natural coupled thermoelastic K = 4 1.0 (val) | L2 Error1.04 | 10 |