LFNO: Bridging Laplace and Fourier via Transient-Steady Decomposition
About
We introduce the Laplace-Fourier Neural Operator (LFNO), a unified framework for modeling dynamical systems across transient and steady-state regimes by integrating the spectral advantages of Laplace and Fourier Neural Operators. LFNO employs a dual-branch architecture that explicitly decomposes system dynamics into transient and steady-state components. We evaluate LFNO on nine benchmarks, including three ODE systems (Duffing, Lorenz, and Pendulum) and six PDE systems (Euler-Bernoulli beam, Heat, Reaction-diffusion, Brusselator, Burgers, and Navier-Stokes). LFNO significantly outperforms existing operators on ODE systems, where transient dynamics dominate, and consistently surpasses LNO while achieving performance competitive with FNO on PDE benchmarks. Furthermore, LFNO offers improved stability and physical interpretability through its component-wise decomposition. These results demonstrate that LFNO provides a robust and unified approach for learning complex dynamical systems across multiple temporal scales.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| PDE solving | Navier-Stokes (NS) problem | Relative L2 Error0.04 | 20 | |
| PDE solving | Burgers' equation | L2 Relative Error0.0026 | 18 | |
| ODE solving | Lorenz (ρ = 10) | L2 Error0.1731 | 6 | |
| ODEs | Lorenz ρ = 5 | L-infinity Error0.2245 | 6 | |
| ODE solving | Duffing c = 0 | L2 Error0.0518 | 3 | |
| ODE solving | Duffing c = 0.5 | L2 Error0.0142 | 3 | |
| ODE solving | Pendulum c = 0.5 | L2 Error0.0315 | 3 | |
| ODEs | Duffing c = 0.5 | L∞ Error0.0218 | 3 | |
| ODEs | Lorenz (ρ = 10) | L-infinity Error0.2646 | 3 | |
| ODEs | Pendulum c = 0.5 | L-infinity Error0.0337 | 3 |