# SNODE: Spectral Discretization of Neural ODEs for System Identification

arXiv: https://arxiv.org/abs/1906.07038 · HTML: https://mlsanity.com/paper/1906.07038/

## Key facts

- Cited by 26 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 17 Jun 2019.

## Built on by (most cited first, 26 of 26)

- [Liquid Structural State-Space Models](https://mlsanity.com/paper/2209.12951/) (2022)
- [Machine learning structure preserving brackets for forecasting irreversible processes](https://mlsanity.com/paper/2106.12619/) (2021)
- [MALI: A memory efficient and reverse accurate integrator for Neural ODEs](https://mlsanity.com/paper/2102.04668/) (2021)
- [Heavy Ball Neural Ordinary Differential Equations](https://mlsanity.com/paper/2110.04840/) (2021)
- [Deep networks for system identification: a Survey](https://mlsanity.com/paper/2301.12832/) (2023)
- [Constructing Neural Network-Based Models for Simulating Dynamical Systems](https://mlsanity.com/paper/2111.01495/) (2021)
- [Physics-Informed Machine Learning for Modeling and Control of Dynamical Systems](https://mlsanity.com/paper/2306.13867/) (2023)
- [OCT-GAN: Neural ODE-based Conditional Tabular GANs](https://mlsanity.com/paper/2105.14969/) (2021)
- [Neural Delay Differential Equations](https://mlsanity.com/paper/2102.10801/) (2021)
- [Causal Navigation by Continuous-time Neural Networks](https://mlsanity.com/paper/2106.08314/) (2021)
- [Unleashing the Potential of Fractional Calculus in Graph Neural Networks with FROND](https://mlsanity.com/paper/2404.17099/) (2024)
- [Neural Ordinary Differential Equations for Nonlinear System Identification](https://mlsanity.com/paper/2203.00120/) (2022)
- [How Does Momentum Benefit Deep Neural Networks Architecture Design? A Few Case Studies](https://mlsanity.com/paper/2110.07034/) (2021)
- [TO-FLOW: Efficient Continuous Normalizing Flows with Temporal Optimization adjoint with Moving Speed](https://mlsanity.com/paper/2203.10335/) (2022)
- [Advancing Generalization in PINNs through Latent-Space Representations](https://mlsanity.com/paper/2411.19125/) (2024)
- [Neuromorphic Imaging and Classification with Graph Learning](https://mlsanity.com/paper/2309.15627/) (2023)
- [Training Neural ODEs Using Fully Discretized Simultaneous Optimization](https://mlsanity.com/paper/2502.15642/) (2025)
- [Method of Manufactured Learning for Solver-free Training of Neural Operators](https://mlsanity.com/paper/2511.12890/) (2025)
- [Parameterized Temporal Extrapolation of PDEs via Disentangled Latent Dynamics Manifold Fusion](https://mlsanity.com/paper/2603.12676/) (2026)
- [Understanding the Theoretical Foundations of Deep Neural Networks through Differential Equations](https://mlsanity.com/paper/2603.18331/) (2026)
- [Neural Delay Differential Equations: System Reconstruction and Image Classification](https://mlsanity.com/paper/2304.05310/) (2023)
- [Directed Chain Generative Adversarial Networks](https://mlsanity.com/paper/2304.13131/) (2023)
- [LILAD: Learning In-context Lyapunov-stable Adaptive Dynamics Models](https://mlsanity.com/paper/2511.21846/) (2025)
- [Learning partially observed systems with neural Hamiltonian ordinary differential equations](https://mlsanity.com/paper/2605.23510/) (2026)
- [Frequency-Domain Neural ODEs for Modeling Non-Linear Dynamical Systems](https://mlsanity.com/paper/2606.22075/) (2026)
- [Neural Controlled Differential Equations for EMT-Level Surrogate Modeling of Grid-Forming Inverters](https://mlsanity.com/paper/2607.16258/) (2026)

## Abstract (arXiv)

This paper proposes the use of spectral element methods citepcanutospectral1988 for fast and accurate training of Neural Ordinary Differential Equations (ODE-Nets; citealpChen2018NeuralOD) for system identification. This is achieved by expressing their dynamics as a truncated series of Legendre polynomials. The series coefficients, as well as the network weights, are computed by minimizing the weighted sum of the loss function and the violation of the ODE-Net dynamics. The problem is solved by coordinate descent that alternately minimizes, with respect to the coefficients and the weights, two unconstrained sub-problems using standard backpropagation and gradient methods. The resulting optimization scheme is fully time-parallel and results in a low memory footprint. Experimental comparison to standard methods, such as backpropagation through explicit solvers and the adjoint technique citepChen2018NeuralOD, on training surrogate models of small and medium-scale dynamical systems shows that it is at least one order of magnitude faster at reaching a comparable value of the loss function. The corresponding testing MSE is one order of magnitude smaller as well, suggesting generalization capabilities increase.