Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2603.23861 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910070610067456 |
|---|---|
| author | Yu, Fangzhou Su, Yiqi Lee, Ray Cheng, Shenfeng Ramakrishnan, Naren |
| author_facet | Yu, Fangzhou Su, Yiqi Lee, Ray Cheng, Shenfeng Ramakrishnan, Naren |
| contents | Neural ODEs are increasingly used as continuous-time models for scientific and sensor data, but unconstrained neural ODEs can drift and violate domain invariants (e.g., conservation laws), yielding physically implausible solutions. In turn, this can compound error in long-horizon prediction and surrogate simulation. Existing solutions typically aim to enforce invariance by soft penalties or other forms of regularization, which can reduce overall error but do not guarantee that trajectories will not leave the constraint manifold. We introduce the invariant compiler, a framework that enforces invariants by construction: it treats invariants as first-class types and uses an LLM-driven compilation workflow to translate a generic neural ODE specification into a structure-preserving architecture whose trajectories remain on the admissible manifold in continuous time (and up to numerical integration error in practice). This compiler view cleanly separates what must be preserved (scientific structure) from what is learned from data (dynamics within that structure). It provides a systematic design pattern for invariant-respecting neural surrogates across scientific domains. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_23861 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | An Invariant Compiler for Neural ODEs in AI-Accelerated Scientific Simulation Yu, Fangzhou Su, Yiqi Lee, Ray Cheng, Shenfeng Ramakrishnan, Naren Machine Learning Neural ODEs are increasingly used as continuous-time models for scientific and sensor data, but unconstrained neural ODEs can drift and violate domain invariants (e.g., conservation laws), yielding physically implausible solutions. In turn, this can compound error in long-horizon prediction and surrogate simulation. Existing solutions typically aim to enforce invariance by soft penalties or other forms of regularization, which can reduce overall error but do not guarantee that trajectories will not leave the constraint manifold. We introduce the invariant compiler, a framework that enforces invariants by construction: it treats invariants as first-class types and uses an LLM-driven compilation workflow to translate a generic neural ODE specification into a structure-preserving architecture whose trajectories remain on the admissible manifold in continuous time (and up to numerical integration error in practice). This compiler view cleanly separates what must be preserved (scientific structure) from what is learned from data (dynamics within that structure). It provides a systematic design pattern for invariant-respecting neural surrogates across scientific domains. |
| title | An Invariant Compiler for Neural ODEs in AI-Accelerated Scientific Simulation |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2603.23861 |