Preprint · 2026 · Open access
Learning Physics from an Imperfect Ancestor
arXiv (Cornell University) · Published 21 September 2026 · DOI 10.48550/arxiv.2609.24947
In brief
“Learning Physics from an Imperfect Ancestor” is a 2026 preprint by S. Mohammad Mousavi and 3 co-authors, published in arXiv (Cornell University). OpenAlex counts 1 citations as of 16 November 2026. It is open access: the full text is free to read at arXiv (Cornell University).
- Citations
- 1
- Most cited in
- 2026 (1)
- References
- 7
- Authors
- 4
Abstract
Neural operators evaluate parametric partial differential equations cheaply but degrade sharply outside their training distribution. Physics-informed neural networks avoid dependence on labeled data, yet their optimization can be basin-fragile: when the governing residual admits multiple solutions, a PINN trained from scratch may converge to a physically incorrect state despite achieving a small residual. We show that these failure modes can be addressed jointly: an imperfect NO provides the structural prior needed to place a PINN in the correct solution basin, while the PDE residual refines the solution beyond the operator's accuracy. We introduce a three-stage framework that freezes the spatial basis of a physics-informed NO, extrapolates its solution branch to an out-of-distribution parameter using a polynomial continuation prior, and distills the resulting field into a fresh PINN. The NO need not be accurate at the target; it transfers solution-branch information, while PDE residual minimization in the PINN governs convergence. We evaluate the framework on three nonlinear PDEs: 1D viscous Burgers, 2D steady Allen-Cahn near a pitchfork bifurcation, and 2D steady lid-driven cavity flow. For Allen-Cahn, where the trivial solution satisfies the PDE residual exactly, a standard PINN collapses to the trivial zero branch, whereas distillation from the crude extrapolated operator recovers the non-trivial branch that matches the finite-difference reference. For the lid-driven cavity, extrapolating to a Reynolds number of Re = 3200 accelerates convergence to the correct physical state, achieving competitive accuracy using fewer parameters and optimization steps than recent literature baselines. These results establish a simple principle: an NO need not accurately predict the solution to be useful; it only needs to identify the correct basin from which PINN optimization can recover it.
Cite this paper
Formatted from the catalogue record. Check names, capitals and page numbers against the paper before you submit.
APA 7
Mousavi, S. M., Kadeethum, T., Bouklas, N., & Goswami, S. (2026). Learning Physics from an Imperfect Ancestor. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2609.24947
MLA 9
Mousavi, S. Mohammad, et al. “Learning Physics from an Imperfect Ancestor.” arXiv (Cornell University), 2026. https://doi.org/10.48550/arxiv.2609.24947.
Chicago
Mousavi, S. Mohammad, Teeratorn Kadeethum, Nikolaos Bouklas, and Somdatta Goswami. 2026. “Learning Physics from an Imperfect Ancestor.” arXiv (Cornell University). https://doi.org/10.48550/arxiv.2609.24947.
Harvard
Mousavi, S. M. et al. (2026) ‘Learning Physics from an Imperfect Ancestor’, arXiv (Cornell University). doi:10.48550/arxiv.2609.24947.
IEEE
S. M. Mousavi, T. Kadeethum, N. Bouklas, and S. Goswami, “Learning Physics from an Imperfect Ancestor,” arXiv (Cornell University), 2026, doi: 10.48550/arxiv.2609.24947.
BibTeX
@misc{mousavi2026learning,
title = {{Learning Physics from an Imperfect Ancestor}},
author = {Mousavi, S. Mohammad and Kadeethum, Teeratorn and Bouklas, Nikolaos and Goswami, Somdatta},
howpublished = {arXiv (Cornell University)},
year = {2026},
doi = {10.48550/arxiv.2609.24947},
url = {https://doi.org/10.48550/arxiv.2609.24947}
}
Download for a reference manager: BibTeX (.bib) · RIS for Zotero, Mendeley and EndNote (.ris)
Topic and keywords
Physical Sciences › Physics and Astronomy › Statistical and Nonlinear Physics › Model Reduction and Neural Networks
- physics-informed neural networks
- neural operators
- knowledge distillation
- parametric partial differential equations
- Allen-Cahn equation
- viscous Burgers equation
- lid-driven cavity flow
- pitchfork bifurcation
Work with this paper
Most-cited papers in Physics and Astronomy
- Generalized Gradient Approximation Made SimpleJohn P. Perdew, Kieron Burke, Matthias Ernzerhof · Physical Review Letters · 1996 · 217,345 citations
- Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis setGeorg Kresse, Jürgen Furthmüller · Physical review. B, Condensed matter · 1996 · 125,164 citations
- Density-functional thermochemistry. III. The role of exact exchangeAxel D. Becke · The Journal of Chemical Physics · 1993 · 104,772 citations
- Development of the Colle-Salvetti correlation-energy formula into a functional of the electron densityChengteh Lee, Weitao Yang, Robert G. Parr · Physical review. B, Condensed matter · 1988 · 101,730 citations
- Projector augmented-wave methodPeter E. Blöchl · Physical review. B, Condensed matter · 1994 · 93,520 citations
Questions about this paper
How do I cite this paper in APA style?
Mousavi, S. M., Kadeethum, T., Bouklas, N., & Goswami, S. (2026). Learning Physics from an Imperfect Ancestor. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2609.24947
How many times has this paper been cited?
1 times, according to OpenAlex on 16 November 2026. Google Scholar and Scopus count different sets of sources, so their numbers differ.
Is this paper free to read?
Yes. It is open access: read the free full text at arXiv (Cornell University).
Who wrote this paper?
S. Mohammad Mousavi, Teeratorn Kadeethum, Nikolaos Bouklas, Somdatta Goswami.
What is the DOI of this paper?
10.48550/arxiv.2609.24947. The permanent link is https://doi.org/10.48550/arxiv.2609.24947.
Bibliographic data from OpenAlex (CC0), last checked 16 November 2026. ResearchGuru does not host this paper; the links go to the publisher or the repository that does.