dorsal/arxiv
View SchemaShallow-KAN Based Solution of Moving Boundary PDEs
| Authors | Tarus Pande, V M S K Minnikanti, Shyamprasad Karagadde |
|---|---|
| Categories | |
| ArXiv ID | 2601.09818vv1 |
| URL | https://arxiv.org/abs/2601.09818 |
| License | http://creativecommons.org/licenses/by/4.0/ |
Abstract
Kolmogorov-Arnold Networks (KANs) require significantly smaller architectures compared to multilayer perceptron (MLP)-based approaches, while retaining expressive power through spline-based activations. We propose a shallow KAN framework that directly approximates the temperature distribution T(x,t) and the moving interface $\Gamma(t)$, enforcing the governing PDEs, phase equilibrium, and Stefan condition through physics-informed residuals. To enhance accuracy, we employ interface-focused collocation resampling. Numerical experiments in one and two dimensions show that the framework achieves accurate reconstructions of both temperature fields and interface dynamics, highlighting the potential of KANs as a compact and efficient alternative for moving boundary PDEs. First, we validate the model with semi-infinite analytical solutions. Subsequently, the model is extended to 2D using a level-set based formulation for interface propagation, which is solved within the KAN framework. This work demonstrates that KANs are capable of solving complex moving boundary problems without the need for measurement data.
{
"annotation_id": "6ac64139-c515-4b70-9173-296a0ee8f74a",
"date_created": "2026-02-17T05:53:24.053000Z",
"date_modified": "2026-02-17T05:53:24.053000Z",
"file_hash": "59ea042860e49dae8c80a9d74dbd7ce3b8b4c43a4c812b3eff2294580101586e",
"private": false,
"record": {
"abstract": "Kolmogorov-Arnold Networks (KANs) require significantly smaller architectures compared to multilayer perceptron (MLP)-based approaches, while retaining expressive power through spline-based activations. We propose a shallow KAN framework that directly approximates the temperature distribution T(x,t) and the moving interface $\\Gamma(t)$, enforcing the governing PDEs, phase equilibrium, and Stefan condition through physics-informed residuals. To enhance accuracy, we employ interface-focused collocation resampling. Numerical experiments in one and two dimensions show that the framework achieves accurate reconstructions of both temperature fields and interface dynamics, highlighting the potential of KANs as a compact and efficient alternative for moving boundary PDEs. First, we validate the model with semi-infinite analytical solutions. Subsequently, the model is extended to 2D using a level-set based formulation for interface propagation, which is solved within the KAN framework. This work demonstrates that KANs are capable of solving complex moving boundary problems without the need for measurement data.",
"arxiv_id": "2601.09818",
"authors": [
"Tarus Pande",
"V M S K Minnikanti",
"Shyamprasad Karagadde"
],
"categories": [
"math-ph",
"cs.NA",
"math.MP",
"math.NA",
"physics.comp-ph"
],
"license": "http://creativecommons.org/licenses/by/4.0/",
"title": "Shallow-KAN Based Solution of Moving Boundary PDEs",
"url": "https://arxiv.org/abs/2601.09818",
"version": "v1"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "da0895f5-1907-488a-9f5e-f303471b53be",
"id": "arXiv Dataset",
"type": "Model",
"variant": "snapshot-2026-01-17",
"version": "0.1.0"
},
"user_id": 1000002
}