High-Order Variable-Scaled Energetic Variational Neural Networks for Phase-Field Gradient Flows
Abstract
In this paper, we develop a high-order variable-scaled energetic variational neural network (VS-EVNN) method for phase-field gradient flows. Each discretization stage uses an updated-state network whose spatial inputs are centered and divided by the interface scale, while the sampled energy and movement penalties remain unchanged relative to the baseline EVNN formulation. This rescaling is motivated by a coordinate change in the energy that normalizes the gradient coefficient of the transformed functional. We combine the scaled representation with high-order, multi-stage variational extrapolation schemes and a heuristic adaptive time-stepping controller. Each stage is solved as a network minimization problem, in which the earlier stages serve as fixed anchors on the quadrature grid, yielding a neural realization of these multi-stage variational schemes, including a mobility-weighted, mass-conserving version for Cahn--Hilliard flows. The method is tested on one- and two-dimensional Allen--Cahn and Cahn--Hilliard flows, the latter in the mass-conserving metric. Under the reported optimization settings, the scaled networks yield lower solution errors than the unscaled EVNN baseline, and the higher-order schemes improve the accuracy in the test examples.
Related
- Energetic Variational Deep Learning for Solving High-Dimensional Stationary Fokker-Planck Equations
- Scaled-Gradient Hard Thresholding Algorithms for Sparse Signal Recovery
- Metric gradient flows and variational principles
- Amortized Structured Stochastic Variational Inference for Gaussian Process Latent Variable Models
- Variational Bayes for Variable Selection with Adaptive Sharing across Populations and Responses
BibTeX
@misc{jing2026highorder,
title = {{High-Order Variable-Scaled Energetic Variational Neural Networks for Phase-Field Gradient Flows}},
author = {Xiaobo Jing and Leiyi Dong and Jia Zhao},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.25489},
eprint = {2609.25489},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.25489},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - High-Order Variable-Scaled Energetic Variational Neural Networks for Phase-Field Gradient Flows
AU - Xiaobo Jing
AU - Leiyi Dong
AU - Jia Zhao
PY - 2026
DA - 2026/09/21
PB - arXiv
DO - 10.48550/arxiv.2609.25489
UR - https://arxiv.org/abs/2609.25489
AB - In this paper, we develop a high-order variable-scaled energetic variational neural network (VS-EVNN) method for phase-field gradient flows. Each discretization stage uses an updated-state network whose spatial inputs are centered and divided by the interface scale, while the sampled energy and movement penalties remain unchanged relative to the baseline EVNN formulation. This rescaling is motivated by a coordinate change in the energy that normalizes the gradient coefficient of the transformed functional. We combine the scaled representation with high-order, multi-stage variational extrapolation schemes and a heuristic adaptive time-stepping controller. Each stage is solved as a network minimization problem, in which the earlier stages serve as fixed anchors on the quadrature grid, yielding a neural realization of these multi-stage variational schemes, including a mobility-weighted, mass-conserving version for Cahn--Hilliard flows. The method is tested on one- and two-dimensional Allen--Cahn and Cahn--Hilliard flows, the latter in the mass-conserving metric. Under the reported optimization settings, the scaled networks yield lower solution errors than the unscaled EVNN baseline, and the higher-order schemes improve the accuracy in the test examples.
ER -