Neptuna: A Comprehensive Machine Learning Framework for Benchmarking Complex Multiphase Flows
Abstract
Compressible multiphase flows involving shocks and material interfaces arise in applications such as bubble collapse and droplet breakup, where strong nonlinear interactions produce complex interface deformation, mixing, and multiscale dynamics. Developing reliable machine learning surrogates for these flows remains challenging due to the simultaneous presence of compressibility, sharp discontinuities, and multiphase effects. In this work, we introduce the first large-scale benchmark specifically designed for shock-driven compressible multiphase flows, comprising 3 TB of high-fidelity 2D and 3D datasets featuring shock-induced bubble collapse and droplet breakup. We evaluate diverse surrogate model families on our benchmarking framework: Neptuna, including convolutional, spectral, transformer-based, and pre-trained PDE foundation models. Beyond standard MSE training, we investigate composite losses that combine MSE with Sobolev, interface-aware, and structure-aware terms, along with adaptive loss balancing using SoftAdapt and GradNorm. Evaluation includes pointwise, spectral, feature-focused, structural, and physics-informed metrics. Results show that no single model performs best across all datasets and metrics, while composite losses significantly improve interface preservation and spectral fidelity. Among adaptive weighting strategies, SoftAdapt provides the most consistent improvements with almost no overhead compared to MSE-only training. The benchmarking repository and the dataset are made available at \url{https://tumaer.github.io/Neptuna/}
Related
- Scalable machine learning framework for multiphase identification from powder X-ray diffraction
- PoreML: A Data-Driven Framework for Learning Multiphase Flow in Porous Media
- XDecomposer: Learning Prior-Free Set Decomposition for Multiphase X-ray Diffraction
- QBugLM: An Agentic Benchmarking Framework for LLM-based Quantum Software Debugging
- Benchmarking machine-learning interatomic potentials using classical force fields
BibTeX
@misc{ramachandran2026neptuna,
title = {{Neptuna: A Comprehensive Machine Learning Framework for Benchmarking Complex Multiphase Flows}},
author = {Harish Ramachandran and Björn Kimpel and Thomas Paula and Josef Winter and Steffen Schmidt and Nikolaus Adams},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2607.22280},
eprint = {2607.22280},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2607.22280},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Neptuna: A Comprehensive Machine Learning Framework for Benchmarking Complex Multiphase Flows
AU - Harish Ramachandran
AU - Björn Kimpel
AU - Thomas Paula
AU - Josef Winter
AU - Steffen Schmidt
AU - Nikolaus Adams
PY - 2026
DA - 2026/10/18
PB - arXiv
DO - 10.48550/arxiv.2607.22280
UR - https://arxiv.org/abs/2607.22280
AB - Compressible multiphase flows involving shocks and material interfaces arise in applications such as bubble collapse and droplet breakup, where strong nonlinear interactions produce complex interface deformation, mixing, and multiscale dynamics. Developing reliable machine learning surrogates for these flows remains challenging due to the simultaneous presence of compressibility, sharp discontinuities, and multiphase effects. In this work, we introduce the first large-scale benchmark specifically designed for shock-driven compressible multiphase flows, comprising 3 TB of high-fidelity 2D and 3D datasets featuring shock-induced bubble collapse and droplet breakup. We evaluate diverse surrogate model families on our benchmarking framework: Neptuna, including convolutional, spectral, transformer-based, and pre-trained PDE foundation models. Beyond standard MSE training, we investigate composite losses that combine MSE with Sobolev, interface-aware, and structure-aware terms, along with adaptive loss balancing using SoftAdapt and GradNorm. Evaluation includes pointwise, spectral, feature-focused, structural, and physics-informed metrics. Results show that no single model performs best across all datasets and metrics, while composite losses significantly improve interface preservation and spectral fidelity. Among adaptive weighting strategies, SoftAdapt provides the most consistent improvements with almost no overhead compared to MSE-only training. The benchmarking repository and the dataset are made available at \url{https://tumaer.github.io/Neptuna/}
ER -