Neptuna: A Comprehensive Machine Learning Framework for Benchmarking Complex Multiphase Flows

2026-07-24

AI Summary

This work addresses the challenge of constructing reliable surrogate models for compressible multiphase flows, where shock waves and material interfaces coexist. It presents the first large-scale benchmark specifically designed for shock-driven multiphase flows, comprising 2.4 TB of high-fidelity 2D and 3D data encompassing canonical scenarios such as bubble collapse and droplet breakup. The authors propose a composite loss function integrating Sobolev, interface-aware, and structure-aware terms, combined with SoftAdapt and GradNorm for efficient adaptive weighting. A systematic evaluation of diverse architectures—including convolutional networks, spectral methods, Transformers, and pre-trained PDE foundation models—demonstrates that the composite loss significantly enhances interface preservation and spectral fidelity. SoftAdapt delivers the most stable performance gains with negligible computational overhead, while no single model universally outperforms others across all tasks.

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 2.4 TB of high-fidelity 2D and 3D datasets \footnote{Dataset repo: https://huggingface.co/FluidVerse. Dataset sample videos, metadata.json, inference rollout plots from autoregressive rollout of the trained baselines are provided in the supplementary\_material.zip } featuring shock-induced bubble collapse and droplet breakup. We evaluate diverse surrogate model families on our benchmarking framework: Neptuna \footnote{Benchmarking repo: https://anonymous.4open.science/r/neptuna-A4E3}, including convolutional, spectral, transformer-based, and pre-trained PDE foundation models. Beyond standard MSE training, we investigate composite losses combining MSE with Sobolev, interface-aware, and structure-aware terms, together 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.

Problem

• compressible multiphase flows
• shock-driven flows
• machine learning surrogates
• interface deformation
• multiscale dynamics

Innovation

• multiphase flows
• machine learning surrogate
• composite loss
• adaptive loss balancing
• shock-driven dynamics