GeoSim: Benchmarking Multi-Field Neural Operators for Geodynamics
Abstract
Learning reliable surrogates for geodynamic models requires benchmarks that resolve coupled fields and temporal stability. We present GeoSim, a benchmark of high-resolution two-dimensional mantle-plume simulations generated with I2VIS, comprising 237 trajectories across two parameter regimes, with six coupled fields, physical timestamps, and trajectory-level splits. GeoSim defines short-window prediction and autonomous rollout, and evaluates representative neural operators field by field. Across models, we observe a pronounced scalar--velocity gap: scalar fields are reconstructed accurately while velocity errors remain large, and short-window accuracy does not imply stable rollout. We also evaluate MFGNO (Multi-Field Graph Neural Operator), a joint multi-field graph baseline designed for this benchmark; on GeoSim-Plume1, under a fixed per-field objective, it reduces horizontal and vertical velocity errors by 58.4% and 38.5% relative to an independent-field graph baseline, while an explicit field-difference input shows no measurable benefit at this scale. GeoSim provides a reproducible benchmark for field-wise diagnosis and long-horizon evaluation of geodynamic surrogates.
Related
- How Sparse and How Noisy? Systematic Benchmarking of Inverse Physics-Informed Neural Networks for Manning Friction Estimation in Shallow Water Equations
- High-performance geodynamics on GPUs using the PETSc CUDA backend (GAUZZ v1.0.0)
- Neural Operators Reveal Scale-Dependent Limits on Recovering Greenland Ice Loss from Sea Level Fingerprints
- Optimization of UV-C irradiation time for biologically treated saline wastewater using artificial neural networks and a hybrid multi-criteria decision-making framework
- Spherical Physics-informed Neural Operator with Multi-scale Coupling for Meteorological Downscaling
BibTeX
@misc{liu2026geosim,
title = {{GeoSim: Benchmarking Multi-Field Neural Operators for Geodynamics}},
author = {Hailong Liu and Dinghui Yang and Mengxue Liu and Rui Qi and Meng Zhao and Yihang Tang and Jin Wen and Zhilong Fang},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.04517},
eprint = {2611.04517},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.04517},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - GeoSim: Benchmarking Multi-Field Neural Operators for Geodynamics
AU - Hailong Liu
AU - Dinghui Yang
AU - Mengxue Liu
AU - Rui Qi
AU - Meng Zhao
AU - Yihang Tang
AU - Jin Wen
AU - Zhilong Fang
PY - 2026
DA - 2026/11/03
PB - arXiv
DO - 10.48550/arxiv.2611.04517
UR - https://arxiv.org/abs/2611.04517
AB - Learning reliable surrogates for geodynamic models requires benchmarks that resolve coupled fields and temporal stability. We present GeoSim, a benchmark of high-resolution two-dimensional mantle-plume simulations generated with I2VIS, comprising 237 trajectories across two parameter regimes, with six coupled fields, physical timestamps, and trajectory-level splits. GeoSim defines short-window prediction and autonomous rollout, and evaluates representative neural operators field by field. Across models, we observe a pronounced scalar--velocity gap: scalar fields are reconstructed accurately while velocity errors remain large, and short-window accuracy does not imply stable rollout. We also evaluate MFGNO (Multi-Field Graph Neural Operator), a joint multi-field graph baseline designed for this benchmark; on GeoSim-Plume1, under a fixed per-field objective, it reduces horizontal and vertical velocity errors by 58.4% and 38.5% relative to an independent-field graph baseline, while an explicit field-difference input shows no measurable benefit at this scale. GeoSim provides a reproducible benchmark for field-wise diagnosis and long-horizon evaluation of geodynamic surrogates.
ER -