Conditional Mean Flows for Frequency-Domain Scattered Wavefield Generation
Abstract
Frequency-domain wavefield modeling underpins seismic imaging and inversion, where repeated Helmholtz solutions become costly as velocity models, source locations, and frequencies vary. Neural operators learn mappings between physical fields, and their accuracy can decline outside the training distribution. Diffusion models learn wavefield distributions and generally require repeated sampling evaluations to produce accurate fields. A conditional mean flows model is introduced for complex scattered-wavefield generation. The analytical background wavefield serves as a condition alongside the velocity model, supplying spatially organized information about the source location and frequency. Learning the scattered wavefield avoids the point-source singularity in the target and supports generalization across these physical conditions. The model estimates the average transport velocity over a finite interval connecting Gaussian noise to the scattered field. This representation captures the accumulated displacement and permits both full-interval one-step generation and interval-partitioned sampling. Trained on OpenFWI, the model is evaluated on in-distribution and out-of-distribution velocity structures, frequency generalization, and inference efficiency. In most evaluated settings, one-step mean flows sampling achieves lower wavefield error than multi-step diffusion. Because each sampling step requires another network evaluation, accurate one-step generation also reduces inference time. These results indicate that mean flows can provide accurate and efficient frequency-domain scattered-wavefield generation.
Related
- Conditional Flow Matching for Generation of 3D Multi-variable Instantaneous Urban Microclimate Fields
- Earthquake Seismogram Denoising Across Time, Time‐Frequency, and Hybrid Domain Approaches
- Frequency‐Domain Airborne Electromagnetic Imaging of Permafrost Reveals Heterogeneity and Drivers of Permafrost Change in the Discontinuous Permafrost Zone of Northwestern Canada
- Scattered data for the Scattered Islands: presence of the Yellow-bellied sea snake Hydrophis platura on Tromelin and first record for Juan de Nova, Western Indian Ocean Region
- Conditional reliability of gridded rainfall datasets in Nigeria
BibTeX
@misc{lai2026conditional,
title = {{Conditional Mean Flows for Frequency-Domain Scattered Wavefield Generation}},
author = {Lipeng Lai and Shaohuan Zu},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.31071},
eprint = {2610.31071},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.31071},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Conditional Mean Flows for Frequency-Domain Scattered Wavefield Generation
AU - Lipeng Lai
AU - Shaohuan Zu
PY - 2026
DA - 2026/10/22
PB - arXiv
DO - 10.48550/arxiv.2610.31071
UR - https://arxiv.org/abs/2610.31071
AB - Frequency-domain wavefield modeling underpins seismic imaging and inversion, where repeated Helmholtz solutions become costly as velocity models, source locations, and frequencies vary. Neural operators learn mappings between physical fields, and their accuracy can decline outside the training distribution. Diffusion models learn wavefield distributions and generally require repeated sampling evaluations to produce accurate fields. A conditional mean flows model is introduced for complex scattered-wavefield generation. The analytical background wavefield serves as a condition alongside the velocity model, supplying spatially organized information about the source location and frequency. Learning the scattered wavefield avoids the point-source singularity in the target and supports generalization across these physical conditions. The model estimates the average transport velocity over a finite interval connecting Gaussian noise to the scattered field. This representation captures the accumulated displacement and permits both full-interval one-step generation and interval-partitioned sampling. Trained on OpenFWI, the model is evaluated on in-distribution and out-of-distribution velocity structures, frequency generalization, and inference efficiency. In most evaluated settings, one-step mean flows sampling achieves lower wavefield error than multi-step diffusion. Because each sampling step requires another network evaluation, accurate one-step generation also reduces inference time. These results indicate that mean flows can provide accurate and efficient frequency-domain scattered-wavefield generation.
ER -