Measuring and Mitigating Forgetting when Extending a PDE Foundation Model to Simulated Flow Regimes

Abstract Foundation models for partial differential equations adapt efficiently to new tasks, but their effect on the original pretraining distribution is rarely measured. We study this retention problem on Poseidon, a neural operator pretrained on periodicdomain compressible Euler and incompressible Navier–Stokes families. We adapt it to two regimes absent from pretraining, a compressible Kelvin–Helmholtz flow and wall-bounded flow past a circular cylinder. Fine-tuning adapts efficiently but degrades the pretraining families. LoRA still forgets substantially, and freezing the whole trunk reduces forgetting without preventing it. We show that low-weight replay and joint training resolve this trade-off, extending the model to the new regimes while keeping the pretraining families near their original accuracy. These results suggest that lightweight replay can extend a PDE foundation model while substantially preserving its existing capabilities.

Authors

Publication Details

Journal
Research Square
Published
2026-11-17
DOI
https://doi.org/10.21203/rs.3.rs-11745898/v1
Primary Topic
Model Reduction and Neural Networks
Type
preprint

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Measuring and Mitigating Forgetting when Extending a PDE Foundation Model to Simulated Flow Regimes

Shoaib Noor, Wasif Masood
1 links
Research Square
Model Reduction and Neural Networks
preprint

Measuring and Mitigating Forgetting when Extending a PDE Foundation Model to Simulated Flow Regimes

Shoaib Noor, Wasif Masood
preprint en

Abstract

Abstract Foundation models for partial differential equations adapt efficiently to new tasks, but their effect on the original pretraining distribution is rarely measured. We study this retention problem on Poseidon, a neural operator pretrained on periodicdomain compressible Euler and incompressible Navier–Stokes families. We adapt it to two regimes absent from pretraining, a compressible Kelvin–Helmholtz flow and wall-bounded flow past a circular cylinder. Fine-tuning adapts efficiently but degrades the pretraining families. LoRA still forgets substantially, and freezing the whole trunk reduces forgetting without preventing it. We show that low-weight replay and joint training resolve this trade-off, extending the model to the new regimes while keeping the pretraining families near their original accuracy. These results suggest that lightweight replay can extend a PDE foundation model while substantially preserving its existing capabilities.

Research Square
Österreichische Forschungsförderungsgesellschaft
Model Reduction and Neural Networks
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon