MORPH

PDE Foundation Models with Arbitrary Data Modality

Mahindra Singh Rautela · Alexander Most · Siddharth Mansingh · Bradley C. Love
Alexander Scheinker · Diane Oyen · Nathan Debardeleben · Earl Lawrence · Ayan Biswas

Los Alamos National Laboratory

Overview

MORPH is a foundation model for PDEs that learns across 1D, 2D, and 3D physics data with different fields and resolutions. It can forecast system dynamics and adapt to inverse problems such as property estimation, damage detection, and sparse-field reconstruction. Pretraining improves accuracy and data efficiency, especially when task-specific data are limited.

Architecture

MORPH maps heterogeneous observations to a common representation, reasons over spatial and temporal structure, and decodes the next physical state.

MORPH architecture diagram showing heterogeneous 1D, 2D, and 3D data unified into a common tensor format, processed by convolution and axial-attention modules, then decoded autoregressively.

Capabilities & downstream tasks

MORPH supports efficient autoregressive forecasting and adapts to forward and inverse scientific learning tasks.

Data efficiency

Pretraining lets MORPH outperform standalone models with substantially fewer task-specific trajectories.

<1%

FNS-KF data can exceed the standalone reference.

Autoregressive prediction

MORPH produces stable multi-step forecasts from an initial state on representative shallow-water and forced Navier–Stokes systems.

Terminal key-frame prediction

Predict a system’s terminal state directly from its initial conditions.

Forward modeling

Composite material property estimation

Infer material properties from observed physical responses.

Inverse modeling

Aerospace structural damage detection

Identify damage in aerospace structures from their measured behavior.

Inverse modeling

Inertial-confinement-fusion parameter estimation

Estimate physical parameters governing inertial-confinement-fusion systems.

Inverse modeling

Sparse sea-surface-temperature reconstruction

Reconstruct complete sea-surface-temperature fields from sparse observations.

Inverse modeling

Data universe

Pretraining · 6 datasets

  • MHD-3D
  • DR-2D
  • CFD-1D
  • CFD2D-IC
  • CFD-3D
  • SW-2D

Fine-tuning · 7 datasets

  • DR-1D
  • CFD-2D
  • CFD3D-TURB
  • BE-1D
  • GSDR-2D
  • TGC-3D
  • FNS-KF-2D

Autoregressive rollouts

Ground-truth dynamics and MORPH-FM-S* predictions across a 20-step autoregressive rollout.

Fine-tuning trajectories
128 (<1% of data)
Epochs
100
Compute
1 GPU

NS-SL

NS-SL horizontal velocity autoregressive rollout comparing ground truth with the MORPH prediction over 20 time steps.
Horizontal velocity vx
NS-SL vertical velocity autoregressive rollout comparing ground truth with the MORPH prediction over 20 time steps.
Vertical velocity vy

PwC

PwC horizontal velocity autoregressive rollout comparing ground truth with the MORPH prediction over 20 time steps.
Horizontal velocity vx
PwC vertical velocity autoregressive rollout comparing ground truth with the MORPH prediction over 20 time steps.
Vertical velocity vy

CE-RM

CE-RM pressure autoregressive rollout comparing ground truth with the MORPH prediction over 20 time steps.
Pressure P
CE-RM horizontal velocity autoregressive rollout comparing ground truth with the MORPH prediction over 20 time steps.
Horizontal velocity vx

Citation

@article{rautela2025morph,
  title   = {MORPH: PDE Foundation Models with Arbitrary Data Modality},
  author  = {Rautela, Mahindra Singh and Most, Alexander and Mansingh, Siddharth and Love, Bradley C and Scheinker, Alexander and Oyen, Diane and Debardeleben, Nathan and Lawrence, Earl and Biswas, Ayan},
  journal = {arXiv preprint arXiv:2509.21670},
  year    = {2025}
}