Generative Learning on Manifolds for Probabilistic Seismic Performance Assessment and Structural Identifiability
Abstract
Seismic performance assessment relies on computationally expensive nonlinear simulations, limiting uncertainty quantification and inference of structural properties from observed responses. We develop a manifold-based generative framework that learns the joint distribution of uncertain structural parameters and seismic-response trajectories from limited high-fidelity simulations. Combining Diffusion Maps, score-based Diffusion Modeling, and Geometric Harmonics, the framework enables both forward probabilistic prediction and inverse inference. Applied to a nine-story steel moment-resisting frame, the results reveal which structural properties are identifiable from seismic performance and how uncertainty in these properties influences response. In particular, the learned representation captures post-capping degradation characteristics more effectively than individual strength parameters. Inverse inference identifies distinct structural configurations that produce similar seismic performance, with representative solutions independently validated using nonlinear simulations. These findings demonstrate how generative learning can support uncertainty-aware seismic assessment while revealing structural non-identifiability and alternative mechanisms underlying observed performance
Related
- Technical note: Integrating generative AI with physics-based simulations for probabilistic modelling of volcanic clouds
- Probabilistic assessment of tropical cyclone recovery burden across US Atlantic and Gulf Coast counties
- Multi-scenario probabilistic hazard assessment of flash flood disasters in Chengkou County based on a multi-state system model
- Regional climate risk assessment from climate models using probabilistic machine learning
- Probabilistic environmental risk assessment of nondissolvable nanomaterials (titanium dioxide, carbon nanotubes, graphene-based materials, polystyrene nanoplastics) in the terrestrial ecosystem
BibTeX
@misc{giovanis2026generative,
title = {{Generative Learning on Manifolds for Probabilistic Seismic Performance Assessment and Structural Identifiability}},
author = {Dimitris G Giovanis and Ioannis G Kevrekidis and Roger G Ghanem},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.12908},
eprint = {2611.12908},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.12908},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Generative Learning on Manifolds for Probabilistic Seismic Performance Assessment and Structural Identifiability
AU - Dimitris G Giovanis
AU - Ioannis G Kevrekidis
AU - Roger G Ghanem
PY - 2026
DA - 2026/11/01
PB - arXiv
DO - 10.48550/arxiv.2611.12908
UR - https://arxiv.org/abs/2611.12908
AB - Seismic performance assessment relies on computationally expensive nonlinear simulations, limiting uncertainty quantification and inference of structural properties from observed responses. We develop a manifold-based generative framework that learns the joint distribution of uncertain structural parameters and seismic-response trajectories from limited high-fidelity simulations. Combining Diffusion Maps, score-based Diffusion Modeling, and Geometric Harmonics, the framework enables both forward probabilistic prediction and inverse inference. Applied to a nine-story steel moment-resisting frame, the results reveal which structural properties are identifiable from seismic performance and how uncertainty in these properties influences response. In particular, the learned representation captures post-capping degradation characteristics more effectively than individual strength parameters. Inverse inference identifies distinct structural configurations that produce similar seismic performance, with representative solutions independently validated using nonlinear simulations. These findings demonstrate how generative learning can support uncertainty-aware seismic assessment while revealing structural non-identifiability and alternative mechanisms underlying observed performance
ER -