| Autor: |
He, Wenbin, Wang, Junpeng, Guo, Hanqi, Wang, Ko-Chih, Shen, Han-Wei, Raj, Mukund, Nashed, Youssef S. G., Peterka, Tom |
| Rok vydání: |
2019 |
| Předmět: |
|
| Druh dokumentu: |
Working Paper |
| DOI: |
10.1109/TVCG.2019.2934312 |
| Popis: |
We propose InSituNet, a deep learning based surrogate model to support parameter space exploration for ensemble simulations that are visualized in situ. In situ visualization, generating visualizations at simulation time, is becoming prevalent in handling large-scale simulations because of the I/O and storage constraints. However, in situ visualization approaches limit the flexibility of post-hoc exploration because the raw simulation data are no longer available. Although multiple image-based approaches have been proposed to mitigate this limitation, those approaches lack the ability to explore the simulation parameters. Our approach allows flexible exploration of parameter space for large-scale ensemble simulations by taking advantage of the recent advances in deep learning. Specifically, we design InSituNet as a convolutional regression model to learn the mapping from the simulation and visualization parameters to the visualization results. With the trained model, users can generate new images for different simulation parameters under various visualization settings, which enables in-depth analysis of the underlying ensemble simulations. We demonstrate the effectiveness of InSituNet in combustion, cosmology, and ocean simulations through quantitative and qualitative evaluations. |
| Databáze: |
arXiv |
| Externí odkaz: |
|
