The JAG inertial confinement fusion simulation dataset for multi-modal scientific deep learning
The repository snapshot available here was captured in December 2019. For more recent updates to the code, see https://github.com/rushilanirudh/icf-jag-cycleGAN.
- Cite This Work
Gaffney, Jim A.; Anirudh, Rushil; Bremer, Peer-Timo; Hammer, Jim; Hysom, David; Jacobs, Sam A.; Peterson, J. Luc; Robinson, Peter; Spears, Brian K.; Springer, Paul T.; Thiagarajan, Jayaraman J.; Van Essen, Brian; Yeom, Jae-Seung (2020). The JAG inertial confinement fusion simulation dataset for multi-modal scientific deep learning. In Lawrence Livermore National Laboratory (LLNL) Open Data Initiative. UC San Diego Library Digital Collections. https://doi.org/10.6075/J0RV0M27
Anirudh, Rushil, Bremer, Peer-Timo, and Thiagrarjan, Jayaraman J. Cycle Consistent Surrogate for Inertial Confinement Fusion. Computer Software. https://github.com/rushilanirudh/icf-jag-cycleGAN. USDOE National Nuclear Security Administration (NNSA). 01 Feb. 2019. Web. https://doi.org/10.11578/dc.20190503.2
The goal of this project is to build better surrogate models for Inertial Confinement Fusion (ICF) using neural networks. Particularly, we are interested in replicating the behavior of the JAG 1D Semi-analytic simulator for ICF.
The JAG model has been designed to give a rapid description of the observables from ICF experiments, which are all generated very late in the implosion. In this way the very complex and computationally expensive transport models needed to describe the capsule drive can be avoided, allowing a single solution in $ ilde$ seconds. The trade-off is that JAG inputs do not relate to actual experimental observables, rather the state of the implosion once the laser drive has switched off. At that point, an analytic description of the spatial profile inside the hotspot can be found [1,2], leaving only a set of coupled ODEs describing the temporal energy balance inside the entire problem which can be solved easily . The various terms in the energy balance equation relate to different physics processes (radiation, electron conduction, heating by alpha particles, etc), making JAG useful for investigating the role of various potentially uncertain physics models. Combined with a thin-shell model describing the 3D hydrodynamic evolution of the hotspot , JAG has a detailed description of the spatial and temporal evolution of all thermodynamic variables which can be post-process to predict a full range of experimental observables.
1. Betti et al., Physics of Plasmas 9, 2277 (2002)
2. Springer et al., EPJ Web. Conferences 59:04001 (2013)
3. Betti et al., Physical Review Letters 114:255003 (2015)
4. Ott et al., Physical Review letters 29:1429 (1995)
- Creation Date
- July 2018
- Date Issued
- Technical Details
A dataset is provided to test/train the models. This is a tarball inside 'data/', which contains .npy files for 10K images, scalars, and the corresponding input parameters. The size of the dataset provided (in 'data/') are as follows: Input: (9984, 5), Output/Scalars: (9984, 22), Output/Images: (9984, 16384). Images are interpreted as (-1,64,64,4).
This package was built and tested using Tensorflow 1.8.0. It also depends on standard Python packages such as NumPy, Matplotlib for basic data loading and plotting utilities.
We also provide a Python Jupyter Notebook, that is a self-contained script to load, process, and test the dataset described above. In particular, we include a Neural Network designed to act as a surrogate for the JAG 1D Simulator. The neural network is implemented in Tensorflow.
The notebook allows a user to load the dataset, load the neural network and train it such that given just the 5 input parameters, it predicts the scalars and images accurately. This can be done directly in the notebook, without any additional modifications. During training, intermediate predictions are also saved to disk (as specified by the user). We hope this serves as a starting point to build, test and play with the ICF-JAG simulation dataset.
- Related Publications
R. Betti et al. (2002), Deceleration phase of inertial confinement fusion implosions, Physics of Plasmas, https://doi.org/10.1063/1.1459458
Betti et al. (2015), Alpha heating and burning plasmas in inertial confinement fusion, Physical Review Letters, https://doi.org/10.1103/PhysRevLett.114.255003
E. Ott, “Nonlinear Evolution of the Rayleigh-Taylor Instability of a Thin Layer” Physical Review Letters, vol. 29, no. 21, pp. 1429, 1972. https://doi.org/10.1103/PhysRevLett.29.1429
P.T. Springer et al. (2013), Integrated thermodynamic model for ignition target performance, EPJ Web of Conferences, https://doi.org/10.1051/epjconf/20135904001
Is Referenced By:
Anirudh, R., Thiagarajan, J. J., Bremer, P. T., & Spears, B. K. (2019). Improved Surrogates in Inertial Confinement Fusion with Manifold and Cycle Consistencies. arXiv preprint arXiv:1912.08113. (https://arxiv.org/abs/1912.08113)
Anirudh, Rushil; Thiagarajan, Jayaraman J.; Liu, Shusen; Bremer, Peer-Timo; Spears, Brian K. 2019. Exploring Generative Physics Models with Scientific Priors
in Inertial Confinement Fusion. arXiv:1910.01666v1 [physics.comp-ph] (https://arxiv.org/abs/1910.01666)
Gaffney et al. (2014), Thermodynamic modeling of uncertainties in NIF ICF implosions due to underlying microphysics models, APS Meeting Abstracts. http://meetings.aps.org/link/BAPS.2014.DPP.PO5.11
Liu et al. (2019), Scalable topological data analysis and visualization for evaluating data-driven models in scientific applications, IEEE VIS 2019. https://arxiv.org/abs/1907.08325. https://doi.org/10.1109/TVCG.2019.2934594
Sam Ade Jacobs, Brian Van Essen, David Hysom, Jae-Seung Yeom, Tim Moon, Rushil Anirudh, Jayaraman J. Thiagaranjan, Shusen Liu, Peer-Timo Bremer, Jim Gaffney, Tom Benson, Peter Robinson, Luc Peterson, Brian Spears. 2019. Parallelizing Training of Deep Generative Models on Massive Scientific Datasets. arXiv:1910.02270v1 [cs.DC] (https://arxiv.org/abs/1910.02270)
- Related Resource
- Rights Holder
- Lawrence Livermore National Laboratory
Under copyright (US)
Use: This work is available from the UC San Diego Library. This digital copy of the work is intended to support research, teaching, and private study.
Constraint(s) on Use: This work is protected by the U.S. Copyright Law (Title 17, U.S.C.). Use of this work beyond that allowed by "fair use" or any license applied to this work requires written permission of the copyright holder(s). Responsibility for obtaining permissions and any use and distribution of this work rests exclusively with the user and not the UC San Diego Library. Inquiries can be made to the UC San Diego Library program having custody of the work.
- Digital Object Made Available By
Research Data Curation Program, UC San Diego, La Jolla, 92093-0175 (https://lib.ucsd.edu/rdcp)
- Last Modified