StOKeDMD: Streaming Occupation kernel dynamic mode decomposition
Dynamic mode decomposition (DMD) has become a common technique for constructing surrogate models for dynamical systems from observed system states. The Occupation Kernel DMD (OKDMD) method proposed in (Rosenfeld et al., 2022) and (Rosenfeld et al., 2024) is a Liouville operator based method that builds surrogate models from system state trajectories. Here, this paper proposes an extension of OKDMD to the case when the system states are observed in a streaming fashion, i.e., only a small fraction of the state trajectory is available at a given time. The developed method, Streaming Occupation Kernel DMD (StOKeDMD), accommodates the streaming data input by leveraging properties of specific choices of kernel functions and occupation kernels. We apply the StoKeDMD method as a compression method for streaming data, analyze the memory complexity, and demonstrate the performance of StoKeDMD in the compression of streaming data generated from a Lorenz system and a fluid flow simulation.
- Research Organization:
- Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (United States); Sandia National Laboratories (SNL-NM), Albuquerque, NM (United States)
- Sponsoring Organization:
- USDOE National Nuclear Security Administration (NNSA); USDOE Office of Science (SC), Advanced Scientific Computing Research (ASCR)
- Grant/Contract Number:
- AC05-00OR22725; NA0003525
- OSTI ID:
- 2481176
- Report Number(s):
- SAND--2024-16668J
- Journal Information:
- Applied Mathematics for Modern Challenges, Journal Name: Applied Mathematics for Modern Challenges Journal Issue: 4 Vol. 2; ISSN 2994-7669
- Publisher:
- American Institute of Mathematical SciencesCopyright Statement
- Country of Publication:
- United States
- Language:
- English
Similar Records
Fault Detection via Occupation Kernel Principal Component Analysis
Dynamic Mode Decomposition for Compressive System Identification