Mathematics > Dynamical Systems
[Submitted on 1 Nov 2022 (this version), latest version 4 Jan 2024 (v2)]
Title:Generalized Quadratic-Embeddings for Nonlinear Dynamics using Deep Learning
View PDFAbstract:The engineering design process (e.g., control and forecasting) relies on mathematical modeling, describing the underlying dynamic behavior. For complex dynamics behavior, modeling procedures, as well as models, can be intricated, which can make the design process cumbersome. Therefore, it is desirable to have a common model structure, which is also simple enough, for all nonlinear dynamics to enhance design processes. The simplest dynamical model -- one can think of -- is linear, but linear models are often not expressive enough to apprehend complex dynamics. In this work, we propose a modeling approach for nonlinear dynamics and discuss a common framework to model nonlinear dynamic processes, which is built upon a \emph{lifting-principle}. The preeminent idea of the principle is that smooth nonlinear systems can be written as quadratic systems in an appropriate lifted coordinate system without any approximation error. Hand-designing these coordinates is not straightforward. In this work, we utilize deep learning capabilities and discuss suitable neural network architectures to find such a coordinate system using data. We present innovative neural architectures and the corresponding objective criterion to achieve our goal. We illustrate the approach using data coming from applications in engineering and biology.
Submission history
From: Pawan Goyal [view email][v1] Tue, 1 Nov 2022 10:03:34 UTC (11,975 KB)
[v2] Thu, 4 Jan 2024 17:51:13 UTC (7,524 KB)
Current browse context:
math.DS
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.