Prediction of flow dynamics using point processes.

Area of Science:

• Dynamical Systems
• Time Series Analysis
• Chaos Theory

Background:

• Parsimonious description of time series is crucial for studying underlying dynamics.
• Generating partitions offer compact descriptions for discrete systems but lack a basis for continuous systems.

Purpose of the Study:

• To propose a novel method for describing time-continuous time series.
• To enable accurate prediction of system dynamics using time series data.

Main Methods:

• Describing time-continuous series via a local cross section and orbit crossing times.
• Utilizing crossing times and past observations for prediction.
• Demonstrating reconstructability with varying cross section parameters and database length.

Main Results:

• The proposed method allows for accurate prediction of system dynamics.
• Reconstructability is robust to the size and placement of the local cross section with sufficient data.
• The method was successfully demonstrated on the Lorenz model and real-world wind speed data.

Conclusions:

• The novel method provides an effective way to describe and predict continuous time series dynamics.
• This approach offers a robust alternative to traditional methods for continuous systems.
• The findings have implications for analyzing complex dynamical systems.