|
|
Attractors in the complex dynamics of turbulent convection
Kašný, Jakub ; Nechvátal, Luděk (referee) ; Macek, Michal (advisor)
This Bachelor's thesis deals with an application of the HAVOK (Hankel Alternative View of Koopman) numerical method, which seeks attractors and predicts intermittent phenomena in dynamical systems, to data from Rayleigh-Bénard convection (RBC), which are measured at Brno Institute of Scientific Instruments in the group of Cryogenics and Superconductivity. This thesis discusses the theory on which the HAVOK is built and further deepens it compared to the article [2]. Furthermore, it enlightens some issues as the best selection of the embedding dimension r, which we selected based on the quality of regression that HAVOK creates, or the use of the Koopman operator and Taken's embedding theorem, that weren't explicitly explained in the article [2]. We discovered three different methods to compute HAVOK regressions based on and using the codes attached to the article. In the thesis, we inspect the matrices of ordinary differential equations, their behaviour when the initial values are changed and their stability for the different regression models and embedding dimensions. The solution with different initial conditions is plotted so that the attractivity can be seen. Part of the thesis contains description of RBC, its equations of motion and characteristic dimensionless numbers that describe the convection. Moreover, the thesis describes how the data are obtained and processed normally and how are processed in new ways based on the HAVOK method.
|
|
|
Attractors in the complex dynamics of turbulent convection
Kašný, Jakub ; Nechvátal, Luděk (referee) ; Macek, Michal (advisor)
This Bachelor's thesis deals with an application of the HAVOK (Hankel Alternative View of Koopman) numerical method, which seeks attractors and predicts intermittent phenomena in dynamical systems, to data from Rayleigh-Bénard convection (RBC), which are measured at Brno Institute of Scientific Instruments in the group of Cryogenics and Superconductivity. This thesis discusses the theory on which the HAVOK is built and further deepens it compared to the article [2]. Furthermore, it enlightens some issues as the best selection of the embedding dimension r, which we selected based on the quality of regression that HAVOK creates, or the use of the Koopman operator and Taken's embedding theorem, that weren't explicitly explained in the article [2]. We discovered three different methods to compute HAVOK regressions based on and using the codes attached to the article. In the thesis, we inspect the matrices of ordinary differential equations, their behaviour when the initial values are changed and their stability for the different regression models and embedding dimensions. The solution with different initial conditions is plotted so that the attractivity can be seen. Part of the thesis contains description of RBC, its equations of motion and characteristic dimensionless numbers that describe the convection. Moreover, the thesis describes how the data are obtained and processed normally and how are processed in new ways based on the HAVOK method.
|
guest ::
