Low-order dynamical models for non-linear fluid behaviour in quasi two-dimensional plasmas. Two complex systems in which a magnetic field imposes two-dimensional fluid motions are turbulent fusion plasmas and magnetospheric plasmas. A distinctive property of 2D flows is the inverse energy cascade, whereby energy streaming into large-scale vortices, coherent structures, or sheared flows gives a remarkable propensity for self-organizing behaviour. This can be exploited to govern or guide our respo ....Low-order dynamical models for non-linear fluid behaviour in quasi two-dimensional plasmas. Two complex systems in which a magnetic field imposes two-dimensional fluid motions are turbulent fusion plasmas and magnetospheric plasmas. A distinctive property of 2D flows is the inverse energy cascade, whereby energy streaming into large-scale vortices, coherent structures, or sheared flows gives a remarkable propensity for self-organizing behaviour. This can be exploited to govern or guide our response to such systems. We propose to investigate the dynamics of momentum and energy exchange in these plasmas, using reduced dynamical models and bifurcation and stability mathematics. Expected outcomes are improved prediction of magnetospheric substorms and confinement of fusion plasmas.
Read moreRead less
Dynamics of eigenvalue/eigenspace algorithms with applications to signal processing. Many problems in signal and systems lead naturally to an eigenvalue/eigenspace determination and tracking problem; for example (acoustic) echo-cancellation, crosstalk suppression in ADSL modems, direction of arrival determination with an array of sensors, linear system identification etc. Exploiting methods from global analysis and dynamical systems theory we will study the available algorithms for eigenspace de ....Dynamics of eigenvalue/eigenspace algorithms with applications to signal processing. Many problems in signal and systems lead naturally to an eigenvalue/eigenspace determination and tracking problem; for example (acoustic) echo-cancellation, crosstalk suppression in ADSL modems, direction of arrival determination with an array of sensors, linear system identification etc. Exploiting methods from global analysis and dynamical systems theory we will study the available algorithms for eigenspace determination to characterise their computational efficiency, accuracy and effectiveness in various data scenarios. The analysis will lead to improved designs for eigenvalue/eigenspace algorithms, as well as design tools to engineer algorithms to specific situations.Read moreRead less