A tale of two vortices: how numerical ergodic theory and transfer operators reveal fundamental changes to coherent structures in non-autonomous dynamical systems (Q2194428)

The authors develop and implement singular value decomposition algorithms to provide multilayered descriptions of time-dependent systems. In numerical applications, infinite-dimensional transfer operators are approximated by finite-rank stochastic matrices. Further, these algorithms rely on singular value decompositions associated to Ulam-type discretisations of transfer operators induced by dynamical systems. The authors provide the following algorithms: (1) Forward time rolling windows and singular value decompositions; (2) Tracking modes through time with singular values; (3) Tracking modes through time using singular vectors; (4) Animating tracked modes over time, (5) Equivariance mismatch in forward time. For the Ulam scheme, all models seed 100 test points per bin and integrate the vector field using standard Runge-Kutta methods, interpolating linearly in space and time. The algorithms are tested on three models of periodically and quasi-periodically driven systems, as well as on a geophysical dataset corresponding to the splitting of the Southern Polar Vortex. The authors mention that, in all of these three cases, their methods provide important information regarding the location and the time windows within which coherent structures merge and/or separate in the presence of an underlying time-dependent and possibly chaotic flow.