Cluster-Weighted EDMD (CW-EDMD) jointly learns a soft phase-space partition and per-cluster EDMD operators, improving prediction accuracy over standard Extended Dynamic Mode Decomposition (EDMD) when local dynamics vary across state-space regions. Tested on Lorenz, damped pendulum, and Duffing systems across 288 paired comparisons, CW-EDMD achieved significant error reductions in 258 cases. Median one-step error reductions were 57x, 2.7x, and 12x on pendulum, Duffing, and Lorenz, respectively.
No score is assigned. Sources and their independence are shown in the citation chain below.