“…Along the line of low rank algorithms based on tensor decomposition, dynamical low rank algorithm (DLRA) [35,14,34]and the proper generalized decomposition (PGD) [1,37,13] have been designed. Projection based ROMs have also been actively developed in the recent few years, for example the proper orthogonal decomposition (POD) and its variations [5,11,12,40,3,10,19], the dynamical mode decomposition (DMD) [26,27]. Among those work, the POD methods in [5,41,19] and our previous work in reduced basis method (RBM) for the steady state problem [30] make explicit use of the low rank structure of the solution manifold induced by the angular variable, namely, the ROM built is based on treating the angular variable as the "parameter" of the model.…”