1: \begin{abstract}

2: \noindent

3: We study the principal component analysis (PCA) based approach introduced by Reisinger \& Wittum \cite{RW07} 

4: for the approximation of Bermudan basket option values via partial differential equations (PDEs).

5: This highly efficient approximation approach requires the solution of only a limited number of low-dimensional 

6: PDEs complemented with optimal exercise conditions.

7: It is demonstrated by ample numerical experiments that a common discretization of the pertinent PDE problems 

8: yields a second-order convergence behaviour in space and time, which is as desired.

9: It is also found that this behaviour can be somewhat irregular, and insight into this phenomenon is obtained.

10: \end{abstract}

11: