Transformation into optimal parallelism in Euclidean spaces (or: How to explain the shape of the electron-density distribution inside a crystal) (Q5954129)

The authors consider a problem that has its origin in crystallographic diffraction experiments. Let \((x_1,x_2,\ldots,x_n)\) and \((y_1,y_2,\ldots,y_n)\) be \(n\)-tuples of vectors in a \(d\)-dimensional Euclidean vector space. Find a linear transformation \(U\) in the (special) orthogonal group such that \(\sum_i \langle Ux_i,y_i\rangle\) is maximal. Thus, loosely speaking, we seek a transformation \(U\) such that the vectors \(Ux_1,Ux_2,\ldots,Ux_n\) and \(y_1,y_2,\ldots,y_n\) are ``as parallel as possible''. The authors give a complete solution of this problem in terms of the singular value decomposition of a \(d\times d\)-matrix associated with the given vectors and discuss various generalizations, e.g., the infinite-dimensional case.