Szegő limit theorems (Q1403369)

Szegő's limit theorem says that the ratio of the determinants of two successive leading principal submatrices of an infinite Toeplitz matrix is equal to the geometric mean of the symbol of the Toeplitz matrix. The Toeplitz matrix may consist of the Fourier coefficients of some positive measure \(\mu\) and the symbol is the Radon-Nykodym derivative of the absolutely continuous part of \(\mu\). Several extensions were given [\textit{D. Bump} and \textit{P. Diaconis}, J. Comb. Theory, Ser. A 97, No. 2, 252--271 (2002; Zbl 1005.47030); \textit{C. A. Tracy} and \textit{H. Widom}, SIAM J. Matrix Anal. Appl. 23, 1194--1196 (2002; Zbl 1010.47019); \textit{H. Helson} and \textit{D. Lowdenslager}, Acta Math. 99, 165--202 (1958; Zbl 0082.28201)]. These extensions concern a complex measure \(\mu\), or they take the ratio of two determinants of more general submatrices, and results of that type were also generalized to higher dimensions. In this paper, a more geometric approach is taken to derive all these results in a general setting allowing all the previous generalizations simultaneously. The author's approach is based on the analysis of projections on various subspaces.