Beresford Parlett's "The Symmetric Eigenvalue Problem" is a foundational, SIAM-reprinted text (1980) focusing on numerical methods for real symmetric matrices. The text covers dense matrix methods, including QR algorithms, and extensive coverage of Lanczos algorithms for large sparse matrices, with a critical, in-depth approach to practical numerical analysis. For a detailed overview of the book's structure and contents, visit SIAM Publications Library.
He then introduces the canonical angles between subspaces (the sin(Θ) metric) to measure how close two invariant subspaces are. This geometric viewpoint directly informs algorithms: if you only need the subspace (e.g., for PCA), you can stop early without computing individual eigenvectors. parlett the symmetric eigenvalue problem pdf
🔗 Full details & Series info: SIAM Classics in Applied Mathematics Beresford Parlett's "The Symmetric Eigenvalue Problem" is a
Searching for "Parlett the symmetric eigenvalue problem pdf" is more than a hunt for a free file; it is a recognition of a masterpiece. Parlett’s work stands alongside Wilkinson’s The Algebraic Eigenvalue Problem as one of the two pillars of eigenvalue computation. While Wilkinson emphasizes rounding error analysis, Parlett emphasizes mathematical structure and algorithmic geometry. He then introduces the canonical angles between subspaces
The symmetric eigenvalue problem remains an active area of research, with many open problems and challenges. Future research directions include:
Berkeley professor Beresford N. Parlett has made significant contributions to the field of numerical linear algebra, particularly in the area of eigenvalue problems. His book, "The Symmetric Eigenvalue Problem," provides a comprehensive treatment of the symmetric eigenvalue problem, covering both theoretical and practical aspects. The book is written in a clear and concise manner, making it accessible to researchers and practitioners alike.
Rayleigh Quotient Iteration (RQI): Known for its cubic convergence, this is a central theme in the text for refining eigenvalue approximations.