While the book is densely packed with theoretical rigor, it is structured to guide the reader through the foundational concepts to advanced algorithms. 1. Fundamental Concepts of Symmetric Matrices
Parlett emphasizes the impact of floating-point arithmetic on computations, teaching readers to distinguish between algorithms that look good on paper and those that are numerically stable. Core Topics Covered in the Book
The algorithms, error bounds, and mathematical philosophy detailed in Parlett's book serve as the theoretical blueprint behind LAPACK (Linear Algebra Package), the foundational Fortran library that powers the numerical backends of modern tools like MATLAB, NumPy, SciPy, and R. Final Thoughts: A Timeless Classic
“Parlett’s book is the definitive treatment of the symmetric eigenvalue problem – a masterpiece of clarity, depth, and numerical wisdom.” – common sentiment among numerical analysts. parlett the symmetric eigenvalue problem pdf
The first half covers transformations for dense matrices, while the latter half tackles the complex world of large, sparse matrices and Krylov subspaces.
A crucial property for many applications is that symmetric matrices possess mutually orthogonal eigenvectors.
Beresford N. Parlett's book "The Symmetric Eigenvalue Problem" provides a comprehensive treatment of the symmetric eigenvalue problem, including the QR algorithm and other methods. The book covers the following topics: While the book is densely packed with theoretical
Eigenvectors corresponding to distinct eigenvalues are strictly orthogonal.
Because the original book was published in 1980, it predates some modern developments:
See a comparing dense vs. tridiagonal solvers Share public link Core Topics Covered in the Book The algorithms,
Parlett’s text is celebrated because it does not just present algorithms; it explains the underlying mathematical structure that makes those algorithms work. 1. The Power of Variational Characterization
| Chapter | Focus | |---------|-------| | 4–5 | Perturbation theory and error analysis | | 6–8 | Reduction to tridiagonal form (Householder, Lanczos) | | 9–11 | The symmetric QR algorithm | | 12–13 | Bisection and inverse iteration | | 14–15 | Lanczos method in depth (including practical issues) |
These focus on "storable" matrices—dense matrices where we can perform transformations explicitly with minimal error beyond inexact arithmetic. The Scale (Chapters 10–14):