Parlett The Symmetric Eigenvalue Problem Pdf May 2026

Beresford N. Parlett’s The Symmetric Eigenvalue Problem is considered a definitive authority on the numerical analysis of real symmetric matrices. Originally published in 1980 and later reprinted by SIAM in its Classics in Applied Mathematics series (1998), the book bridges the gap between pure matrix theory and practical computer implementation. Key Highlights

Grab the amended version from SIAM Publications or find a copy on Amazon to see why it's been a staple for over 40 years. parlett the symmetric eigenvalue problem pdf

1. The QR algorithm: The QR algorithm is a popular method for computing the eigenvalues and eigenvectors of a symmetric matrix. The algorithm involves iteratively applying a sequence of orthogonal similarity transformations to the matrix, which converges to a diagonal matrix containing the eigenvalues.
2. The divide-and-conquer eigenvalue algorithm: This algorithm is a fast and efficient method for computing the eigenvalues and eigenvectors of a symmetric matrix. The algorithm involves dividing the matrix into smaller submatrices, solving the eigenvalue problem for each submatrix, and then combining the solutions.
3. Eigenvalue decomposition: Parlett discusses the eigenvalue decomposition of a symmetric matrix, which involves expressing the matrix as a product of three matrices: an orthogonal matrix of eigenvectors, a diagonal matrix of eigenvalues, and the transpose of the eigenvector matrix.

• Small-to-medium dense matrices: tridiagonal QR or divide-and-conquer.
• Large dense where many eigenpairs required: divide-and-conquer or MRRR.
• Sparse large matrices with a few eigenpairs: Lanczos or implicitly restarted Lanczos (ARPACK).

3. Key Topics and Highlights

• QR Algorithm: The crown jewel of the book is its treatment of the QR algorithm. Parlett does an exceptional job explaining the mechanics of QR with shifts, particularly the Wilkinson shift. He explains how the convergence rates relate to the gaps between eigenvalues, offering deep insight into why the algorithm is essentially "infinitely fast" in practice.
• Tridiagonalization: The book meticulously details the reduction of dense symmetric matrices to tridiagonal form (Householder reflections) and why this step is crucial for efficiency.
• Perturbation Theory: Before diving into computation, Parlett provides a rigorous look at how small changes in matrix entries affect eigenvalues (condition numbers). This is essential for understanding the accuracy of results.
• Bisection and Inverse Iteration: For finding specific eigenvalues rather than the whole spectrum, the book covers the bisection method using Sturm sequences (sylvester's law of inertia) and the method of inverse iteration, complete with strategies for finding orthogonal eigenvectors for clustered eigenvalues.
• The Lanczos Algorithm: The later chapters provide one of the best early treatments of the Lanczos algorithm, explaining its ability to handle sparse matrices that are too large to tridiagonalize. He addresses the "spurious eigenvalue" problem and the loss of orthogonality with characteristic clarity.

11. Suggested Study Path

1. Review linear algebra fundamentals: spectral theorem, orthogonality, conditioning.
2. Learn Householder tridiagonalization and blocked implementations.
3. Study QR algorithm for symmetric tridiagonal matrices.
4. Read Cuppen’s divide-and-conquer and MRRR papers for advanced methods.
5. Experiment with LAPACK routines on test matrices and validate residuals.
6. For sparse problems, study Lanczos, Arnoldi, and practical reorthogonalization.

4. Tridiagonal Eigenvalue Algorithms

Main algorithms covered/used in practice: Beresford N