Eigenproblems on symmetric tridiagonal matrices are a cornerstone of numerical linear algebra, with applications spanning physics, engineering, and data science. LAPACK, a widely adopted library for linear algebra computations, offers multiple routines for solving these eigenproblems, each leveraging distinct algorithms with varying computational complexities and performance characteristics. This report provides a comparative analysis of the LAPACK routines STEV, STEVD, STEVX, and STEVR. Through synthetic and real-world matrices, we evaluate their performance and accuracy under different scenarios.
Check out the full report for more details and the repository for reproducing the experiment results.