A comparative study of numerical algorithms in calculating eigenpairs of the master equation for protein folding kinetics

In this paper, we apply different eigenvalue algorithms to calculate some larger nonpositive eigenvalues and their corresponding eigenvectors of the matrix A coming from the master equation for a protein folding problem. In terms of the accuracy, stability, and robustness, different algorithms, such as power method, implicitly restarted Arnoldi method, Jacobi-Davidson method, and QR algorithm are compared with respect to different problem size.