Assess students' ability to analyze and predict convergence behavior of Conjugate Gradient (CG) and GMRES methods for large sparse linear systems using spectral diagnostics — including eigenvalue distributions, condition number, clustering, nonnormality/field of values, Ritz/value approximations, and the impact of preconditioning and restarts. Items should require interpretation of spectra, estimation of iteration counts, identification of stagnation causes, and selection or design of mitigation strategies for realistic sparse problems.