Evaluate the learner's ability to design a novel adaptive preconditioned iterative solver for large-scale non-symmetric sparse linear systems arising from anisotropic finite-element discretizations. Task requires a precise algorithm, adaptive preconditioner strategy, rigorous convergence proof under stated matrix assumptions (e.g., field-of-values or pseudospectrum bounds), complexity and scalability analysis, and a plan for numerical experiments validating theoretical rates.