MML-05-06: Matrix Condition Number, Ill-Conditioning & Numerical Regularization – Practice Quiz

30 worked single-select and multiselect questions. Tier 2–3. Prerequisites: QS 5.5. Learning goals: Compute Euclidean condition numbers from extreme singular values; Distinguish absolute scale from relative sensitivity; Apply right-hand-side perturbation bounds with fixed invertible matrix; Diagnose anisotropic noise amplification and numerical rank; Compare normal-equation conditioning with the original system; Derive ridge normal equations and singular-direction filters; Quantify regularization stability and bias tradeoffs. Column gradients and output-by-input Jacobians are used where relevant. Detailed solutions explain every choice. References: Source 1, Source 2, Source 3. Ridge reference: UBC Stat406.