Questions: Iterative Methods for Linear Systems

5 questions to test your understanding

Score: 0 / 5
Question 1 Multiple Choice

Jacobi iteration is applied to a 10,000 × 10,000 sparse linear system. After many iterations, the residual does not decrease — the error stays roughly constant or oscillates. What is the most likely cause?

AThe matrix is too large; Jacobi only works for systems with fewer than 1,000 unknowns
BThe spectral radius of the Jacobi iteration matrix is greater than or equal to 1, so the iteration does not converge
CThe right-hand side vector b contains numerical errors that prevent convergence
DJacobi requires a symmetric matrix; the system must not be symmetric
Question 2 Multiple Choice

Why is the conjugate gradient method typically preferred over Jacobi or Gauss-Seidel for large symmetric positive definite (SPD) systems?

AConjugate gradient does not require storing the matrix A, while Jacobi and Gauss-Seidel both require dense factorizations
BConjugate gradient converges in at most n steps in exact arithmetic and selects optimal update directions from a Krylov subspace, while Jacobi and Gauss-Seidel use simple component-wise updates with slower convergence rates
CConjugate gradient is the only iterative method guaranteed to work on sparse matrices
DJacobi and Gauss-Seidel cannot handle SPD matrices because the diagonal is always positive
Question 3 True / False

Gauss-Seidel typically converges faster than Jacobi on the same system because it uses the most recently computed values of x immediately within each iteration sweep.

TTrue
FFalse
Question 4 True / False

Iterative methods are typically preferable to direct methods (like Gaussian elimination) for solving large linear systems, since direct methods are too slow for any system of practical size.

TTrue
FFalse
Question 5 Short Answer

Explain why the spectral radius ρ(M) of the iteration matrix governs convergence of an iterative method, and what happens geometrically when ρ(M) ≥ 1.

Think about your answer, then reveal below.