Sparse iterative solvers (gmres, cg) in SciPy - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is the main purpose of sparse iterative solvers like GMRES and CG?

They efficiently solve large systems of linear equations where the matrix is sparse, meaning most elements are zero, saving memory and computation time.

beginner

What type of matrices is the Conjugate Gradient (CG) method best suited for?

CG is best for symmetric and positive definite matrices.

intermediate

How does GMRES differ from CG in terms of matrix requirements?

GMRES can solve general non-symmetric matrices, while CG requires the matrix to be symmetric and positive definite.

beginner

In SciPy, which module provides the gmres and cg functions?

The functions gmres and cg are available in the scipy.sparse.linalg module.

intermediate

Why are iterative solvers preferred over direct solvers for very large sparse systems?

Iterative solvers use less memory and can be faster because they avoid computing full matrix factorizations, which are costly for large sparse matrices.

Which solver is suitable for a symmetric positive definite sparse matrix?

A Conjugate Gradient (CG)

B GMRES

C LU decomposition

D Jacobi method

CG is designed for symmetric positive definite matrices, making it the best choice here.

What does GMRES stand for?

A Gaussian Matrix Residual

B Gradient Method for Residuals

C General Matrix Residual Solver

D Generalized Minimal Residual

GMRES stands for Generalized Minimal Residual, an iterative method for solving non-symmetric systems.

Which SciPy module contains the gmres and cg functions?

A scipy.linalg

B scipy.sparse.linalg

C scipy.optimize

D scipy.integrate

The gmres and cg functions are in scipy.sparse.linalg, specialized for sparse matrix operations.

Why are iterative solvers preferred for large sparse systems?

A They compute exact solutions instantly

B They require dense matrices

C They use less memory and can be faster

D They do not need initial guesses

Iterative solvers save memory and time by avoiding full matrix factorizations.

Which of these is NOT a characteristic of the CG method?

A Can solve non-symmetric matrices

B Works only on symmetric matrices

C Requires positive definiteness

D Is an iterative solver

CG cannot solve non-symmetric matrices; GMRES is used for those.

Explain when and why you would use the GMRES solver instead of CG.

Think about matrix symmetry and solver flexibility.

Describe the advantages of using sparse iterative solvers for large linear systems.

Consider memory and speed benefits.