Iterative Methods for Large Linear Systems
For a large sparse system, elimination costs too much and the solution is approached instead of computed. Convergence is decided by the spectrum of the iteration, and preconditioning is the act of changing that spectrum.
What a learner can do afterwards
- Run one step of a splitting method and state its convergence condition
- Say why the conjugate gradient method needs a symmetric positive definite matrix
- Explain what a preconditioner changes and why that helps
1 · Read
For a large sparse system, elimination costs too much, so you approach the solution with a splitting method. Jacobi reuses only old values: solve each equation for its own variable, then plug in the previous guess. For 5x plus y = 11 with x plus 4y = 9, start from (0, 0): x1 is 11 over 5, which is 2.2, and y1 is 9 over 4, which is 2.25.
Keep going with the same formulas. Feed y1 = 2.25 into the x formula: 11 minus 2.25 is 8.75, over 5 is 1.75, so x2 is 1.75. The system is strictly diagonally dominant, since 5 exceeds 1 and 4 exceeds 1, which guarantees Jacobi converges from any start. In general a splitting converges exactly when its iteration matrix has spectral radius below 1.
Conjugate gradients are faster but pickier: they demand a symmetric positive definite matrix. Each step minimizes a quadratic energy built from the matrix, symmetry aligns that energy with the inner product used, and positive definiteness supplies one unique minimum to land on. Without both, the guarantee evaporates.
A preconditioner does not change the solution; it changes the spectrum. By clustering the eigenvalues it makes the iteration converge far sooner. So the workflow is: skip costly elimination, pick a splitting or gradients, and precondition to reshape the spectrum.
Split and repeat for huge sparse systems, watch the iteration spectrum for convergence, and let symmetry plus definiteness unlock gradients.
2 · Watch
Take it off screen
Where it sits
Learn first
This opens up
Nothing builds on it yet.
8 questions wait behind this lesson, each with its answer explained. Every answer feeds the sky: stars light as they are learned, and dim when it is time to come back.