You are on page 1of 5

07‐09‐2021

LU Decomposition
(Chapter 20, Section 20.2 Kreyszig)

Manjesh K. Singh, manjesh@iitk.ac.in

1
Matrix decomposition (factorization)
representing matrix as a product of several matrices

LU decomposition refers to factorization of an invertible matrix A such that

where L ‐ lower‐triangular matrix, and U ‐ upper‐triangular matrix

If we know LU decomposition of matrix A, solution of Ax = b


may be solved easily
we can now split the system in two parts
and the methods of finding L and U will be
discussed later, we will get the idea
find y find x using Gaussian elimination

Manjesh K. Singh, ME, IIT Kanpur


07‐09‐2021

Recall Gaussian elimination

Forward Elimination Back Substitution


Forward Elimination in matrix form we also know

E A U
The goal of forward elimination is to reduce the coefficient matrix A
to upper‐triangular matrix U via multiplication of matrix E

3
we can
easily find
E A U
lower‐triangular
Thus

In general, for an invertible matrix A,


Gaussian elimination provides
E-1 should be lower‐triangular

Finding the LU decomposition from gaussian Elimination is not helpful for solving
linear equations since Gaussian elimination itself solves linear systems;
we must look for other methods
LU decomposition is not unique; for same A, you may get many L and U

Manjesh K. Singh, ME, IIT Kanpur


07‐09‐2021

Solve using LU Decomposition

Solve the linear system with the following augmented matrices,


use LU decomposition

You might also like