Scribd
Upload
2 views

Lecture 8

The document discusses matrix factorization, specifically LU decomposition, which expresses a square matrix as a product of a lower triangular matrix and an upper triangular matrix. LU factorization simplifies many matrix operations and is particularly useful in computational contexts. It also covers definitions of triangular and diagonal matrices, conditions for invertibility, and provides examples and algorithms for performing LU and LDU factorizations.

Uploaded by

Abebayehu Endale
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
2 views

Lecture 8

The document discusses matrix factorization, specifically LU decomposition, which expresses a square matrix as a product of a lower triangular matrix and an upper triangular matrix. LU factorization simplifies many matrix operations and is particularly useful in computational contexts. It also covers definitions of triangular and diagonal matrices, conditions for invertibility, and provides examples and algorithms for performing LU and LDU factorizations.

Uploaded by

Abebayehu Endale
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 15

Lecture 8.

Matrix Factorization: LU Decomposition


The Factorization

• A factorization of a matrix A is an equation that express A as a product of two or


more matrices.
• Whereas matrix multiplication involves a synthesis of data, matrix factorization is
an analysis of data.
• In the language of computer science, the expression of A as a product amounts to
a preprocessing of the data in A, organizing that data into two or more parts
whose structures are more useful in some way, perhaps more accessible for
computation.
• While there are several factorizations in matrix algebra, we focus on an important
factorization widely used in applications, the LU factorization (or decomposition).
LU Factorization (or Decomposition) of a matrix, introduced by Alan Turing in
1948, is the process of factoring a square matrix into two special types of matrices.
These two matrices are a lower triangular matrix and an upper triangular matrix.
A lot of matrix operations are easier for triangular matrices. “Easier” here means that
the time-complexity for a computer to calculate the result will be lower. While working
with a particular matrix, obtaining its LU decomposition will likely speed things up.
Diagonal and Triangular Matrices
• An n × n matrix A is said to be upper triangular if its entries below the main
diagonal are 0’s (i.e., aij = 0 for i > j) and lower triangular if its entries above the
main diagonal are 0’s (i.e., aij = 0 for i < j).
Diagonal and Triangular Matrices
• An n × n matrix A is said to be upper triangular if its entries below the main
diagonal are 0’s (i.e., aij = 0 for i > j) and lower triangular if its entries above the
main diagonal are 0’s (i.e., aij = 0 for i < j).
• For example,
   
3 2 1 1 0 0
 0 2 1  and  6 0 0 
0 0 5 1 4 3
Both are triangular matrices. The first is upper triangular and the second is lower
triangular.
Diagonal and Triangular Matrices
• An n × n matrix A is said to be upper triangular if its entries below the main
diagonal are 0’s (i.e., aij = 0 for i > j) and lower triangular if its entries above the
main diagonal are 0’s (i.e., aij = 0 for i < j).
• For example,
   
3 2 1 1 0 0
 0 2 1  and  6 0 0 
0 0 5 1 4 3
Both are triangular matrices. The first is upper triangular and the second is lower
triangular.
• An n × n matrix A is diagonal if aij = 0 whenever i ̸= j. For example, the matrices
 
3 0 0  
 0 2 1 0
0  and
0 2
0 0 5

are diagonal. A diagonal matrix is both upper triangular and lower triangular.
• Question. When is a square upper (or lower) triangular matrix invertible? Justify
your answer.
• Question. When is a square upper (or lower) triangular matrix invertible? Justify
your answer.
• Solution. If a square upper (or lower) triangular n × n matrix has nonzero diagonal
entries, then because it is already in echelon form, the matrix is row equivalent to In
and hence is invertible. Conversely, if the matrix is invertible, it has n pivots (leading
entries) on the diagonal and hence the diagonal entries are nonzero.
LU Factorization
• Assume that A is an m × n matrix that can be row reduced to echelon form,
without row interchanges. Then A can be written in the form of

A = LU

where L is an m × m lower triangular matrix with 1’s on the diagonal and U is an


m × n matrix in row echelon form. 1

1
Caution: Some doesn’t require 1s along the diagonal entries.
LU Factorization
• Assume that A is an m × n matrix that can be row reduced to echelon form,
without row interchanges. Then A can be written in the form of

A = LU

where L is an m × m lower triangular matrix with 1’s on the diagonal and U is an


m × n matrix in row echelon form. 1

• Such a factorization is called an LU factorization of A. The matrix L is invertible


and is called a unit lower triangular matrix.
1
Caution: Some doesn’t require 1s along the diagonal entries.
LU Factorization: Algorithm

• Suppose A can be reduced to an echelon form U using only row replacements that
add a multiple of one row to another row below it. In this case, there exist unit
lower triangular elementary matrices E1 , E2 , ..., Ep such that

Ep · · · E 1 A = U

• Then
A = (Ep · · · E1 )−1 U = LU
where
L = (Ep · · · E1 )−1
• It can be shown that products and inverse of unit lower triangular matrices are
also unit lower triangular. Thus L is unit lower triangular.
Example 8.1. Find an LU factorization of
 
2 4 −1 5 −2
 −4 −5 3 −8 1 
 .
 2 −5 −4 1 8 
−6 0 7 −3 1
Example 8.1. Find an LU factorization of
 
2 4 −1 5 −2
 −4 −5 3 −8 1 
 .
 2 −5 −4 1 8 
−6 0 7 −3 1

Solution.
    
2 4 −1 5 −2 1 0 0 0 2 4 −1 5 −2
 −4 −5 3 −8 1   −2 1 0 0 
 0 3 1 2 3 
 
 =
 2 −5 −4 1 8   1 −3 1 0   0 0 0 2 1 
−6 0 7 −3 1 −3 4 2 1 0 0 0 0 5
Practice 8.2. Let
 
  1 2 3 2
6 9
A= and B =  1 3 2 1 
4 5
5 0 1 3

Find an LU factorization of each matrix.


LDU Factorization
If A = LU where the diagonal entries of L are all 1, then we can multiply row i of
matrix U by 1/uii and produce a row echelon matrix U ∗ with diagonal entries of 1.
We then have a factorization of A as A = LDU ∗ , where the diagonal entries of L and
U ∗ are all 1. (Note: When such a factorization of a matrix exists, it is unique.)
Example 8.3. Let
    
1 3 −1 1 0 0 1 3 −1
A =  2 8 4  = LU =  2 1 0   0 2 6 
−1 3 4 −1 3 1 0 0 −15

Factorize A into LDU.

You might also like