Lecture 15

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

3.

3 LU factorisation 63

3.3.3 Cholesky Factorisation

If there exists a lower triangular matrix L with positive


diagonal elements such that A = LL T , we can say that A
has a Cholesky factorization.

Theorem 3.3.3. If A is real symmetric, and positive definite


matrix, then it has a unique factorisation A = LL T , in which L is
lower triangular with a positive diagonal.

Proof. Clearly, A have a LU factorization (why?), that is, we


have lower triangular matrix L and upper triangular matrix
U such that A = LU. Since A is symmetric, we have

A = AT
=) LU = U T LT
1
=) U ( LT ) = L 1U T
=) lower triangular= upper triangular
1
=) U ( LT ) = L 1 U T = D, where D is diagonal
=) A = LDLT

Now since A is positive definite and L is non-singular,


D is also positive definite. Therefore, we can write A =
1 1 1
LD 2 D 2 L T = L̃ L̃ T , where L̃ = LD 2 .

Theorem 3.3.4. Show that Cholesky Factorisation of a matrix A


is unique.
64 matrices and system of linear equations

Proof. Suppose A = LL T = MM T where L and U are lower


triangular. Then

1 1
L M = LT ( MT )
1 1
=) L M = LT ( MT )
=) lower triangular= upper triangular
1 1
=) L M = LT ( MT ) =D

Now

1 1
L M = LT ( MT )
1 1 1
=) L M = (( L M)T )
1
=) D = ( D T )
=) DD T = I
=) D = I (why?)
1
=) L M = I =) L = M.

The elements of the lower triangular matrix can be found as,

8 q
<lk,k = ak,k Âkj=11 lk,j
2
⇣ ⌘
:li,k 1
= lk,k ai,k Âkj=11 li,j lk,j
2 , for i = k + 1, k + 2, . . . n

where k = 1, 2 . . . n. A pseudocode can be found in Algo-


rithm 11.
3.3 LU factorisation 65

Algorithm 11 Cholesky Factorization


Require: Symmetric positive definite matrix A 2 R n⇥n
Ensure: Lower triangular matrix L 2 R n⇥n such that A =
LL T
1: procedure CholeskyFactorization(A)
2: for k = 1 toqn do
3: Lk,k Ak,k Âkj=11 L2k,j
4: for i = k + 1 to n do
Ai,k Âkj=11 Li,j Lk,j
5: Li,k Lk,k
6: end for
7: end for
8: return L
9: end procedure

Example 3.10. Find the Cholesky factorisation of the following


matrix

2 3
4 2 2
A = 42 5 15 .
2 1 6

You might also like