Gram-Schmidt and QR Decomposition (Factorization) of Matrices
1. Gram-Schmidt and QR Decompostion
(Factorization) of Matrices
Isaac Amornortey Yowetu
NIMS-GHANA
September 24, 2020
2. 1 Introduction
2 Types of Matrices under consideration
3 QR Decompostion
4 Application of Gram-Schmidt Process to QR Decomposition
Isaac Amornortey Yowetu (NIMS-GHANA)Gram-Schmidt and QR Decompostion (Factorization) of MatricesSeptember 24, 2020 2 / 10
3. Introduction
Gram-Schmidt Process is one of the principal numerical
algorithms for computing QR Factorization.
Householder Triangularization is also one of the other algorithms
for computing QR Factorization.
In QR Decomposition;
Q is an Orthogonal Matrix with orthonormal basis.
R is a Right Upper Triangular Matrix.
A matrix A can be decomposed as A = QR.
Isaac Amornortey Yowetu (NIMS-GHANA)Gram-Schmidt and QR Decompostion (Factorization) of MatricesSeptember 24, 2020 3 / 10
4. Cases of Matrices
Considering An×m
Q-Factor
In a real space, if A is a square matrix(m=n), then
Q is an orthogonal unit vectors such that,
QT
Q = QQT
= I
if Q m × n and (m = n) with orthonormal columns, then
QT
Q = I
R-Factor
R is n × n right upper triangular matrix with rii = 0 (nonzero)
elements
R is nonsinular matrix
Isaac Amornortey Yowetu (NIMS-GHANA)Gram-Schmidt and QR Decompostion (Factorization) of MatricesSeptember 24, 2020 4 / 10
5. QR Decomposition
A = QR
a1 a2 · · · an = q1 q2 · · · qn
r11 r12 · · · r1n
0 r22 · · · r2n
...
...
...
...
0 0 · · · rnn
Isaac Amornortey Yowetu (NIMS-GHANA)Gram-Schmidt and QR Decompostion (Factorization) of MatricesSeptember 24, 2020 5 / 10
6. Question
Example 1
Consider the matrix
B =
−1 −1 1
1 3 3
−1 −1 5
1 3 7
using Gram-Schmidt process, determine the QR Factorization.
Isaac Amornortey Yowetu (NIMS-GHANA)Gram-Schmidt and QR Decompostion (Factorization) of MatricesSeptember 24, 2020 6 / 10
7. solution
Rewrite B as B = [b1, b2, b3] where
b1 =
−1
1
−1
1
, b2 =
−1
3
−1
3
, b3 =
1
3
5
7
Suppose {v1, v2, v3} ∈ V a set of linearly independent vectors and
{q1, q2, q3} ∈ Q a set of othornormal basis.
Let v1 = b1 =
−1
1
−1
1
, r11 = ||v1|| =
√
4 = 2
Isaac Amornortey Yowetu (NIMS-GHANA)Gram-Schmidt and QR Decompostion (Factorization) of MatricesSeptember 24, 2020 7 / 10