iterative techniques in matrix algebra...
TRANSCRIPT
![Page 1: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/1.jpg)
Iterative Techniques in Matrix Algebra
Eigenvalues & Eigenvectors
Numerical Analysis (9th Edition)R L Burden & J D Faires
Beamer Presentation Slidesprepared byJohn Carroll
Dublin City University
c© 2011 Brooks/Cole, Cengage Learning
![Page 2: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/2.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Outline
1 Introduction
2 The Characteristic Polynomial of a Matrix
3 The Spectral Radius of a Matrix
4 Convergent Matrices
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 2 / 33
![Page 3: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/3.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Outline
1 Introduction
2 The Characteristic Polynomial of a Matrix
3 The Spectral Radius of a Matrix
4 Convergent Matrices
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 2 / 33
![Page 4: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/4.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Outline
1 Introduction
2 The Characteristic Polynomial of a Matrix
3 The Spectral Radius of a Matrix
4 Convergent Matrices
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 2 / 33
![Page 5: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/5.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Outline
1 Introduction
2 The Characteristic Polynomial of a Matrix
3 The Spectral Radius of a Matrix
4 Convergent Matrices
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 2 / 33
![Page 6: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/6.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Outline
1 Introduction
2 The Characteristic Polynomial of a Matrix
3 The Spectral Radius of a Matrix
4 Convergent Matrices
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 3 / 33
![Page 7: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/7.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Matrix-Vector MultiplicationAn n ×m matrix can be considered as a function that uses matrixmultiplication to take m-dimensional column vectors inton-dimensional column vectors.
So an n ×m matrix is actually a linear function from IRm to IRn.A square matrix A takes the set of n-dimensional vectors intoitself, which gives a linear function from IRn to IRn.In this case, certain nonzero vectors x might be parallel to Ax,which means that a constant λ exists with
Ax = λx
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 4 / 33
![Page 8: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/8.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Matrix-Vector MultiplicationAn n ×m matrix can be considered as a function that uses matrixmultiplication to take m-dimensional column vectors inton-dimensional column vectors.So an n ×m matrix is actually a linear function from IRm to IRn.
A square matrix A takes the set of n-dimensional vectors intoitself, which gives a linear function from IRn to IRn.In this case, certain nonzero vectors x might be parallel to Ax,which means that a constant λ exists with
Ax = λx
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 4 / 33
![Page 9: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/9.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Matrix-Vector MultiplicationAn n ×m matrix can be considered as a function that uses matrixmultiplication to take m-dimensional column vectors inton-dimensional column vectors.So an n ×m matrix is actually a linear function from IRm to IRn.A square matrix A takes the set of n-dimensional vectors intoitself, which gives a linear function from IRn to IRn.
In this case, certain nonzero vectors x might be parallel to Ax,which means that a constant λ exists with
Ax = λx
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 4 / 33
![Page 10: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/10.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Matrix-Vector MultiplicationAn n ×m matrix can be considered as a function that uses matrixmultiplication to take m-dimensional column vectors inton-dimensional column vectors.So an n ×m matrix is actually a linear function from IRm to IRn.A square matrix A takes the set of n-dimensional vectors intoitself, which gives a linear function from IRn to IRn.In this case, certain nonzero vectors x might be parallel to Ax,which means that a constant λ exists with
Ax = λx
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 4 / 33
![Page 11: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/11.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Ax = λx
Matrix-Vector Multiplication (Cont’d)
For these vectors, we have
(A− λI)x = 0
There is a close connection between these numbers λ and thelikelihood that an iterative method will converge.We will consider this connection in this section.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 5 / 33
![Page 12: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/12.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Ax = λx
Matrix-Vector Multiplication (Cont’d)For these vectors, we have
(A− λI)x = 0
There is a close connection between these numbers λ and thelikelihood that an iterative method will converge.We will consider this connection in this section.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 5 / 33
![Page 13: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/13.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Ax = λx
Matrix-Vector Multiplication (Cont’d)For these vectors, we have
(A− λI)x = 0
There is a close connection between these numbers λ and thelikelihood that an iterative method will converge.
We will consider this connection in this section.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 5 / 33
![Page 14: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/14.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Ax = λx
Matrix-Vector Multiplication (Cont’d)For these vectors, we have
(A− λI)x = 0
There is a close connection between these numbers λ and thelikelihood that an iterative method will converge.We will consider this connection in this section.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 5 / 33
![Page 15: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/15.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Outline
1 Introduction
2 The Characteristic Polynomial of a Matrix
3 The Spectral Radius of a Matrix
4 Convergent Matrices
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 6 / 33
![Page 16: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/16.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & EigenvectorsDefinition: Characteristic PolynomialIf A is a square matrix, the characteristic polynomial of A is defined by
p(λ) = det(A− λI)
CommentsIt is not difficult to show that p is an nth-degree polynomial and,consequently, has at most n distinct zeros, some of which mightbe complex.If λ is a zero of p, then, since det(A− λI) = 0, we can prove thatthe linear system defined by
(A− λI)x = 0
has a solution with x 6= 0.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 7 / 33
![Page 17: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/17.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & EigenvectorsDefinition: Characteristic PolynomialIf A is a square matrix, the characteristic polynomial of A is defined by
p(λ) = det(A− λI)
CommentsIt is not difficult to show that p is an nth-degree polynomial and,consequently, has at most n distinct zeros, some of which mightbe complex.
If λ is a zero of p, then, since det(A− λI) = 0, we can prove thatthe linear system defined by
(A− λI)x = 0
has a solution with x 6= 0.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 7 / 33
![Page 18: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/18.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & EigenvectorsDefinition: Characteristic PolynomialIf A is a square matrix, the characteristic polynomial of A is defined by
p(λ) = det(A− λI)
CommentsIt is not difficult to show that p is an nth-degree polynomial and,consequently, has at most n distinct zeros, some of which mightbe complex.If λ is a zero of p, then, since det(A− λI) = 0, we can prove thatthe linear system defined by
(A− λI)x = 0
has a solution with x 6= 0.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 7 / 33
![Page 19: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/19.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Definition: Eigenvalues & EigenvectorsIf p is the characteristic polynomial of the matrix A, the zeros of pare eigenvalues, or characteristic values, of the matrix A.
If λ is an eigenvalue of A and x 6= 0 satisfies
(A− λI)x = 0
then x is an eigenvector, or characteristic vector, of Acorresponding to the eigenvalue λ.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 8 / 33
![Page 20: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/20.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Definition: Eigenvalues & EigenvectorsIf p is the characteristic polynomial of the matrix A, the zeros of pare eigenvalues, or characteristic values, of the matrix A.If λ is an eigenvalue of A and x 6= 0 satisfies
(A− λI)x = 0
then x is an eigenvector, or characteristic vector, of Acorresponding to the eigenvalue λ.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 8 / 33
![Page 21: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/21.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Finding the Eigenvalues & EigenvectorsTo determine the eigenvalues of a matrix, we can use the fact thatλ is an eigenvalue of A if and only if
det(A− λI) = 0
Once an eigenvalue λ has been found, a correspondingeigenvector x 6= 0 is determined by solving the system
(A− λI)x = 0
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 9 / 33
![Page 22: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/22.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Finding the Eigenvalues & EigenvectorsTo determine the eigenvalues of a matrix, we can use the fact thatλ is an eigenvalue of A if and only if
det(A− λI) = 0
Once an eigenvalue λ has been found, a correspondingeigenvector x 6= 0 is determined by solving the system
(A− λI)x = 0
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 9 / 33
![Page 23: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/23.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Example
Show that there are no nonzero vectors x in IR2 with Ax parallel to x if
A =
[0 1
−1 0
]
Solution (1/2)The eigenvalues of A are the solutions to the characteristic polynomial
0 = det(A− λI) = det[−λ 1−1 −λ
]= λ2 + 1
so the eigenvalues of A are the complex numbers λ1 = i and λ2 = −i .
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 10 / 33
![Page 24: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/24.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Example
Show that there are no nonzero vectors x in IR2 with Ax parallel to x if
A =
[0 1
−1 0
]
Solution (1/2)The eigenvalues of A
are the solutions to the characteristic polynomial
0 = det(A− λI) = det[−λ 1−1 −λ
]= λ2 + 1
so the eigenvalues of A are the complex numbers λ1 = i and λ2 = −i .
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 10 / 33
![Page 25: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/25.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Example
Show that there are no nonzero vectors x in IR2 with Ax parallel to x if
A =
[0 1
−1 0
]
Solution (1/2)The eigenvalues of A are the solutions to the characteristic polynomial
0 = det(A− λI) = det[−λ 1−1 −λ
]= λ2 + 1
so the eigenvalues of A are the complex numbers λ1 = i and λ2 = −i .
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 10 / 33
![Page 26: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/26.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Example
Show that there are no nonzero vectors x in IR2 with Ax parallel to x if
A =
[0 1
−1 0
]
Solution (1/2)The eigenvalues of A are the solutions to the characteristic polynomial
0 = det(A− λI) = det[−λ 1−1 −λ
]= λ2 + 1
so the eigenvalues of A are the complex numbers λ1 = i and λ2 = −i .
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 10 / 33
![Page 27: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/27.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Solution (2/2)A corresponding eigenvector x for λ1
needs to satisfy[00
]=
[−i 1−1 −i
] [x1x2
]=
[−ix1 + x2−x1 − ix2
]that is, 0 = −ix1 + x2, so x2 = ix1, and 0 = −x1 − ix2.Hence if x is an eigenvector of A, then exactly one of itscomponents is real and the other is complex.
As a consequence, there are no nonzero vectors x in IR2 with Axparallel to x.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 11 / 33
![Page 28: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/28.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Solution (2/2)A corresponding eigenvector x for λ1 needs to satisfy[
00
]=
[−i 1−1 −i
] [x1x2
]=
[−ix1 + x2−x1 − ix2
]
that is, 0 = −ix1 + x2, so x2 = ix1, and 0 = −x1 − ix2.Hence if x is an eigenvector of A, then exactly one of itscomponents is real and the other is complex.
As a consequence, there are no nonzero vectors x in IR2 with Axparallel to x.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 11 / 33
![Page 29: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/29.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Solution (2/2)A corresponding eigenvector x for λ1 needs to satisfy[
00
]=
[−i 1−1 −i
] [x1x2
]=
[−ix1 + x2−x1 − ix2
]that is, 0 = −ix1 + x2, so x2 = ix1, and 0 = −x1 − ix2.
Hence if x is an eigenvector of A, then exactly one of itscomponents is real and the other is complex.
As a consequence, there are no nonzero vectors x in IR2 with Axparallel to x.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 11 / 33
![Page 30: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/30.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Solution (2/2)A corresponding eigenvector x for λ1 needs to satisfy[
00
]=
[−i 1−1 −i
] [x1x2
]=
[−ix1 + x2−x1 − ix2
]that is, 0 = −ix1 + x2, so x2 = ix1, and 0 = −x1 − ix2.Hence if x is an eigenvector of A, then exactly one of itscomponents is real and the other is complex.
As a consequence, there are no nonzero vectors x in IR2 with Axparallel to x.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 11 / 33
![Page 31: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/31.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors
Solution (2/2)A corresponding eigenvector x for λ1 needs to satisfy[
00
]=
[−i 1−1 −i
] [x1x2
]=
[−ix1 + x2−x1 − ix2
]that is, 0 = −ix1 + x2, so x2 = ix1, and 0 = −x1 − ix2.Hence if x is an eigenvector of A, then exactly one of itscomponents is real and the other is complex.
As a consequence, there are no nonzero vectors x in IR2 with Axparallel to x.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 11 / 33
![Page 32: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/32.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & EigenvectorsGeometric Interpretation of λ
If λ is real and λ > 1, then A has the effect of stretching x by afactor of λ (see (a)).If 0 < λ < 1, then A shrinks x by a factor of λ (see (b)).If λ < 0, the effects are similar (see (c) and (d)), although thedirection of Ax is reversed.
Ax
x
Ax 5 lx
(a) l . 1 (b) 1 . l . 0 (c) l , 21 (d) 21 , l , 0
x
xx
Ax
Ax
Ax
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 12 / 33
![Page 33: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/33.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
ExampleDetermine the eigenvalues and eigenvectors for the matrix
A =
2 0 01 1 21 −1 4
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 13 / 33
![Page 34: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/34.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (1/4)The characteristic polynomial of A is
p(λ) = det(A− λI)
= det
2− λ 0 01 1− λ 21 −1 4− λ
= −(λ3 − 7λ2 + 16λ− 12)
= −(λ− 3)(λ− 2)2
so there are two eigenvalues of A: λ1 = 3 and λ2 = 2.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 14 / 33
![Page 35: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/35.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (1/4)The characteristic polynomial of A is
p(λ) = det(A− λI) = det
2− λ 0 01 1− λ 21 −1 4− λ
= −(λ3 − 7λ2 + 16λ− 12)
= −(λ− 3)(λ− 2)2
so there are two eigenvalues of A: λ1 = 3 and λ2 = 2.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 14 / 33
![Page 36: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/36.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (1/4)The characteristic polynomial of A is
p(λ) = det(A− λI) = det
2− λ 0 01 1− λ 21 −1 4− λ
= −(λ3 − 7λ2 + 16λ− 12)
= −(λ− 3)(λ− 2)2
so there are two eigenvalues of A: λ1 = 3 and λ2 = 2.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 14 / 33
![Page 37: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/37.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (1/4)The characteristic polynomial of A is
p(λ) = det(A− λI) = det
2− λ 0 01 1− λ 21 −1 4− λ
= −(λ3 − 7λ2 + 16λ− 12)
= −(λ− 3)(λ− 2)2
so there are two eigenvalues of A: λ1 = 3 and λ2 = 2.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 14 / 33
![Page 38: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/38.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (1/4)The characteristic polynomial of A is
p(λ) = det(A− λI) = det
2− λ 0 01 1− λ 21 −1 4− λ
= −(λ3 − 7λ2 + 16λ− 12)
= −(λ− 3)(λ− 2)2
so there are two eigenvalues of A: λ1 = 3 and λ2 = 2.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 14 / 33
![Page 39: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/39.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (2/4)An eigenvector x1 corresponding to the eigenvalue λ1 = 3 is a solutionto the vector-matrix equation (A− 3 · I)x1 = 0,
so 000
=
−1 0 01 −2 21 −1 1
· x1
x2x3
which implies that x1 = 0 and x2 = x3.
Any nonzero value of x3 produces an eigenvector for the eigenvalueλ1 = 3. For example, when x3 = 1 we have the eigenvectorx1 = (0, 1, 1)t , and any eigenvector of A corresponding to λ = 3 is anonzero multiple of x1.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 15 / 33
![Page 40: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/40.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (2/4)An eigenvector x1 corresponding to the eigenvalue λ1 = 3 is a solutionto the vector-matrix equation (A− 3 · I)x1 = 0, so 0
00
=
−1 0 01 −2 21 −1 1
· x1
x2x3
which implies that x1 = 0 and x2 = x3.
Any nonzero value of x3 produces an eigenvector for the eigenvalueλ1 = 3. For example, when x3 = 1 we have the eigenvectorx1 = (0, 1, 1)t , and any eigenvector of A corresponding to λ = 3 is anonzero multiple of x1.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 15 / 33
![Page 41: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/41.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (2/4)An eigenvector x1 corresponding to the eigenvalue λ1 = 3 is a solutionto the vector-matrix equation (A− 3 · I)x1 = 0, so 0
00
=
−1 0 01 −2 21 −1 1
· x1
x2x3
which implies that x1 = 0 and x2 = x3.
Any nonzero value of x3 produces an eigenvector for the eigenvalueλ1 = 3.
For example, when x3 = 1 we have the eigenvectorx1 = (0, 1, 1)t , and any eigenvector of A corresponding to λ = 3 is anonzero multiple of x1.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 15 / 33
![Page 42: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/42.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (2/4)An eigenvector x1 corresponding to the eigenvalue λ1 = 3 is a solutionto the vector-matrix equation (A− 3 · I)x1 = 0, so 0
00
=
−1 0 01 −2 21 −1 1
· x1
x2x3
which implies that x1 = 0 and x2 = x3.
Any nonzero value of x3 produces an eigenvector for the eigenvalueλ1 = 3. For example, when x3 = 1 we have the eigenvectorx1 = (0, 1, 1)t , and any eigenvector of A corresponding to λ = 3 is anonzero multiple of x1.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 15 / 33
![Page 43: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/43.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (3/4)An eigenvector x 6= 0 of A associated with λ2 = 2 is a solution of thesystem (A− 2 · I)x = 0,
so 000
=
0 0 01 −1 21 −1 2
· x1
x2x3
In this case the eigenvector has only to satisfy the equation
x1 − x2 + 2x3 = 0
which can be done in various ways.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 16 / 33
![Page 44: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/44.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (3/4)An eigenvector x 6= 0 of A associated with λ2 = 2 is a solution of thesystem (A− 2 · I)x = 0, so 0
00
=
0 0 01 −1 21 −1 2
· x1
x2x3
In this case the eigenvector has only to satisfy the equation
x1 − x2 + 2x3 = 0
which can be done in various ways.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 16 / 33
![Page 45: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/45.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (3/4)An eigenvector x 6= 0 of A associated with λ2 = 2 is a solution of thesystem (A− 2 · I)x = 0, so 0
00
=
0 0 01 −1 21 −1 2
· x1
x2x3
In this case the eigenvector has only to satisfy the equation
x1 − x2 + 2x3 = 0
which can be done in various ways.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 16 / 33
![Page 46: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/46.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (4/4)For example, when x1 = 0 we have x2 = 2x3, so one choice wouldbe x2 = (0, 2, 1)t .
We could also choose x2 = 0, which requires that x1 = −2x3.Hence x3 = (−2, 0, 1)t gives a second eigenvector for theeigenvalue λ2 = 2 that is not a multiple of x2.The eigenvectors of A corresponding to the eigenvalue λ2 = 2generate an entire plane. This plane is described by all vectors ofthe form
αx2 + βx3 = (−2β, 2α, α + β)t
for arbitrary constants α and β, provided that at least one of theconstants is nonzero.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 17 / 33
![Page 47: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/47.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (4/4)For example, when x1 = 0 we have x2 = 2x3, so one choice wouldbe x2 = (0, 2, 1)t .We could also choose x2 = 0, which requires that x1 = −2x3.
Hence x3 = (−2, 0, 1)t gives a second eigenvector for theeigenvalue λ2 = 2 that is not a multiple of x2.The eigenvectors of A corresponding to the eigenvalue λ2 = 2generate an entire plane. This plane is described by all vectors ofthe form
αx2 + βx3 = (−2β, 2α, α + β)t
for arbitrary constants α and β, provided that at least one of theconstants is nonzero.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 17 / 33
![Page 48: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/48.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (4/4)For example, when x1 = 0 we have x2 = 2x3, so one choice wouldbe x2 = (0, 2, 1)t .We could also choose x2 = 0, which requires that x1 = −2x3.Hence x3 = (−2, 0, 1)t gives a second eigenvector for theeigenvalue λ2 = 2 that is not a multiple of x2.
The eigenvectors of A corresponding to the eigenvalue λ2 = 2generate an entire plane. This plane is described by all vectors ofthe form
αx2 + βx3 = (−2β, 2α, α + β)t
for arbitrary constants α and β, provided that at least one of theconstants is nonzero.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 17 / 33
![Page 49: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/49.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Finding the Eigenvalues & Eigenvectors of A
Solution (4/4)For example, when x1 = 0 we have x2 = 2x3, so one choice wouldbe x2 = (0, 2, 1)t .We could also choose x2 = 0, which requires that x1 = −2x3.Hence x3 = (−2, 0, 1)t gives a second eigenvector for theeigenvalue λ2 = 2 that is not a multiple of x2.The eigenvectors of A corresponding to the eigenvalue λ2 = 2generate an entire plane. This plane is described by all vectors ofthe form
αx2 + βx3 = (−2β, 2α, α + β)t
for arbitrary constants α and β, provided that at least one of theconstants is nonzero.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 17 / 33
![Page 50: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/50.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Outline
1 Introduction
2 The Characteristic Polynomial of a Matrix
3 The Spectral Radius of a Matrix
4 Convergent Matrices
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 18 / 33
![Page 51: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/51.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
Definition: Spectral RadiusThe spectral radius ρ(A) of a matrix A is defined by
ρ(A) = max |λ|, where λ is an eigenvalue of A
(For complex λ = α + βi , we define |λ| = (α2 + β2)1/2.)
For the matrix in the previous example, namely
A =
2 0 01 1 21 −1 4
note that
ρ(A) = max{2, 3} = 3
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 19 / 33
![Page 52: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/52.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
Definition: Spectral RadiusThe spectral radius ρ(A) of a matrix A is defined by
ρ(A) = max |λ|, where λ is an eigenvalue of A
(For complex λ = α + βi , we define |λ| = (α2 + β2)1/2.)
For the matrix in the previous example, namely
A =
2 0 01 1 21 −1 4
note that
ρ(A) = max{2, 3} = 3
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 19 / 33
![Page 53: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/53.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
Definition: Spectral RadiusThe spectral radius ρ(A) of a matrix A is defined by
ρ(A) = max |λ|, where λ is an eigenvalue of A
(For complex λ = α + βi , we define |λ| = (α2 + β2)1/2.)
For the matrix in the previous example, namely
A =
2 0 01 1 21 −1 4
note that
ρ(A) = max{2, 3} = 3
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 19 / 33
![Page 54: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/54.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
The spectral radius is closely related to the norm of a matrix, as shownin the following theorem.
TheoremIf A is an n × n matrix, then
(i) ‖A‖2 = [ρ(AtA)]1/2
(ii) ρ(A) ≤ ‖A‖for any natural norm ‖ · ‖
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 20 / 33
![Page 55: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/55.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
The spectral radius is closely related to the norm of a matrix, as shownin the following theorem.
TheoremIf A is an n × n matrix,
then(i) ‖A‖2 = [ρ(AtA)]1/2
(ii) ρ(A) ≤ ‖A‖for any natural norm ‖ · ‖
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 20 / 33
![Page 56: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/56.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
The spectral radius is closely related to the norm of a matrix, as shownin the following theorem.
TheoremIf A is an n × n matrix, then
(i) ‖A‖2 = [ρ(AtA)]1/2
(ii) ρ(A) ≤ ‖A‖for any natural norm ‖ · ‖
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 20 / 33
![Page 57: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/57.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
The spectral radius is closely related to the norm of a matrix, as shownin the following theorem.
TheoremIf A is an n × n matrix, then
(i) ‖A‖2 = [ρ(AtA)]1/2
(ii) ρ(A) ≤ ‖A‖
for any natural norm ‖ · ‖
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 20 / 33
![Page 58: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/58.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
The spectral radius is closely related to the norm of a matrix, as shownin the following theorem.
TheoremIf A is an n × n matrix, then
(i) ‖A‖2 = [ρ(AtA)]1/2
(ii) ρ(A) ≤ ‖A‖for any natural norm ‖ · ‖
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 20 / 33
![Page 59: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/59.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
‖A‖2 = [ρ(AtA)]1/2
Proof (1/2)
The proof of part (i) requires more information concerningeigenvalues than we presently have available.For the details involved in the proof, see p. 21 of Ortega, J. M.,Numerical Analysis; a Second Course, Academic Press, NewYork, 1972, 201 pp.
Note: It can be shown that this result implies that if A is symmetric,then ‖A‖2 = ρ(A).
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 21 / 33
![Page 60: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/60.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
‖A‖2 = [ρ(AtA)]1/2
Proof (1/2)The proof of part (i) requires more information concerningeigenvalues than we presently have available.
For the details involved in the proof, see p. 21 of Ortega, J. M.,Numerical Analysis; a Second Course, Academic Press, NewYork, 1972, 201 pp.
Note: It can be shown that this result implies that if A is symmetric,then ‖A‖2 = ρ(A).
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 21 / 33
![Page 61: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/61.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
‖A‖2 = [ρ(AtA)]1/2
Proof (1/2)The proof of part (i) requires more information concerningeigenvalues than we presently have available.For the details involved in the proof, see p. 21 of Ortega, J. M.,Numerical Analysis; a Second Course, Academic Press, NewYork, 1972, 201 pp.
Note: It can be shown that this result implies that if A is symmetric,then ‖A‖2 = ρ(A).
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 21 / 33
![Page 62: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/62.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
‖A‖2 = [ρ(AtA)]1/2
Proof (1/2)The proof of part (i) requires more information concerningeigenvalues than we presently have available.For the details involved in the proof, see p. 21 of Ortega, J. M.,Numerical Analysis; a Second Course, Academic Press, NewYork, 1972, 201 pp.
Note: It can be shown that this result implies that if A is symmetric,then ‖A‖2 = ρ(A).
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 21 / 33
![Page 63: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/63.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
ρ(A) ≤ ‖A‖
Proof (2/2)To prove part (ii), suppose λ is an eigenvalue of A with eigenvector xand ‖x‖ = 1. Then Ax = λx and
|λ| = |λ| · ‖x‖ = ‖λx‖ = ‖Ax‖ ≤ ‖A‖‖x‖ = ‖A‖
Thusρ(A) = max |λ| ≤ ‖A‖
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 22 / 33
![Page 64: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/64.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
ρ(A) ≤ ‖A‖
Proof (2/2)To prove part (ii), suppose λ is an eigenvalue of A with eigenvector xand ‖x‖ = 1.
Then Ax = λx and
|λ| = |λ| · ‖x‖ = ‖λx‖ = ‖Ax‖ ≤ ‖A‖‖x‖ = ‖A‖
Thusρ(A) = max |λ| ≤ ‖A‖
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 22 / 33
![Page 65: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/65.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
ρ(A) ≤ ‖A‖
Proof (2/2)To prove part (ii), suppose λ is an eigenvalue of A with eigenvector xand ‖x‖ = 1. Then Ax = λx and
|λ| = |λ| · ‖x‖ = ‖λx‖ = ‖Ax‖ ≤ ‖A‖‖x‖ = ‖A‖
Thusρ(A) = max |λ| ≤ ‖A‖
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 22 / 33
![Page 66: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/66.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
ρ(A) ≤ ‖A‖
Proof (2/2)To prove part (ii), suppose λ is an eigenvalue of A with eigenvector xand ‖x‖ = 1. Then Ax = λx and
|λ| = |λ| · ‖x‖ = ‖λx‖ = ‖Ax‖ ≤ ‖A‖‖x‖ = ‖A‖
Thusρ(A) = max |λ| ≤ ‖A‖
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 22 / 33
![Page 67: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/67.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
ExampleDetermine the l2 norm of
A =
1 1 01 2 1
−1 1 2
Note: We will apply part (i) of the theorem, namely that
‖A‖2 = [ρ(AtA)]1/2
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 23 / 33
![Page 68: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/68.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
ExampleDetermine the l2 norm of
A =
1 1 01 2 1
−1 1 2
Note: We will apply part (i) of the theorem, namely that
‖A‖2 = [ρ(AtA)]1/2
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 23 / 33
![Page 69: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/69.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
‖A‖2 = [ρ(AtA)]1/2
Solution (1/3)We first need the eigenvalues of AtA, where
AtA =
1 1 −11 2 10 1 2
1 1 01 2 1
−1 1 2
=
3 2 −12 6 4
−1 4 5
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 24 / 33
![Page 70: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/70.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
‖A‖2 = [ρ(AtA)]1/2
Solution (2/3)If
0 = det(AtA− λI)
= det
3− λ 2 −12 6− λ 4−1 4 5− λ
= −λ3 + 14λ2 − 42λ
= −λ(λ2 − 14λ + 42)
then λ = 0 or λ = 7±√
7.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 25 / 33
![Page 71: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/71.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
‖A‖2 = [ρ(AtA)]1/2
Solution (2/3)If
0 = det(AtA− λI) = det
3− λ 2 −12 6− λ 4−1 4 5− λ
= −λ3 + 14λ2 − 42λ
= −λ(λ2 − 14λ + 42)
then λ = 0 or λ = 7±√
7.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 25 / 33
![Page 72: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/72.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
‖A‖2 = [ρ(AtA)]1/2
Solution (2/3)If
0 = det(AtA− λI) = det
3− λ 2 −12 6− λ 4−1 4 5− λ
= −λ3 + 14λ2 − 42λ
= −λ(λ2 − 14λ + 42)
then λ = 0 or λ = 7±√
7.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 25 / 33
![Page 73: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/73.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
‖A‖2 = [ρ(AtA)]1/2
Solution (2/3)If
0 = det(AtA− λI) = det
3− λ 2 −12 6− λ 4−1 4 5− λ
= −λ3 + 14λ2 − 42λ
= −λ(λ2 − 14λ + 42)
then λ = 0 or λ = 7±√
7.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 25 / 33
![Page 74: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/74.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
‖A‖2 = [ρ(AtA)]1/2
Solution (2/3)If
0 = det(AtA− λI) = det
3− λ 2 −12 6− λ 4−1 4 5− λ
= −λ3 + 14λ2 − 42λ
= −λ(λ2 − 14λ + 42)
then λ = 0 or λ = 7±√
7.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 25 / 33
![Page 75: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/75.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
‖A‖2 = [ρ(AtA)]1/2
Solution (3/3)By part (i) of the theorem, we have
||A||2 =√
ρ(AtA)
=
√max{0, 7−
√7, 7 +
√7}
=
√7 +
√7
≈ 3.106
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 26 / 33
![Page 76: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/76.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
‖A‖2 = [ρ(AtA)]1/2
Solution (3/3)By part (i) of the theorem, we have
||A||2 =√
ρ(AtA)
=
√max{0, 7−
√7, 7 +
√7}
=
√7 +
√7
≈ 3.106
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 26 / 33
![Page 77: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/77.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
‖A‖2 = [ρ(AtA)]1/2
Solution (3/3)By part (i) of the theorem, we have
||A||2 =√
ρ(AtA)
=
√max{0, 7−
√7, 7 +
√7}
=
√7 +
√7
≈ 3.106
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 26 / 33
![Page 78: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/78.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Spectral Radius
‖A‖2 = [ρ(AtA)]1/2
Solution (3/3)By part (i) of the theorem, we have
||A||2 =√
ρ(AtA)
=
√max{0, 7−
√7, 7 +
√7}
=
√7 +
√7
≈ 3.106
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 26 / 33
![Page 79: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/79.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Outline
1 Introduction
2 The Characteristic Polynomial of a Matrix
3 The Spectral Radius of a Matrix
4 Convergent Matrices
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 27 / 33
![Page 80: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/80.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Convergent Matrices
In studying iterative matrix techniques, it is of particular importance toknow when powers of a matrix become small (that is, when all theentries approach zero). Matrices of this type are called convergent.
Definition: Convergent MatrixWe call an n × n matrix A convergent if
limk→∞
(Ak )ij = 0, for each i = 1, 2, . . . , n and j = 1, 2, . . . , n
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 28 / 33
![Page 81: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/81.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Convergent Matrices
In studying iterative matrix techniques, it is of particular importance toknow when powers of a matrix become small (that is, when all theentries approach zero). Matrices of this type are called convergent.
Definition: Convergent MatrixWe call an n × n matrix A convergent if
limk→∞
(Ak )ij = 0, for each i = 1, 2, . . . , n and j = 1, 2, . . . , n
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 28 / 33
![Page 82: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/82.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Convergent Matrices
ExampleShow that
A =
[12 014
12
]is a convergent matrix.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 29 / 33
![Page 83: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/83.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Convergent Matrices
SolutionComputing powers of A, we obtain:
A2 =
[14 014
14
]A3 =
[18 0316
18
]A4 =
[116 018
116
]
and, in general,
Ak =
[(1
2)k 0k
2k+1 (12)k
]So A is a convergent matrix because
limk→∞
(12
)k
= 0 and limk→∞
k2k+1 = 0
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 30 / 33
![Page 84: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/84.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Convergent Matrices
SolutionComputing powers of A, we obtain:
A2 =
[14 014
14
]A3 =
[18 0316
18
]A4 =
[116 018
116
]
and, in general,
Ak =
[(1
2)k 0k
2k+1 (12)k
]
So A is a convergent matrix because
limk→∞
(12
)k
= 0 and limk→∞
k2k+1 = 0
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 30 / 33
![Page 85: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/85.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Convergent Matrices
SolutionComputing powers of A, we obtain:
A2 =
[14 014
14
]A3 =
[18 0316
18
]A4 =
[116 018
116
]
and, in general,
Ak =
[(1
2)k 0k
2k+1 (12)k
]So A is a convergent matrix because
limk→∞
(12
)k
= 0 and limk→∞
k2k+1 = 0
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 30 / 33
![Page 86: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/86.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Convergent Matrices
Notice that the convergent matrix A in the last example hasρ(A) = 1
2 , because 12 is the only eigenvalue of A.
This illustrates an important connection that exists between thespectral radius of a matrix and the convergence of the matrix,asdetailed in the following result.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 31 / 33
![Page 87: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/87.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Convergent Matrices
Notice that the convergent matrix A in the last example hasρ(A) = 1
2 , because 12 is the only eigenvalue of A.
This illustrates an important connection that exists between thespectral radius of a matrix and the convergence of the matrix,asdetailed in the following result.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 31 / 33
![Page 88: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/88.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Convergent Matrices
TheoremThe following statements are equivalent.
(i) A is a convergent matrix.(ii) limn→∞ ‖An‖ = 0, for some natural norm.(iii) limn→∞ ‖An‖ = 0, for all natural norms.(iv) ρ(A) < 1.(v) limn→∞ Anx = 0, for every x.
The proof of this theorem can be found on p. 14 of Issacson, E. and H.B. Keller, Analysis of Numerical Methods, John Wiley & Sons, NewYork, 1966, 541 pp.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 32 / 33
![Page 89: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/89.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Convergent Matrices
TheoremThe following statements are equivalent.
(i) A is a convergent matrix.
(ii) limn→∞ ‖An‖ = 0, for some natural norm.(iii) limn→∞ ‖An‖ = 0, for all natural norms.(iv) ρ(A) < 1.(v) limn→∞ Anx = 0, for every x.
The proof of this theorem can be found on p. 14 of Issacson, E. and H.B. Keller, Analysis of Numerical Methods, John Wiley & Sons, NewYork, 1966, 541 pp.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 32 / 33
![Page 90: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/90.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Convergent Matrices
TheoremThe following statements are equivalent.
(i) A is a convergent matrix.(ii) limn→∞ ‖An‖ = 0, for some natural norm.
(iii) limn→∞ ‖An‖ = 0, for all natural norms.(iv) ρ(A) < 1.(v) limn→∞ Anx = 0, for every x.
The proof of this theorem can be found on p. 14 of Issacson, E. and H.B. Keller, Analysis of Numerical Methods, John Wiley & Sons, NewYork, 1966, 541 pp.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 32 / 33
![Page 91: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/91.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Convergent Matrices
TheoremThe following statements are equivalent.
(i) A is a convergent matrix.(ii) limn→∞ ‖An‖ = 0, for some natural norm.(iii) limn→∞ ‖An‖ = 0, for all natural norms.
(iv) ρ(A) < 1.(v) limn→∞ Anx = 0, for every x.
The proof of this theorem can be found on p. 14 of Issacson, E. and H.B. Keller, Analysis of Numerical Methods, John Wiley & Sons, NewYork, 1966, 541 pp.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 32 / 33
![Page 92: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/92.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Convergent Matrices
TheoremThe following statements are equivalent.
(i) A is a convergent matrix.(ii) limn→∞ ‖An‖ = 0, for some natural norm.(iii) limn→∞ ‖An‖ = 0, for all natural norms.(iv) ρ(A) < 1.
(v) limn→∞ Anx = 0, for every x.
The proof of this theorem can be found on p. 14 of Issacson, E. and H.B. Keller, Analysis of Numerical Methods, John Wiley & Sons, NewYork, 1966, 541 pp.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 32 / 33
![Page 93: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/93.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Convergent Matrices
TheoremThe following statements are equivalent.
(i) A is a convergent matrix.(ii) limn→∞ ‖An‖ = 0, for some natural norm.(iii) limn→∞ ‖An‖ = 0, for all natural norms.(iv) ρ(A) < 1.(v) limn→∞ Anx = 0, for every x.
The proof of this theorem can be found on p. 14 of Issacson, E. and H.B. Keller, Analysis of Numerical Methods, John Wiley & Sons, NewYork, 1966, 541 pp.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 32 / 33
![Page 94: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/94.jpg)
Introduction Characteristic Polynomial Spectral Radius Convergent Matrices
Eigenvalues & Eigenvectors: Convergent Matrices
TheoremThe following statements are equivalent.
(i) A is a convergent matrix.(ii) limn→∞ ‖An‖ = 0, for some natural norm.(iii) limn→∞ ‖An‖ = 0, for all natural norms.(iv) ρ(A) < 1.(v) limn→∞ Anx = 0, for every x.
The proof of this theorem can be found on p. 14 of Issacson, E. and H.B. Keller, Analysis of Numerical Methods, John Wiley & Sons, NewYork, 1966, 541 pp.
Numerical Analysis (Chapter 7) Eigenvalues & Eigenvectors R L Burden & J D Faires 32 / 33
![Page 95: Iterative Techniques in Matrix Algebra …homen.vsb.cz/~lud0016/NM/Lecture_Notes_05-Eigenvalues.pdf · Iterative Techniques in Matrix Algebra Eigenvalues & Eigenvectors Numerical](https://reader031.vdocument.in/reader031/viewer/2022020304/5bb6c04109d3f23d358be62a/html5/thumbnails/95.jpg)
Questions?