an inverse eigenvalue problem for jacobi matrices · 2010. 11. 25. · a jacobi matrix inverse...

Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2011, Article ID 571781, 11 pagesdoi:10.1155/2011/571781

Research ArticleAn Inverse Eigenvalue Problem for Jacobi Matrices

Zhengsheng Wang1 and Baojiang Zhong2

1 Department of Mathematics, Nanjing University of Aeronautics and Astronautics,Nanjing 210016, China

2 School of Computer Science and Technology, Soochow University, Suzhou 215006, China

Correspondence should be addressed to Zhengsheng Wang, [email protected]

Received 25 November 2010; Revised 26 February 2011; Accepted 1 April 2011

Academic Editor: Jaromir Horacek

Copyright q 2011 Z. Wang and B. Zhong. This is an open access article distributed under theCreative Commons Attribution License, which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.

A kind of inverse eigenvalue problem is proposed which is the reconstruction of a Jacobi matrixby given four or five eigenvalues and corresponding eigenvectors. The solvability of the problemis discussed, and some sufficient conditions for existence of the solution of this problem areproposed. Furthermore, a numerical algorithm and two examples are presented.

1. Introduction

An n × n matrix J is called a Jacobi matrix if it is of the following form:

J �

⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣

a1 b1

b1 a2 b2

b2 a3 b3

. . . . . . . . .

bn−2 an−1 bn−1

bn−1 an

⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦

, bi > 0. �1.1�

A Jacobi matrix inverse eigenvalue problem, roughly speaking, is how to determinethe elements of Jacobi matrix from given eigen data. This kind of problem has great valuefor many applications, including vibration theory and structural design, for example, thevibrating rod model �1, 2�. In recent years, some new results have been obtained on the

2 Mathematical Problems in Engineering

construction of a Jacobi matrix �3, 4�. However, the problem of constructing a Jacobi matrixfrom its four or five eigenpairs has not been considered yet. The problem is as follows.

Problem 1. Given four different real scalars λ, μ, ξ, and η �supposed λ > μ > ξ > η�and four real orthogonal vectors of size nx � �x1, x2, . . . , xn�

T , y � �y1, y2, . . . , yn�T , m �

�m1, m2, . . . , mn�T , r � �r1, r2, . . . , rn�T , finding a Jacobi matrix J of size n such that�λ, x�, �μ, y�, �ξ,m�, and �η, r� are its four eigenpairs.

Problem 2. Given five different real scalars λ, μ, ν, ξ, and η �supposed λ > μ > ν > ξ >η� and five real orthogonal vectors of size nx � �x1, x2, . . . , xn�T , y � �y1, y2, . . . , yn�T , z ��z1, z2, . . . , zn�T , m � �m1, m2, . . . , mn�T , r � �r1, r2, . . . , rn�T , finding a Jacobi matrix J of size nsuch that �λ, x�, �μ, y�, �ν, z�, �ξ, m�, and �η, r� are its five eigenpairs.

In Sections 2 and 3, the sufficient conditions for the existence and uniqueness ofthe solution of Problems 1 and 2 are derived, respectively. Numerical algorithms and twonumerical examples are given in Section 4. We give conclusion and remarks in Section 5.

2. The Solvability Conditions of Problem 1

Lemma 2.1 �see �5, 6��. Given two different real scalars λ, μ �supposed λ > μ� and two realorthognal vectors of size n, x � �x1, x2, . . . , xn�T , y � �y1, y2, . . . , yn�T , there is a unique Jacobimatrix J such that �λ, x�, �μ, y� are its two eigenpairs if the following condition is satisfied:

dkDk

> 0, �k � 1, 2, . . . , n − 1�, �2.1�

where

dk �k∑i�1

xiyi, �k � 1, 2, . . . , n�,

Dk �

∣∣∣∣∣xk xk1

yk yk1

∣∣∣∣∣/� 0, �k � 1, 2, . . . , n − 1�.

�2.2�

And the elements of matrix J are

bk �

(λ − μ)dkDk

, �k � 1, 2, . . . , n − 1�,

a1 � λ − b1x2x1

,

Mathematical Problems in Engineering 3

an � λ − bn−1xn−1xn

,

ak �

⎧⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎩

λ − �bk−1xk−1 bkxk1�xk

, xk /� 0,

μ −(bk−1yk−1 bkyk1

)

yk, xk � 0,

�k � 2, 3, . . . , n − 1�.

�2.3�

FromLemma 2.1, we can see that under some conditions two eigenpairs can determinea unique Jacobi matrix. Therefore, for Problem 1, we only prove that the Jacobi matricesdetermined by �λ, x�, �μ, y� and �ξ,m�, �η, r� are the same.

The following theorem gives a sufficient condition for the uniqueness of the solutionof Problem 1.

Theorem 2.2. Problem 1 has a unique solution if the following conditions are satisfied:

�i� �λ − μ�d�1�k/D

�1�k

� �λ − ξ�d�2�k/D

�2�k

� �λ − η�d�3�k/D

�3�k

> 0;

�ii� if xk � 0, then �λ − μ�d�1�j /D�1�j � �μ − ξ�d

�4�j /D

�4�j � �μ − η�d

�5�j /D

�5�j , j � k, k − 1,

where

d�1�k

�k∑i�1

xiyi, d�2�k

�k∑i�1

ximi, d�3�k

�k∑i�1

xiri,

d�4�k

�k∑i�1

yimi, d�5�k

�k∑i�1

yiri, d�6�k

�k∑i�1

miri,

�k � 1, 2, . . . , n�, �2.4�

D�1�k �

∣∣∣∣∣yk yk1

xk xk1

∣∣∣∣∣, D�2�k �

∣∣∣∣∣mk mk1

xk xk1

∣∣∣∣∣, D�3�k �

∣∣∣∣∣rk rk1

xk xk1

∣∣∣∣∣,

D�4�k

�

∣∣∣∣∣mk mk1

yk yk1

∣∣∣∣∣, D�5�k

�

∣∣∣∣∣rk rk1

yk yk1

∣∣∣∣∣, D�6�k

�

∣∣∣∣∣rk rk1

mk mk1

∣∣∣∣∣,

�k � 1, 2, . . . , n − 1�.

�2.5�

Proof. According to Lemma 2.1, under certain condition, �λ, x� and �μ, y�, �λ, x� and �ξ,m�,�λ, x� and �η, r� can determine one unique Jacobi matrix, denoted J, J ′, J ′′, respectively. Their


elements are as follows:

bk �

(λ − μ)d�1�

k

D�1�k

, �k � 1, 2, . . . , n − 1�,

a1 � λ − b1x2x1

,


,

ak �

⎧⎪⎪⎪⎪⎨⎪⎪⎪⎪⎩


, xk /� 0,


)

yk, xk � 0,

�k � 2, 3, . . . , n − 1�,

�2.6�

b′k ��λ − ξ�d�2�k

D�2�k

, �k � 1, 2, . . . , n − 1�,

a′1 � λ −b1x2x1

,

a′n � λ −bn−1xn−1

xn,

a′k �

⎧⎪⎪⎪⎨⎪⎪⎪⎩


, xk /� 0,

ξ − �bk−1mk−1 bkmk1�mk

, xk � 0,

�k � 2, 3, . . . , n − 1�,

�2.7�

b′′k �

(λ − η)d�3�

k

D�3�k

, �k � 1, 2, . . . , n − 1�,

a′′1 � λ −b1x2x1

,

a′′n � λ −bn−1xn−1

xn,

a′′k �

⎧⎪⎪⎪⎨⎪⎪⎪⎩


, xk /� 0,

η − �bk−1rk−1 bkrk1�rk

, xk � 0,

�k � 2, 3, . . . , n − 1�.

�2.8�

From the conditions, we have

bk � b′k � b′′k > 0, k � 1, 2, . . . , n − 1. �2.9�


If xk /� 0, we have ak � a′k � a′′k; if xk � 0,

(λ − μ)d�1�

k

D�1�k

�

(μ − ξ)d�4�

k

D�4�k

,

(λ − μ)d�1�

k−1D

�1�k−1

�

(μ − ξ)d�4�

k−1D

�4�k−1

,

(λ − μ)d�1�kD

�1�k

�

(μ − η)d�4�kD

�4�k

,

(λ − μ)d�1�

k−1D

�1�k−1

�

(μ − η)d�4�

k−1D

�4�k−1

.

�2.10�

Since �2.6�, we have

bkD�4�k

�(μ − ξ)d�4�

k,

bk−1D�4�k−1 �

(μ − ξ)d�4�

k−1.�2.11�

That is,

(μ − ξ)ykmk bk−1D�4�k−1 − bkD

�4�k � 0. �2.12�

Since D�i�k /� 0 and xk � 0, we have yk /� 0, mk /� 0.D

�4�k−1 � mk−1yk −mkyk−1, D

�4�k

� mkyk1 −mk1yk replacingD�4�k−1, D�4�k

in �2.12�, then wehave


)

yk� ξ − �bk−1mk−1 bkmk1�

mk. �2.13�

Thus, if xk � 0, we also have ak � a′k. In the same way, we have ak � a′′k. Then, ak � a′k � a

′′k.

Therefore,

J � J ′ � J ′′ �2.14�

with four eigenpairs �λ, x�, �μ, y�, �ξ,m�, and �η, r�.


3. The Solvability Conditions of Problem 2

Lemma 3.1 �see �7��. Given three different real scalars λ, μ, ν (supposed λ > μ > ν) and threereal orthogonal vectors of size nx � �x1, x2, . . . , xn�

T , y � �y1, y2, . . . , yn�T , z � �z1, z2, . . . , zn�

T ,there is a unique Jacobi matrix J such that �λ, x�, �μ, y�, �ν, z� are its three eigenpairs if the followingconditions are satisfied:

�i� �λ − μ�d�1�k/D

�1�k

� �λ − ν�d�2�k/D

�2�k

> 0;

�ii� if xk � 0, �λ − μ�d�1�j /D�1�j � �μ − ν�d

�3�j /D

�3�j , j � k, k − 1, where

d�1�k �

k∑i�1

xiyi, d�2�k �

k∑i�1

xizi, d�3�k �

k∑i�1

yizi,

D�1�k

�

∣∣∣∣∣yk yk1

xk xk1

∣∣∣∣∣, D�2�k

�

∣∣∣∣∣zk zk1

xk xk1

∣∣∣∣∣, D�3�k

�

∣∣∣∣∣zk zk1

yk yk1

∣∣∣∣∣,

�k � 1, 2, . . . , n − 1�. �3.1�

And the elements of matrix J are

bk �

(λ − μ)dkDk

�k � 1, 2, . . . , n − 1�,

a1 � λ − b1x2x1

,


,

ak �

⎧⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎩


, xk /� 0,


)

yk, xk � 0,

�k � 2, 3, . . . , n − 1�.

�3.2�

From Lemma 3.1, we can see that under some conditions three eigenpairs can determinea unique Jacobi matrix. Therefore, for Problem 2, we only prove that the Jacobi matricesdetermined by �λ, x�, �μ, y�, �ν, z�; �λ, x�, �μ, y�, �ξ,m�, �λ, x�, �μ, y�, �η, r� are the same.

The following theorem gives a sufficient condition for the uniqueness of the solutionof Problem 2.


Theorem 3.2. Problem 2 has a unique solution if the following conditions are satisfied:

�i� �λ − μ�d�1�k /D�1�k � �λ − ν�d

�2�k /D

�2�k � �λ − ξ�d

�3�k /D

�3�k � �λ − η�d

�4�k /D

�4�k > 0;

�ii� if xk � 0, then �λ − μ�d�1�j /D�1�j � �μ − ν�d

�5�j /D

�5�j � �μ − ξ�d

�6�j /D

�6�j � �μ −

η�d�7�j /D�7�j , j � k, k − 1, where

d�1�k

�k∑i�1

xiyi, d�2�k

�k∑i�1

xizi, d�3�k

�k∑i�1

ximi,

d�4�k

�k∑i�1

xini, d�5�k

�k∑i�1

yizi, d�6�k

�k∑i�1

yimi,

d�7�k

�k∑i�1

yini, d�8�k

�k∑i�1

zimi, d�9�k

�k∑i�1

zini,

d�10�k

�k∑i�1mini, �k � 1, 2, . . . , n�

D�1�k

�

∣∣∣∣∣yk yk1

xk xk1

∣∣∣∣∣, D�2�k

�

∣∣∣∣∣zk zk1

xk xk1

∣∣∣∣∣, D�3�k

�

∣∣∣∣∣mk mk1

xk xk1

∣∣∣∣∣,

D�4�k

�

∣∣∣∣∣nk nk1

yk yk1

∣∣∣∣∣, D�5�k

�

∣∣∣∣∣zk zk1

yk yk1

∣∣∣∣∣, D�6�k

�

∣∣∣∣∣mk mk1

yk yk1

∣∣∣∣∣,

D�7�k �

∣∣∣∣∣nk nk1

yk yk1

∣∣∣∣∣, D�8�k �

∣∣∣∣∣mk mk1

zk zk1

∣∣∣∣∣, D�9�k �

∣∣∣∣∣nk nk1

zk zk1

∣∣∣∣∣,

D�10�k �

∣∣∣∣∣nk nk1

mk mk1

∣∣∣∣∣, �k � 1, 2, . . . , n − 1�.

�3.3�

Proof. According to Lemma 3.1, under certain condition, �λ, x�, �μ, y�, �ν, z�; �λ, x�, �μ, y�,�ξ,m�, �λ, x�, �μ, y�, �η, r� can determine one unique Jacobi matrix, denoted J, J ′, J ′′,respectively. Their elements are as follows:

bk �

(λ − μ)d�1�

k

D�1�k

�k � 1, 2, . . . , n − 1�,

a1 � λ − b1x2x1

,


,

ak �

⎧⎪⎪⎪⎨⎪⎪⎪⎩


, xk /� 0,


)

yk, xk � 0,

�k � 2, 3, . . . , n − 1�,


b′k�

(λ − μ)d�1�

k

D�1�k

, �k � 1, 2, . . . , n − 1�,

a′1 � λ −b1x2x1

,

a′n � λ −bn−1xn−1

xn,

a′k�

⎧⎪⎪⎪⎪⎨⎪⎪⎪⎪⎩


, xk /� 0,


)

yk, xk � 0,

�k � 2, 3, . . . , n − 1�,

b′′k�

(λ − μ)d�1�kD

�1�k

, �k � 1, 2, . . . , n − 1�,

a′′1 � λ −b1x2x1

,

a′′n � λ −bn−1xn−1

xn,

a′′k�

⎧⎪⎪⎪⎪⎨⎪⎪⎪⎪⎩


, xk /� 0,


)

yk, xk � 0,

�k � 2, 3, . . . , n − 1�,

�3.4�

From conditions �i� and �ii� we have obviously

bk � b′k � b′′k > 0, k � 1, 2, . . . , n − 1, ak � a′k � a′′k. �3.5�

Therefore,

J � J ′ � J ′′ �3.6�

with five eigenpairs �λ, x�, �μ, y�, �ν, z�, �ξ,m�, and �η, r�.

4. Numerical Algorithms and Examples

The process of the proof of the theorem provides us with a recipe for finding the solution ofProblem 1 if it exists.

From Theorem 2.2, we propose a numerical algorithm for finding the unique solutionof Problem 1 as follows.


Algorithm 1. Input. The real numbers λ > μ > ξ > η and mutually orthogonal vectorsx, y,m, r.

Output. The symmetric Jacobi matrix having the eigenpairs �λ, x�, �μ, y�, �ξ,m�, �η, r�:

�1� compute d�1�k, d

�2�k, d

�3�k, d

�4�k, d

�5�k, d

�6�k

and D�1�k, D

�2�k, D

�3�k, D

�4�k, D

�5�k, D

�6�k;

�2� if any one of D�1�k, D

�2�k, D

�3�k, D

�4�k, D

�5�k, D

�6�k

is zero, the Problem 1 can not be solvedby this method;

�3� for k � 1, 2, . . . , n − 1.�a� When xk � 0, if

(λ − μ)d�1�jD

�1�j

�

(μ − ξ)d�4�jD

�4�j

�

(μ − η)d�5�jD

�5�j

, j � k, k − 1, �4.1�

then

bk �

(λ − μ)d�1�

k

D�1�k

,

ak � μ −(bk−1yk−1 bkyk1

)

yk.

�4.2�

Otherwise, Problem 1 has no solution.

�b� When xk /� 0, if

(λ − μ)d�1�kD

�1�k

��λ − ξ�d�2�k

D�2�k

�

(λ − η)d�3�kD

�3�k

> 0, �4.3�

then

bk �

(λ − μ)d�1�kD

�1�k

,

ak � λ − �bk−1xk−1 bkxk1�xk

.

�4.4�

Otherwise, Problem 1 has no solution;

�4� an � λ − bn−1xn−1/xn.

Note that we can also propose a numerical algorithm from Theorem 3.2. Because ofthe limitation of space, we don’t describe it here in detail.

Now we give two numerical examples here to illustrate that the results obtained inthis paper are correct.


Example 4.1. Given four real numbers λ � 3, μ � 2, ξ � 1, η � 0.2679, and the four vectorsx � �1, 1, 0,−1,−1�T , y � �1, 0,−1, 0, 1�T , m � �1,−1, 0, 1,−1�T , r � �1,−√3, 2,−√3, 1�T , it iseasy to verify that these given data satisfy the conditions of the Theorem 2.2. After calculatingon the microcomputer through making program of Algorithm 1, we have a unique Jacobimatrix:

J �

⎡⎢⎢⎢⎢⎢⎢⎢⎢⎣

2 1

1 2 1

1 2 1

1 2 1

1 2

⎤⎥⎥⎥⎥⎥⎥⎥⎥⎦. �4.5�

Example 4.2. Given five real numbers λ � 7.543, μ � −3.543, ν � 2, ξ � 4.296, andη � −0.296, and the five vectors: x � �0.1913, 0.3536, 0.4619, 0.5000, 0.4619, 0.3536, 0.1913�T ,y � �0.1913,−0.3536, 0.4619,−0.5000, 0.4619,−0.3536, 0.1913�T , z � �0.5000, 0,−0.5000, 0,0.5000, 0,−0.5000�T , m � �0.4619, 0.3536,−0.1913, 0.5000,−0.1913, 0.3536, 0.4619�T , and r ��0.4619, −0.3536,−0.1913, 0.5000,−0.1913,−0.3536, 0.4619�T , it is easy to verify that these givennumbers can not satisfy the conditions of the Theorem 2.2 but Theorem 3.2. After calculatingon the microcomputer through making program of Theorem 3.2, we have a Jacobi matrix:

J �

⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣

2 3

3 2 3

3 2 3

3 2 3

3 2 3

3 2 3

3 2

⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦

. �4.6�

5. Conclusion and Remarks

As a summary, we have presented some sufficient conditions, as well as simple methods toconstruct a Jacobi matrix from its four or five eigenpairs. Numerical examples have beengiven to illustrate the effectiveness of our results and the proposed method. Also, the idea inthis paper may provide some insights for other banded matrix inverse eigenvalue problems.

Acknowledgments

This work is supported by the NUAA Research funding �Grant NS2010202� and theAviation Science Foundation of China �Grant 2009ZH52069�. The authors would like to thankProfessor Hua Dai for his valuable discussions.


References

�1� P. Nylen and F. Uhlig, “Inverse eigenvalue problem: existence of special spring-mass systems,” InverseProblems, vol. 13, no. 4, pp. 1071–1081, 1997.

�2� N. Radwan, “An inverse eigenvalue problem for symmetric and normal matrices,” Linear Algebra andIts Applications, vol. 248, pp. 101–109, 1996.

�3� Z. S. Wang, “Inverse eigenvalue problem for real symmetric five-diagonal matrix,” NumericalMathematics, vol. 24, no. 4, pp. 366–376, 2002.

�4� G. M. L. Gladwell, Inverse Problems in Vibration, vol. 119 of Solid Mechanics and Its Applications, KluwerAcademic Publishers, Dordrecht, The Netherlands, 2nd edition, 2004.

�5� H. Dai, “Inverse eigenvalue problems for Jacobi matrices and symmetric tridiagonal matrices,”Numerical Mathematics, vol. 12, no. 1, pp. 1–13, 1990.

�6� A. P. Liao, L. Zhang, and X. Y. Hu, “Conditions for the existence of a unique solution for inverseeigenproblems of tridiagonal symmetric matrices,” Journal on Numerical Methods and ComputerApplications, vol. 21, no. 2, pp. 102–111, 2000.

�7� X. Y. Hu and X. Z. Zhou, “Inverse eigenvalue problems for symmetric tridiagonal matrices,” Journal onNumerical Methods and Computer Applications, vol. 17, pp. 150–156, 1996.

Submit your manuscripts athttp://www.hindawi.com

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

MathematicsJournal of


Mathematical Problems in Engineering

Hindawi Publishing Corporationhttp://www.hindawi.com

Differential EquationsInternational Journal of

Volume 2014

Applied MathematicsJournal of


Probability and StatisticsHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Journal of


Mathematical PhysicsAdvances in

Complex AnalysisJournal of


OptimizationJournal of


CombinatoricsHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

International Journal of


Operations ResearchAdvances in

Journal of


Function Spaces

Abstract and Applied AnalysisHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

International Journal of Mathematics and Mathematical Sciences


The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014


Algebra

Discrete Dynamics in Nature and Society



Decision SciencesAdvances in

Discrete MathematicsJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com

Volume 2014 Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Stochastic AnalysisInternational Journal of

an inverse eigenvalue problem for jacobi matrices · 2010. 11. 25. · a jacobi matrix inverse...

Documents