Appendix B SOME PROPERTIES OF TRIDIAGONAL MATRICES

B.1 Standard Eigensystem for Simple Tridiagonals

In this work tridiagonal banded matrices are prevalent. It is useful to list some of their properties. Many of these can be derived by solving the simple linear di erence equations that arise in deriving recursion relations. Let us consider a simple tridiagonal matrix, i.e., a tridiagonal with constant scalar elements a,b, and c, see Section 3.4. If we examine the conditions under which the determinant of this matrix is zero, we nd (by a recursion exercise) det B (M : a b c)] = 0 if p m b + 2 ac cos M m=1 2 M +1 =0

From this it follows at once that the eigenvalues of B (a b c) are p m m=1 2 M (B.1) m = b + 2 ac cos M +1 The right-hand eigenvector of B (a b c) that is associated with the eigenvalue m satis es the equation B (a b c)~ xm = m~ xm (B.2) and is given by j;1 a m ~ xm = (xj )m = c 2 sin j M m=1 2 M (B.3) +1 257

258

APPENDIX B. SOME PROPERTIES OF TRIDIAGONAL MATRICES

These vectors are the columns of the right-hand eigenvector matrix, the elements of which are j;1 a jm j=1 2 M X = (xjm) = c 2 sin M (B.4) m=1 2 M +1 Notice that if a = ;1 and c = 1, j;1 a 2 = ei j; 2 (B.5) c
( 1)

B.2 Generalized Eigensystem for Simple Tridiagonals

This system is de ned as follows 2 32 3 2 32 3 x x b c e f 6 7 6 7 6 7 6 x 7 x 7 a b c d e f 6 7 6 6 7 6 7 6 7 6 7 6 7 6 6 7 6 7 6 7 6 7 x 7= 6 d e x 7 a b 6 7 6 7 6 7 . . . . 6 7 6 7 6 7 6 . . c 5 4 .. 5 4 . . f 5 4 .. 7 4 5 a b xM d e xM In this case one can show after some algebra that det B (a ; d b ; e c ; f ] = 0 if q m b ; m e + 2 (a ; md)(c ; m f ) cos M m=1 2 +1 =0 If we de ne m m= M + 1 pm = cos m
1 2 3 1 2 3

The left-hand eigenvector matrix of B (a b c) can be written m;1 2 c mj m=1 2 ; X = M + 1 a 2 sin M j=1 2 +1 In this case notice that if a = ;1 and c = 1 ; 1 c m2 = e;i m; 2 a
1 ( 1)

M M
(B.6)

(B.7)

M (B.8)

B.3. THE INVERSE OF A SIMPLE TRIDIAGONAL

eb 2(cd + af )pm + 2pm (ec fb)(ea bd) + (cd af )pm ] m= e 4fdpm The right-hand eigenvectors are " #j 1 2 a d m =1 2 M ~ xm = c mf sin j m ] j =1 2 M m These relations are useful in studying relaxation methods.
;
2

259
;
2

B.3 The Inverse of a Simple Tridiagonal

The inverse of B (a b c) can also be written in analytic form. Let DM represent the determinant of B (M : a b c)

DM det B (M : a b c)] De ning D to be 1, it is simple to derive the rst few determinants, thus D = 1 D = b D = b ac D = b 2abc (B.9) One can also nd the recursion relation DM = bDM ; acDM ; (B.10) Eq. B.10 is a linear O E the solution of which was discussed in Section 4.2. Its characteristic polynomial P (E ) is P (E bE + ac) and the two roots to P ( ) = 0 result in the solution 8" #M " #M 9 = < 1 b b 4ac b + b 4ac DM = 2 2 b 4ac : M =0 1 2 (B.11) where we have made use of the initial conditions D = 1 and D = b. In the limiting case when b 4ac = 0, one can show that
0 0 1 2 2

; ;

1 ;

+1

+1

b DM = (M + 1) 2

!M

b = 4ac
2

260

APPENDIX B. SOME PROPERTIES OF TRIDIAGONAL MATRICES

2 D3 6 16 ;aD2 B ;1 = D 6 4 a2 D1 4
;

Then for M = 4

aD
3

cD DD aD D aD
;
2 1 2 2 1 1

cD cD D DD aD
2 1 1 2 1 2

;
1

cD cD cD D
3 2 1 3 1

3 7 7 7 5

and for M = 5

aD The general element dmn is

4 0

2 D ;cD3 4 6 ;aD3 D1 D3 16 6 2 B ;1 = D 6 a D ; aD1 D2 6 32 54 2 ;a D1 a D1D1

;

aD
3

cD cD D DD aD D aD
2 2 1 2 2 2 2 2

;
1

cD cDD cD D DD aD
;
2 3 1 2 3 1

1 1

cD cD cD cD D
4 3 0 2 4

3 7 17 7 2 7 7 5 3

Upper triangle:

m=1 2

M 1
;
1

n=m+1 m+2
;

dmn = Dm; DM ;n( c)n;m =DM

Diagonal:

n=m=1 2 M dmm = DM ; DM ;m =DM

1

Lower triangle:

m =n+1 n+2

M
1 ;

n=1 2

M 1
;

dmn = DM ;mDn; ( a)m;n=DM

B.4.1 Standard Tridiagonals

B.4 Eigensystems of Circulant Matrices

Bp(M : a b c )
(B.12)

Consider the circulant (see Section 3.4.4) tridiagonal matrix

B.4. EIGENSYSTEMS OF CIRCULANT MATRICES

The eigenvalues are
m

261

= b + (a + c) cos 2 m M

m i(a c) sin 2M
;
)

m=0 1 2

The right-hand eigenvector that satis es Bp(a b c)~ xm = m~ xm is ~ xm = (xj )m = ei j m=M j=0 1 M ;1 (B.14) p where i ;1, and the right-hand eigenvector matrix has the form 2 m j =0 1 M ;1 X = (xjm) = eij M m =0 1 M ;1 The left-hand eigenvector matrix with elements x0 is ;im 2 j 1 m =0 1 M ;1 ; 0 M X = (xmj ) = M e j =0 1 M ;1 1 X , where X is the Note that both X and X ; are symmetric and that X ; = M conjugate transpose of X .
(2 1 1 1

M 1 (B.13)
;

Notice the remarkable fact that the elements of the eigenvector matrices X and X ; for the tridiagonal circulant matrix given by eq. B.12 do not depend on the elements a b c in the matrix. In fact, all circulant matrices of order M have the same set of linearly independent eigenvectors, even if they are completely dense. An example of a dense circulant matrix of order M = 4 is 2 b b b b 3 6 b b b b 7 6 7 6 (B.15) 4b b b b 7 5 b b b b The eigenvectors are always given by eq. B.14, and further examination shows that the elements in these eigenvectors correspond to the elements in a complex harmonic analysis or complex discrete Fourier series. Although the eigenvectors of a circulant matrix are independent of its elements, the eigenvalues are not. For the element indexing shown in eq. B.15 they have the general form M ; X bj ei jm=M m=
1 0 3 2 1 0 2 1 3 2 3 2 0 3 1 0 1 1 (2 )

B.4.2 General Circulant Systems

of which eq. B.13 is a special case.

j =0

B.5 Special Cases Found From Symmetries

262

APPENDIX B. SOME PROPERTIES OF TRIDIAGONAL MATRICES

Consider a mesh with an even number of interior points such as that shown in Fig. B.1. One can seek from the tridiagonal matrix B (2M : a b a ) the eigenvector subset that has even symmetry when spanning the interval 0 x . For example, we seek the set of eigenvectors ~ xm for which
2 3 2b a 32x 3 x1 1 6 7 6 6 7 x2 7 x2 7 6 6 7 a b a 6 7 6 7 6 7 6 7 . . . 6 7 6 7 . . . 6 7 a . . . 6 7 6 7 6 7 6 7 = 6 7 . m . . 6 .. a 7 . . 6 7 6 7 6 7 . . 6 7 6 7 6 7 6 7 6 4 5 a b a 4 x2 5 4 x2 7 5

a b

This leads to the subsystem of order M which has the form

2 b a 6 6 a b a 6 6 6 a ... B (M : a ~ b a)~ xm = 6 6 ... a 6 6 6 a b 4 3 7 7 7 7 7 ~ 7 xm = m~ xm 7 7 7 7 5

(B.16)

a a b+a

By folding the known eigenvectors of B (2M : a b a) about the center, one can show from previous results that the eigenvalues of eq. B.16 are (2m ; 1) m = b + 2a cos 2M + 1
!

m=1 2

(B.17)

B.6. SPECIAL CASES INVOLVING BOUNDARY CONDITIONS

and the corresponding eigenvectors are m ; 1) ~ xm = sin j (2 2M + 1 j =1 2 M Imposing symmetry about the same interval but for a mesh with an odd number of points, see Fig. B.1, leads to the matrix 2b a 3 6 7 a b a 6 7 6 7 . . 6 7 . a ~ 6 7 B (M : a b a) = 6 . 7 . . 6 a 7 6 4 5 a b a7 2a b By folding the known eigenvalues of B (2M ; 1 : a b a) about the center, one can show from previous results that the eigenvalues of eq. B.17 are (2m ; 1) m = b + 2a cos 2M and the corresponding eigenvectors are 1) ~ xm = sin j (2m 2M
;

263

Line of Symmetry

x =0

x=

j0 = 1 2 3 4 5 6 M0 j= 1 2 3 M a. An even-numbered mesh Line of Symmetry x =0 x= j0 = 1 2 3 4 5 M0 j= 1 2 3 M b. An odd{numbered mesh Figure B.1 { Symmetrical folds for special cases m=1 2 M

j=1 2

B.6 Special Cases Involving Boundary Conditions

We consider two special cases for the matrix operator representing the 3-point central di erence approximation for the second derivative @ =@x at all points away from the boundaries, combined with special conditions imposed at the boundaries.
2 2

264

APPENDIX B. SOME PROPERTIES OF TRIDIAGONAL MATRICES

Note: In both cases
;

m =1 2 M j =1 2 M 2 + 2 cos( ) = 4 sin ( =2)

;
2

When the boundary conditions are Dirichlet on both sides,

2 ;2 6 6 1 6 6 6 6 4

1 ;2 1

1 ;2 1

1 ;2 1

3 7 7 7 7 7 17 5

~ xm

m = ;2 h+ 2 cos M m i+ 1 = sin j M +1

(B.18)

When one boundary condition is Dirichlet and the other is Neumann (and a diagonal preconditioner is applied to scale the last equation),
2 ;2 6 6 1 6 6 6 6 4

1 ;2 1

1 ;2 1

1 ;2 1

3 7 7 7 7 7 17 5

~ xm

m ; 1) = ;2 + 2 cos (2 2M + 1 (2 m ; 1) = sin j 2M + 1

(B.19)