The Best Circulant Preconditioners for Hermitian Toeplitz Systems II: The Multiple-Zero Case Raymond H. Chan Michael K. Ng y Andy M. Yip z Abstract In

Size: px
Start display at page:

Download "The Best Circulant Preconditioners for Hermitian Toeplitz Systems II: The Multiple-Zero Case Raymond H. Chan Michael K. Ng y Andy M. Yip z Abstract In"

Transcription

1 The Best Circulant Preconditioners for Hermitian Toeplitz Systems II: The Multiple-ero Case Raymond H. Chan Michael K. Ng y Andy M. Yip z Abstract In [0, 4], circulant-type preconditioners have been proposed for ill-conditioned Hermitian Toeplitz systems that are generated by nonnegative continuous functions with a zero of even order. The proposed circulant preconditioners can be constructed without requiring explicit knowledge of the generating functions. It was shown that the spectra of the preconditioned matrices are uniformly bounded except for a xed number of outliers and that all eigenvalues are uniformly bounded away from zero. Therefore the conjugate gradient method converges linearly when applied to solving the circulant preconditioned systems. In [0, 4], it was mentioned that this result can be extended to the case where the generating functions have multiple zeros. The main aim of this paper is to give a complete convergence proof for this class of generating functions. Key Words. Toeplitz systems, circulant preconditioners, kernel functions, the preconditioned conjugate gradient method AMS(MOS) Subject Classications. 65F0, 65F5, 65T0 Introduction An n-by-n matrix A n with entries a ij is said to be Toeplitz if a ij a ij for all i; j. Toeplitz systems of the form A n x b occur in a variety of applications in mathematics and engineering [8]. In this paper, we consider the solutions of Hermitian positive denite Toeplitz systems. There are a number of specialized fast direct methods for solving such systems in O(n ) operations, see for instance [9]. Faster methods requiring O(n log n) operations have also been developed, see []. rchan@math.cuhk.edu.hk. Department of Mathematics, The Chinese University of Hong Kong, Shatin, Hong Kong. Research supported in part by Hong Kong Research Grants Council Grant No. CUHK 4/99P and CUHK DAG Grant No y mng@maths.hku.hk. Department of Mathematics, The University of Hong Kong, Pokfulam Road, Hong Kong. Research supported in part by HKU 747/99P and HKU CRCG Grant No z mhyipa@hkusua.hku.hk. Department of Mathematics, The University of Hong Kong, Pokfulam Road, Hong Kong.

2 Strang in [8] proposed using the preconditioned conjugate gradient method with circulant matrices as preconditioners for solving Toeplitz systems. The number of operations per iteration is of order O(n log n) as circulant systems can be solved eciently by fast Fourier transforms. Several successful circulant preconditioners have been introduced and analyzed; see for instance [, 6]. In these papers, the given Toeplitz matrix A n is assumed to be generated by a generating function f, i.e., the diagonals a j of A n are given by the Fourier coecients of f. It was shown that if f is a positive function in the Wiener class (i.e., the Fourier coecients of f are absolutely summable), then the corresponding circulant preconditioned systems converge superlinearly [6]. However, if f has zeros, the Toeplitz systems will be ill-conditioned. In fact, for the Toeplitz matrices generated by a nonnegative function having multiple zeros with maximal order p, their condition numbers grow like O(n p ), see [7, 6]. Hence the number of iterations required for convergence will increase like O(n p ), see [, p.4]. Tyrtyshnikov [0] has proved that the Strang [8] and the T. Chan [] preconditioners both fail in this case. To tackle this problem, non-circulant type preconditioners have been proposed, see [5, 7, 5]. The basic idea behind these preconditioners is to nd a function g that matches the zeros of f. Then the preconditioners are constructed based on this function g. These approaches work when the generating function f is given explicitly. However, when we are given only a nite n-by-n Toeplitz system, i.e., only fa j g jjj<n are given and the underlying f is unknown, then these preconditioners cannot be constructed. Recently, Chan, Yip and Ng [0] and Potts and Steidl [4] have independently proposed circulant-type preconditioners for these ill-conditioned Toeplitz systems. In [0], we construct the preconditioners by approximating f with the convolution product K m;r f that matches the zeros of f, where K m;r is chosen to be the generalized Jackson kernels, see []. We have proved that if f has a zero of order p, then K m;r f matches the zero of f when r > p. Using this result, we showed that the spectra of the circulant preconditioned matrices are uniformly bounded except for a xed number of outliers, and that their smallest eigenvalues are bounded uniformly away from zero. The construction of the preconditioners only requires the entries of A n and does not require explicit knowledge of f. In the numerical examples of [0], it has been shown that these circulant preconditioners are also eective for Toeplitz matrices that are generated by functions with multiple zeros. It was mentioned in both papers [0, 4] that this result can be extended to the case where the generating functions have multiple zeros of even order. The main aim of this paper is to give the convergence proof for this class of generating functions. In particular, we will prove that if f has s zeros with maximal order p, then K m;r f matches the zero of f when r > p. Using this result, we show that the spectra of the circulant preconditioned matrices are uniformly bounded except for a xed number of outliers, and that their smallest eigenvalues are bounded uniformly away from zero. It follows that the conjugate gradient method, when applied to solving these circulant preconditioned systems, will converge linearly. Since the cost per iteration is O(n log n) operations, see [8], the total complexity of solving these ill-conditioned Toeplitz systems is of O(n log n) operations. In contrast, the non-preconditioned systems will have condition number growing like O(n p )

3 and hence the complexity for solving the systems will be of order O(n p+ log n). The outline of the paper is as follows. In x, we introduce our circulant preconditioners from the generalized Jackson kernel K m;r. Some preliminary results on the approximation properties of K m;r are given in x. In x4, we show that K m;r f matches the zeros of f. In x5, we analyze the spectra of the circulant preconditioned matrices and their convergence rates. Finally, numerical examples are given in x6. Preconditioners from Generalized Jackson Kernel Let C be the space of all -periodic continuous real-valued functions. The Fourier coecients of a function f in C are given by a k f()e ik d; k 0; ; ; : Clearly a k a k for all k. Let A n [f] be the n-by-n Hermitian Toeplitz matrix with the (i; j)th entry given by a ij, i; j 0; : : : ; n. We will use C + to denote the space of all nonnegative functions in C which are not identically zero. We remark that the Toeplitz matrices A n [f] generated by f C + are positive denite for all n, see [7, Lemma ]. Conversely, if f C takes both positive and negative values, then A n [f] will be nondenite for large n. In this paper, we consider f C +. We say that 0 is a zero of f of order p if f( 0 ) 0 and p is the smallest positive integer such that f (p) ( 0 ) 6 0 and f (p+) () is continuous in a neighborhood of 0. By Taylor's theorem, f() f (p) ( 0 ) ( 0 ) p + O(( 0 ) p+ ) p! for all in that neighborhood. Since f is nonnegative, f (p) ( 0 ) > 0 and p must be even. In this paper, we consider f having zeros of order p j at j [; ] for j s. Such f can be written in the general form f() j ( j ) pj h(); () where h() is a positive function in C. We remark that the condition number of A n [f] generated by such an f grows like O(n pmax ) where p max max j p j, see [6]. The systems A n [f]x b will be solved by the preconditioned conjugate gradient method with circulant preconditioners. In the following, we will use the generalized Jackson kernel functions K m;r () k m;r m r! sin( m ) r sin( ) ; r ; ; : : : ()

4 to construct our circulant preconditioners. Here k m;r is a normalization constant such that It is known that 8 K m;r ()d : r k r m;r ; () 4 see [, p.57] or [0]. We note that K m; () is the Fejer kernel and K m;4 () is the Jackson kernel [, p.57]. For any m >, the kernel K m;r () can be expressed as K m;r () r(m) X kr(m) b (m;r) k e ik ; where the coecients b (m;r) k can be obtained by convolving the vector m ; ; m m ; ; m m ; ; m with itself for r times and this can be done by fast Fourier transforms, see [7, pp.94{ 96]. Thus the cost of computing the coecients fb (m;r) k g for all jkj r(m ) is of order O(rm log m) operations. In order to guarantee that K m;r ()d ; we normalize b (m;r) 0 to () by dividing all coecients b (m;r) k by b (m;r) 0. For a given function g, we dene the circulant preconditioner C n [g] to be the n-by-n circulant matrix with its j-th eigenvalue given by j j (C n [g]) g ; 0 j < n: (4) n We note that C n [g] Fndiag( 0 ; ; : : : ; n )F n, where F n is the n-by-n discrete Fourier matrix and F n is the conjugate transpose of F n, see [8]. Hence the matrix-vector multiplication C n [g]v, which is required in each iteration of the preconditioned conjugate gradient method, can be done in O(n log n) operations by fast Fourier transforms, see [8]. Clearly by (4), if g is a positive function, then C n [g] is positive denite for all n. For a given n-by-n Toeplitz matrix A n [f], our proposed circulant preconditioner is C n [K m;r f], where m dnre, i.e., r(m ) < n rm: (5) 4

5 Since f P k a ke ik, the convolution product of K m;r f is given by where (K m;r f)() d k ( r(m) X kr(m) a k b (m;r) k e ik n X kn+ a k b (m;r) k ; jkj r(m ); 0; otherwise: d k e ik ; Thus K m;r f depends only on a k for jkj < n, i.e., only on the entries of the given n-by-n Toeplitz matrix A n [f]. Notice that the cost of constructing C n [K m;r f] is of O(n log n) operations, see [0]. We rst note that since K m;r is a positive kernel, the preconditioners are all positive denite. Lemma. [0, Lemma.] Let f C +. The preconditioner C n[k m;r f] is positive denite for all positive integers m, n and r. Preliminary Lemmas In this section, we derive two results on the approximation properties of K m;r that will be required in the subsequent sections. In the following, we will use to denote the function dened on the whole real line R. For clarity, we will use to denote the periodic extension of dened on [; ], i.e. () ~ if ~ (mod ) and ~ [; ]. It is clear that p C+ We rst note that K m;r p matches the order of the zero of p Lemma. [0, Lemma.] Let p and r be positive integers. If r > p, then K m;r p (0) K m;r (t)t p dt c p;r m p ; where p p + for any integer p. at 0 if r > p. 4r 4r c p;r p+ : (6) However, if r p, then the order of the zeros of K m;r p and p may not be equal. Lemma. If r p, then where K m;r (t)t p dt c p;r m r ; p 4r m rp p r+p+ c p;r : (7) p + p r + 5

6 Proof: Since sin() on [0; ], we see from () that K m;r (t)t p dt r+ k m;r m r 0 sin r mt t rp dt p+ k m;r sin r u m p du 0 urp p+ k m;r r m p u p du 0 p+ k m;r r (p + )m p : In view of (), we have the rst inequality of (7). On the other hand, we also have K m;r (t)t p dt r k m;r m r pr+ r k m;r m p 0 p+ k m;r m p sin r mt t rp m 0 r m p+ k m;r m p (p r + ) Now the second inequality of (7) follows from (). 0 dt sin r u u u rp du rp du r m pr+ : Combining Lemmas. and., and noting that K m;r (t) is an even function whereas t j is an odd function for odd j, we have the following corollary. Corollary. For all positive integers j, m and r, we have In the next section, we have to estimate The following lemma will come in handy. c j;r K m;r (t)t j dt ; (8) mminfj;rg where c j;r 0 for odd j and the bounds for c j;r when j is even are given by (6) and (7). i () for [; ]. h Q s K m;r j ( j) p j Lemma.4 Let fp j g s j be positive integers and p P s ( ) p h K m;r Q s j p j. Then i j ( j) p j (( j ) () ) p j () (5 ) p ; (9) " K m;r ( j ) p j j 6 # ()

7 for all, j [; ]. In particular, if j k j, then " ( ) ppk for all, k [; ]. K m;r ( k ) p k " j;j6k ( j ) p j (( j ) () ) p j K m;r ( j ) p j j # () # () (5 ) p ; (0) Proof: For simplicity, let us write j(x) x p j (x ) p j (x + ) p j x p j (x () ) pj : () We rst claim that for each j it holds that ( ) pj j ( t j ) ( t j ) p j Let j j [; ], () is equivalent to ( ) pj j ( j t) ( j t) p j (5 ) pj ; 8t; ; j [; ]: () (5 ) pj ; 8t [; ]; j [; ]: () We prove () only for the case j [; 0]. The other case with j [0; ] can be proved similarly. By the denition of ( j t) p j, for any j [; 0], we have j( j t) (j t ) p j ( j t + ) p j ; t [; j + ]; ( j t) p j ( j t ) p j ( j t) p j ; t [ j + ; ]: For t [; j + ] and j [; 0], we have j t. Hence by considering the maximum and minimum of j(x )(x + )j on [; ], we obtain ( ) pj ( j t ) p j ( j t + ) pj (4 ) pj : For t [ j + ; ] and j [; 0], we have j t. Hence by considering the graph of jx(x )j on [; ], we get ( ) pj ( j t ) p j ( j t) pj (5 ) pj : Thus we have () and hence () too. By (), we see that ( ) p j j j( t j ) ( t j ) p j (5 ) p ; 7 8t; ; j [; ];

8 P s where p j p j. Hence we have 4 Km;r j ( j ) p j 5 () K m;r (t) ( ) p j ( t j ) p j K m;r (t) 4 Km;r ( ) p j j dt j( t j )dt j( j ) 5 (): Similarly, we obtain 4 Km;r j ( j ) p j5 () 4 Km;r (5 ) p j j( j ) 5 (): Hence (9) follows. To prove (0), we rst note that for j k tj, we have For j k tj, we have which implies that ( k t) p k ( k t) p k : (4) ( k t) p k (j k tj ) p k ; ( k t) p k ( k t) p pk k j k tj Together with (4), we have, for all j k j j k j, ( k t) p k ( k t) p k pk (5 ) p k ; Combining this with (), we see that for all j k j, pk : 8t [; ]: ( ) ppk ( t k ) p k Q s j;j6k j( t j ) j ( t j ) p j (5 ) p ; 8t [; ]: By using these two inequalities and the same arguments in establishing (9) above, we easily get (0). 8

9 4 Approximation by Using K m;r In this section, we use the results in x to show that for f given in (), the zeros of K m;r f match the order of the zeros of f and that the two functions are essentially the same away from the zeros of f. For simplicity, we divide the interval [; ] into subintervals according to the location of the zeros f j g s j of f. Let > 0 and L j (j ; j + ); if j 6 ; [; + ) [ ( ; ]; if j ; for j s. Note that there exists such that all the intervals L j for j s are disjoint and contained in [; ]. Clearly is less than half of the minimum gaps between the zeros. Dene, for j s and n >, I j;n [j n ; j + n ]; if j 6 ; [; + ] n [ [ n ; ]; if j ; J j;n L j n I j;n (j ; j n ) [ ( j + n ; j + ); if j 6 ; ( + n ; + ) [ ( ; n ); if j ; L [; ] n [ s j L j: In summary, the interval [; ] is divided in such a way that (a) I j;n is an arbitrary small interval containing the zero j ; (b) L j is a xed interval containing I j;n ; (c) L is a xed closed interval in which f takes positive values. Q s 4. The Function j ( j) p j on [s j J j;n Q s In this subsection, we show that K m;r j ( j) p j the same on [ s j J j;n. Q and s j ( j) p j are essentially Lemma 4. Let fp j g s j and r be positive integers with r > max j p j. Let m dnre. Then there exist positive numbers and independent of n such that for all suciently large n, " K m;r ( j ) p j j j ( j ) p j # () ; 8 [ s j J j;n: (5) 9

10 Proof: Without loss of generality, it suces to prove only for J ;n. In this case, by the construction of J ;n, we have n < j j and j j j for j s. In view of (0), it suces to show that 4 Km;r ( ) p j Q s j ( j) p j j( j ) 5 () ; 8 [ s j J j;n; (6) for some positive constants and. Here j () is dened in (). We begin with the upper bound. For, t, j [; ], we have j t j j. Hence by considering the graph of jx(x 4 )j on [; ], we have a bound for j : j( t j ) ( t j ) p j ( t j ) p j ( t j + ) pj (5 ) pj : Let p P s j p j. We rst note that 4 Km;r ( ) p (5 ) pp j K m;r (t)( t ) p K m;r (t)( t ) p K m;r (t) j( j ) 5 () j j p X p j0 j( t j ) dt (5 ) pj dt j ( ) p j (t) j dt: R Since K m;r(t)t j dt 0 for odd j and j j j < for j s, we have 4 Km;r ( ) p (5 ) pp Q s j ( j) p j 5 5 j Q s j ( j) p j pp pp j( j ) 5 () K m;r (t) K m;r (t) K m;r (t) 0 p X p j0 p p X j j0 p X p j0 j j ( ) p j t j dt ( ) j t j dt n j t j dt

11 5 pp p X j0 p j j n K m;r (t)t j dt: (7) Since r > max j p j p, by Lemma. and (5), j K m;r (t)t j dt n r j cj;r for each j 0; : : : ; p. Hence (7) is bounded above by a constant independent of n. Therefore we get an upper bound for (6). Next we establish a lower bound for (6). We have 4 Km;r ( ) p j K m;r (t)( t ) p K m;r (t) 4 ( ) p j( j ) 5 () j j j( t j ) dt j( j ) + u(t) 5 dt where u(t) ( t ) p j j( t j ) ( ) p j j( j ) 6p4p X j u j t j is a degree 6p 4p polynomial without the constant term. By (8), we have Thus 4 Km;r ( ) p j K m;r (t)u(t)dt pp X j u j j( j ) 5 () ( ) p c j;r m minfj;rg : j pp X u j c j;r j( j ) + m minfj;rg : Since j j and j j j for j 6, we have ( ) p ( ) p and pj ( j ) p j < ()p j for j 6. Hence, we have " K m;r ( ) p j j ( j ) p j j( j ) # () j

12 Q s j j( j ) Q s j ( j) p j ( ) pp () pp () pp p pp X p pp X j j ju j jc j;r m minfj;rg ju j jc j;r ; (8) mminfj;rg where the last inequality follows from () with t 0 there. Clearly, the second term of (8) tends to zero as m tends to innity. Thus for suciently large m (and hence large n), (8) is bounded uniformly from below by a positive constant. We therefore have established a lower bound for (6). We remark that if the gaps between j are small, then is small. Hence the upper bound in (7) (and therefore in (5)) will be large. Q s 4. The Function j ( j) p j on L Q s In this subsection, we show that K m;r j ( j) p j the same on L. Q and s j ( j) p j are essentially Lemma 4. Let fp j g s j, r be positive integers and m dnre. Then there exist positive numbers and independent of n such that for all suciently large n, " K m;r ( j ) p j j j ( j ) p j # () ; 8 L: Proof: First note that by the construction of L, there exist positive constants 0 and 0 such that 0 < 0 j ( j ) p j 0; 8 L: Let be a positive number less than 0. By using the fact that K m;r is a summation kernel [] with the property: lim m! kk m;r f f k 0; 8f C + ; where k k denotes the supremum norm, there exists a positive integer m 0 such that for m m 0 4 Km;r j ( j ) p j 5 () j ( j ) p j ; 8 L:

13 It follows that for all suciently large n (and hence m) 0 4 Km;r j j ( j ) p j5 () ( j ) p j + 0 ; 8 L: 4. Functions in C + on [s j J j;n [ L Now we extend the results of Lemmas 4. and 4. to any functions in C + zeros of even order. with multiple Lemma 4. Let f C + and have s zeros of order p j at j [; ] for j s. Let r > max j p j be any integer and m dnre. Then there exist positive numbers and, independent of n, such that for all suciently large n, (K m;r f) () f() ; 8 [ s j J j;n [ L: (9) Proof: Q s By denition (), f() g() dened on [; ]. Write j ( j) p j g() for some positive continuous function (K m;r f) () f() h K m;r Q s h Q s K m;r j ( j) p j i j ( j) p j g() () i () h Q i s K m;r j ( j) p j () Q s j ( j) p j g() : (0) Clearly the last factor is uniformly bounded above and below by positive constants. By Lemmas 4. and 4., the same holds for the second factor when [ s j J j;n [ L. As for the rst factor, by the mean value theorem for integrals [], there exists a [; ] such that Hence [K m;r 0 < g min j ( j ) p j g()]() g()[k m;r h K m;r Q s h Q s K m;r j ( j) p j j i j ( j) p j g() () i g max ; () ( j ) p j ](): 8 [; ];

14 where g min and g max are the minimum and maximum of g respectively. Thus the theorem follows. We remark that if the gaps between the zeros are small, then the upper bound of the second factor in (0) will be large (see the remark at the end of x4.). Hence in (9) will be large. 4.4 Functions in C + on [s j I j;n So far we have considered only the intervals [ s j J j;n [ L, i.e., away from the zeros of f. For [ s j I j;n, i.e., around the zeros of f, we now show that the convolution product K m;r f matches the order of the zeros of f. Lemma 4.4 Let f C + and have s zeros of order p j at j [; ] for j s. Let r > max j p j be any integer and m dnre. If I k;n for some k s, then we have (K m;r f) () O : Proof: We prove only for the case I ;n. The cases for I j;n for j s can be proved similarly. In this case we have j j n, Q j j j s and j j j 4 for j n. Since by (), f() j ( j) p j g() for some positive continuous function g() dened on [; ], by the mean value theorem for integrals again, we have g min 4 Km;r j n p k ( j ) p j5 () (Km;r f)() Hence, together with (0), it suces to show that 4 Km;r ( ) p j j( j ) 4 Km;r j g max 4 Km;r 5 () O ( j ) p j g() 5 () n p j ( j ) p j5 (): ; 8 I ;n : () We note that 4 Km;r ( ) p j j( j ) 5 () 4

15 j K m;r (t)( t ) p K m;r (t)( t ) p j Y 4 s j( t j )dt j( j ) + v(t) 5 dt j j( j )[K m;r ( ) p ]() K m;r (t)( t ) p v(t)dt; () where v(t) j j( t j ) j j( j ) 6p6p X is a degree 6p 6p polynomial without the constant term. For the rst term of (), since j j j and j j j 4 for j n, we have 6p6p j j v j t j j( j ) ( ) pp ; 8 I ;n : Moreover, by [0, Theorem.7] and the fact that r > max j p j p, we also have [K m;r ( ) p ]() O : Thus j n p j( j )[K m;r ( ) p ]() O For the second term of (), since j j n, we have p X j0 p X X j0 K m;r (t)( t ) p v(t)dt K m;r (t) 6p6p X k 6p6p k jv k j p j n p p X 6p6p p X ( ) pj t j v k t k dt j j0 k p v k ( ) p j K m;r (t)t j+k dt j j j p j K m;r (t)t j+k dt : () 5

16 p X j0 X 6p6p X k p 6p6p j0 X k jv k j p j jv k j p j p j n n p j K m;r (t)t j+k dt : c j+k;r m minfj+k;rg ; where the last equality follows from (8). Thus, we have p K m;r (t)( t ) p v(t)dt X O j0 n p j Putting this together with () into (), we get (). O m j+ O n p : + 5 Spectral Properties of the Preconditioned Matrices With the results in the previous sections, we now analyze the spectra and convergence rates of the preconditioned systems when the generating function has multiple zeros. We begin with the following lemma. Lemma 5. [5, ] Let f C +. Then A n[f] is positive denite for all n. Moreover if g C + is such that 0 < fg for some constants and, then for all n, x A n [f]x x A n [g]x ; 8x 6 0: With that, we have our rst main theorem which states that the spectra of the preconditioned matrices are essentially bounded. Theorem 5. Let f C + and have s zeros of order p j at j for j s. Let r > max j p j, m dnre, and p P s j p j. Then there exist positive numbers ; ( ), independent of n, such that for all suciently large n, at most p + s eigenvalues of C n [K m;r f]a n [f] are outside the interval [; ]. Proof: For any function g C, we let Cn ~ [g] to be the n-by-n circulant matrix with the j-th eigenvalue given by j ( ~ Cn [g]) 8 >< >: n p ; k j g n if j n ; otherwise; k < n for some k ; ; s for j 0; : : : ; n. Since, for large n, there is at most one j such that jjn k j < n for each k s, by (4), ~ Cn [g] C n [g] is a matrix of rank at most s. 6 (4)

17 Q s By assumption (), f() j sinp j (( j ))g() for some positive function g in C. We use the following decomposition of the Rayleigh quotient to prove the theorem: x A n [f]x x C n [K m;r f]x x A n [f]x h x Qs A n j sinp j j x ~ Cn h Qs j sinp j j x ~ C n [f]x h x i Qs A n j sinp j j i x i x x i x x ~ C n h Qs j sinp j x ~ Cn [f]x x ~ C n [K m;r f]x j x ~ Cn [K m;r f]x x C n [K m;r f]x : (5) We remark that by Lemma 5. and the denitions (4) and (4), all matrices in the factors in the right hand side of (5) are positive denite. As g is a positive function in C, by Lemma 5., the rst factor in the right hand side of (5) is uniformly bounded above and below. Similarly, by denition (4), the third factor is also uniformly bounded. The eigenvalues of the two circulant matrices in the fourth factor dier only when jjn k j n for all k s. But by Lemma 4., the ratios of these eigenvalues are all uniformly bounded when n is large. The eigenvalues of the two circulant matrices in the last factor dier only when jjn k j < n for some k s. But by Lemma 4.4, their Q ratios are also uniformly bounded. Regarding the second factor, Q s since A n [ j sinp j ( j )] is a banded Toeplitz matrix with Q s half bandwidth p +, C n [ j sinp j ( j )] is just the Strang preconditioner for s A n [ j sinp j ( j )] when n > p, see [9]. Thus the two matrices dier only by a rank p matrix. It follows that the matrix A n [ Q s j sinp j ( j )] ~ Cn [ Q s j sinp j ( j )] is of rank at most p + s. The rest of the proof now follows exactly the same line of the proof in [0, Theorem 4.]. In view of the remark at the end of x4., we see that if the gaps between the zeros are small, then the lower bound for the fourth factor in (5) will be small. Hence the condition numbers of our preconditioned matrices, althought independent of n, will be large. We will observe this in the numerical examples in x6. Finally, by using exactly the same argument used in [0, Theorem 4.], we show that all the eigenvalues of the preconditioned matrices are uniformly bounded away from zero. More precisely, we have Theorem 5. Let f C + and have s zeros of order p j at j for j s. Let r > max j p j and m dnre. Then there exists a positive constant c independent of n, such that for all n suciently large, all eigenvalues of the preconditioned matrix C n [K m;r f]a n [f] are larger than c. By combining Theorems 5. and 5., the number of preconditioned conjugate gradient (PCG) iterations required for convergence is of O(), see [4]. We recall that each PCG 7

18 iteration requires O(n log n) operations (see [8]) and so is the construction of the preconditioner (see x). Thus the total complexity of the PCG method for solving an n-by-n Toeplitz systems generated by f C + with multiple zeros of even order is of O(n log n) operations. In contrast, for the original matrix A n [f], since its condition number is growing like O(n pmax ), where p max max j p j, the complexity of solving the corresponding system will be of order O(n pmax+ log n). 6 Numerical Experiments In [0], we have already illustrated by numerical examples the eectiveness of the preconditioner C n [K m;r f] in solving Toeplitz systems. In particular, we have tested Toeplitz systems generated by functions having multiple zeros, and found that the circulant preconditioned system converge linearly, i.e., the number of iterations required for convergence is independent of n. This is in line with the theoretical results we have proven in x5 above. In this section, we further consider the function f (x d) (x + d) for dierent values of d. By the remark after Theorem 5., d will aect the condition number of the preconditioned matrices. In the test, m was set to dnre. The right-hand side vectors b were formed by multiplying random vectors to A n [f]. The initial guess is the zero vector and the stopping criteria is jjr q jj jjr 0 jj 0 7 where r q is the residual vector after q iterations. We also compared the performance of our preconditioners with the Strang [8] and the T. Chan [] circulant preconditioners. Tables and show the numbers of iterations required for convergence for dierent choices of preconditioners. In the tables, I denotes no preconditioner, S is the Strang preconditioner [8], K m;r are the preconditioners from the generalized Jackson kernel K m;r dened in () and T K m; is the T. Chan preconditioner [9]. Iteration numbers more than 6,000 are denoted by \y". We note that S in general is not positive denite as the Dirichlet kernel is not positive, see [9]. When some of its eigenvalues are negative, we denote the iteration number by \{" as the PCG method does not apply to non-denite systems and the solution thus obtained may be inaccurate. The test function in Table is a nonnegative function with two well separated zeros of order on [; ]. Thus the condition number of the Toeplitz matrices is O(n ) and hence the number of iterations required for convergence without using any preconditioners is increasing like O(n). We see that the T. Chan preconditioner does not work. This is to be expected from the fact that the order of K m; does not match that of at 0, see (7). However, we see that K m;4, K m;6 and K m;8 all work very well as predicted from our convergence analysis in x5. The zeros of the test function in Table are close together and hence the function behaves like 4 since ( 0:00) ( + 0:00) 4. Thus the eective condition numbers of the Toeplitz matrices is like O(n 4 ) and the systems are very ill-conditioned even for moderate n. As in the rst example, the T. Chan preconditioner does not work. However, K m;6 and K m;8 do work very well. This can also be explained by our Theorems 5. 8

19 ( ) ( + ) n I S 0 { T K N;4 4 K N;6 4 4 K N; Table : Numbers of iterations required for dierent preconditioners. and 5.. Since the separation d in this example is small, the condition numbers of our preconditioned matrices are large, cf. the remark after Theorem 5.. We see from Table that the numbers of iterations required for convergence by using our preconditioners are larger than those in Table. ( 0:00) ( + 0:00) n I y S { { { { { { T K N; K N; K N; Table : Numbers of iterations for dierent preconditioners. Example illustrates that for a function with zeros close together, it has eectively one zero with order being the sum of the orders of the zeros. To further illustrate this, we give in Figures and the spectra of the preconditioned matrices for all ve preconditioners for the two test functions when n 8. Notice the dierence between the spectra of the two preconditioned matrices for K m;4. We also see that the spectra of the preconditioned matrices for K m;6 and K m;8 are clustered in a small interval around except for one to two large outliers and that all the eigenvalues are well bounded away from 0. We note that the Strang preconditioned matrices in both cases have negative eigenvalues that are not depicted in the gures. 9

20 7 6 K Jackson Preconditioner m,8 5 K Jackson Preconditioner m,6 4 K Jackson Preconditioner m,4 T. Chan Preconditioner Strang Preconidtioner No Preconidtioner Figure : Spectra of preconditioned matrices for f() ( ) ( + ) when n K Jackson Preconditioner m,8 5 K Jackson Preconditioner m,6 4 K Jackson Preconditioner m,4 T. Chan Preconditioner Strang Preconidtioner (has negative eigenvalues) No Preconidtioner Figure : Spectra of preconditioned matrices for f() ( 0:00) ( + 0:00) n 8. when 0

21 References [] G. Ammar and W. Gragg, Superfast solution of real positive denite Toeplitz systems, SIAM J. Matrix Anal. Appl., 9 (988), pp. 6{67. [] T. Apostol, Mathematical Analysis, Addison-Wesley, 97. [] O. Axelsson and V. Barker, Finite Element Solution of Boundary Value Problems, Theory and Computation, Academic Press, Orlando, 984. [4] F. Di Benedetto, Analysis of preconditioning techniques for ill-conditioned Toeplitz matrices, SIAM J. Sci. Comput., 6 (995), pp. 68{697. [5] F. Di Benedetto, G. Fiorentino and S. Serra, C.G. preconditioning for Toeplitz matrices, Comput. Math. Appl., 5 (99), pp. 5{45. [6] R. Chan, Circulant Preconditioners for Hermitian Toeplitz Systems, SIAM J. Matrix Anal. Appl., 0 (989), pp. 54{550. [7] R. Chan, Toeplitz Preconditioners for Toeplitz Systems with Nonnegative Generating Functions, IMA J. Numer. Anal., (99), pp. {45. [8] R. Chan and M. Ng, Conjugate Gradient Methods for Toeplitz Systems, SIAM Review, 8 (996), pp. 47{48. [9] R. Chan and M. Yeung, Circulant Preconditioners Constructed from Kernels, SIAM J. Numer. Anal., 9 (99), pp. 09{0. [0] R. Chan, A. Yip and M. Ng, The Best Circulant Preconditioners for Hermitian Toeplitz Systems, Math. Dept. Res. Rept. 99-, The Chinese University of Hong Kong, 999. [] T. Chan, An Optimal Circulant Preconditioner for Toeplitz Systems, SIAM J. Sci. Statist. Comput., 9 (988), pp. 766{77. [] G. Lorentz, Approximation of Functions, Holt, Rinehart and Winston, New York, 966. [] D. Potts and G. Steidl, Preconditioners for Ill-Conditioned Toeplitz Matrices, BIT, 9 (999), pp. 5{5. [4] D. Potts and G. Steidl, Preconditioners for Ill-Conditioned Toeplitz Systems Constructed from Positive Kernels, preprint, 999. [5] S. Serra, Preconditioning Strategies for Hermitian Toeplitz Systems with Nondenite Generating Functions, SIAM J. Matrix Anal. Appl., 7 (996), pp. 007{09.

22 [6] S. Serra, On the extreme eigenvalues of Hermitian (block) Toeplitz matrices, Lin. Alg. Appl., 70 (998), pp. 09{9. [7] G. Strang, Introduction to Applied Mathematics, Wellesley-Cambridge Press, Cambridge, 986. [8] G. Strang, A Proposal for Toeplitz Matrix Calculations, Stud. Appl. Math., 74 (986), pp. 7{76. [9] W. Trench, An Algorithm for the Inversion of Finite Toeplitz Matrices, SIAM J. Appl. Math., (964), pp. 55{5. [0] E. Tyrtyshnikov, Circulant Preconditioners with Unbounded Inverses, Lin. Alg. Appl., 6 (995), pp. {4.

Toeplitz-circulant Preconditioners for Toeplitz Systems and Their Applications to Queueing Networks with Batch Arrivals Raymond H. Chan Wai-Ki Ching y

Toeplitz-circulant Preconditioners for Toeplitz Systems and Their Applications to Queueing Networks with Batch Arrivals Raymond H. Chan Wai-Ki Ching y Toeplitz-circulant Preconditioners for Toeplitz Systems and Their Applications to Queueing Networks with Batch Arrivals Raymond H. Chan Wai-Ki Ching y November 4, 994 Abstract The preconditioned conjugate

More information

Preconditioning of elliptic problems by approximation in the transform domain

Preconditioning of elliptic problems by approximation in the transform domain TR-CS-97-2 Preconditioning of elliptic problems by approximation in the transform domain Michael K. Ng July 997 Joint Computer Science Technical Report Series Department of Computer Science Faculty of

More information

1 Introduction The conjugate gradient (CG) method is an iterative method for solving Hermitian positive denite systems Ax = b, see for instance Golub

1 Introduction The conjugate gradient (CG) method is an iterative method for solving Hermitian positive denite systems Ax = b, see for instance Golub Circulant Preconditioned Toeplitz Least Squares Iterations Raymond H. Chan, James G. Nagy y and Robert J. Plemmons z November 1991, Revised April 1992 Abstract We consider the solution of least squares

More information

Preconditioners for ill conditioned (block) Toeplitz systems: facts a

Preconditioners for ill conditioned (block) Toeplitz systems: facts a Preconditioners for ill conditioned (block) Toeplitz systems: facts and ideas Department of Informatics, Athens University of Economics and Business, Athens, Greece. Email:pvassal@aueb.gr, pvassal@uoi.gr

More information

Approximate Inverse-Free Preconditioners for Toeplitz Matrices

Approximate Inverse-Free Preconditioners for Toeplitz Matrices Approximate Inverse-Free Preconditioners for Toeplitz Matrices You-Wei Wen Wai-Ki Ching Michael K. Ng Abstract In this paper, we propose approximate inverse-free preconditioners for solving Toeplitz systems.

More information

BLOCK DIAGONAL AND SCHUR COMPLEMENT PRECONDITIONERS FOR BLOCK TOEPLITZ SYSTEMS WITH SMALL SIZE BLOCKS

BLOCK DIAGONAL AND SCHUR COMPLEMENT PRECONDITIONERS FOR BLOCK TOEPLITZ SYSTEMS WITH SMALL SIZE BLOCKS BLOCK DIAGONAL AND SCHUR COMPLEMENT PRECONDITIONERS FOR BLOCK TOEPLITZ SYSTEMS WITH SMALL SIZE BLOCKS WAI-KI CHING, MICHAEL K NG, AND YOU-WEI WEN Abstract In this paper we consider the solution of Hermitian

More information

Preconditioners for ill conditioned Toeplitz systems constructed from positive kernels

Preconditioners for ill conditioned Toeplitz systems constructed from positive kernels Preconditioners for ill conditioned Toeplitz systems constructed from positive kernels Daniel Potts Medizinische Universität Lübeck Institut für Mathematik Wallstr. 40 D 23560 Lübeck potts@math.mu-luebeck.de

More information

Cosine transform preconditioners for high resolution image reconstruction

Cosine transform preconditioners for high resolution image reconstruction Linear Algebra and its Applications 36 (000) 89 04 www.elsevier.com/locate/laa Cosine transform preconditioners for high resolution image reconstruction Michael K. Ng a,,, Raymond H. Chan b,,tonyf.chan

More information

Positive Denite Matrix. Ya Yan Lu 1. Department of Mathematics. City University of Hong Kong. Kowloon, Hong Kong. Abstract

Positive Denite Matrix. Ya Yan Lu 1. Department of Mathematics. City University of Hong Kong. Kowloon, Hong Kong. Abstract Computing the Logarithm of a Symmetric Positive Denite Matrix Ya Yan Lu Department of Mathematics City University of Hong Kong Kowloon, Hong Kong Abstract A numerical method for computing the logarithm

More information

Superlinear convergence for PCG using band plus algebra preconditioners for Toeplitz systems

Superlinear convergence for PCG using band plus algebra preconditioners for Toeplitz systems Computers and Mathematics with Applications 56 (008) 155 170 Contents lists available at ScienceDirect Computers and Mathematics with Applications journal homepage: www.elsevier.com/locate/camwa Superlinear

More information

Fraction-free Row Reduction of Matrices of Skew Polynomials

Fraction-free Row Reduction of Matrices of Skew Polynomials Fraction-free Row Reduction of Matrices of Skew Polynomials Bernhard Beckermann Laboratoire d Analyse Numérique et d Optimisation Université des Sciences et Technologies de Lille France bbecker@ano.univ-lille1.fr

More information

STABILITY OF INVARIANT SUBSPACES OF COMMUTING MATRICES We obtain some further results for pairs of commuting matrices. We show that a pair of commutin

STABILITY OF INVARIANT SUBSPACES OF COMMUTING MATRICES We obtain some further results for pairs of commuting matrices. We show that a pair of commutin On the stability of invariant subspaces of commuting matrices Tomaz Kosir and Bor Plestenjak September 18, 001 Abstract We study the stability of (joint) invariant subspaces of a nite set of commuting

More information

Ole Christensen 3. October 20, Abstract. We point out some connections between the existing theories for

Ole Christensen 3. October 20, Abstract. We point out some connections between the existing theories for Frames and pseudo-inverses. Ole Christensen 3 October 20, 1994 Abstract We point out some connections between the existing theories for frames and pseudo-inverses. In particular, using the pseudo-inverse

More information

Vector Space Basics. 1 Abstract Vector Spaces. 1. (commutativity of vector addition) u + v = v + u. 2. (associativity of vector addition)

Vector Space Basics. 1 Abstract Vector Spaces. 1. (commutativity of vector addition) u + v = v + u. 2. (associativity of vector addition) Vector Space Basics (Remark: these notes are highly formal and may be a useful reference to some students however I am also posting Ray Heitmann's notes to Canvas for students interested in a direct computational

More information

Numerical solution of the eigenvalue problem for Hermitian Toeplitz-like matrices

Numerical solution of the eigenvalue problem for Hermitian Toeplitz-like matrices TR-CS-9-1 Numerical solution of the eigenvalue problem for Hermitian Toeplitz-like matrices Michael K Ng and William F Trench July 199 Joint Computer Science Technical Report Series Department of Computer

More information

4.1 Eigenvalues, Eigenvectors, and The Characteristic Polynomial

4.1 Eigenvalues, Eigenvectors, and The Characteristic Polynomial Linear Algebra (part 4): Eigenvalues, Diagonalization, and the Jordan Form (by Evan Dummit, 27, v ) Contents 4 Eigenvalues, Diagonalization, and the Jordan Canonical Form 4 Eigenvalues, Eigenvectors, and

More information

Constrained Leja points and the numerical solution of the constrained energy problem

Constrained Leja points and the numerical solution of the constrained energy problem Journal of Computational and Applied Mathematics 131 (2001) 427 444 www.elsevier.nl/locate/cam Constrained Leja points and the numerical solution of the constrained energy problem Dan I. Coroian, Peter

More information

SIAM Journal on Scientific Computing, 1999, v. 21 n. 3, p

SIAM Journal on Scientific Computing, 1999, v. 21 n. 3, p Title A Fast Algorithm for Deblurring Models with Neumann Boundary Conditions Author(s) Ng, MKP; Chan, RH; Tang, WC Citation SIAM Journal on Scientific Computing, 1999, v 21 n 3, p 851-866 Issued Date

More information

Linear Algebra (part 1) : Vector Spaces (by Evan Dummit, 2017, v. 1.07) 1.1 The Formal Denition of a Vector Space

Linear Algebra (part 1) : Vector Spaces (by Evan Dummit, 2017, v. 1.07) 1.1 The Formal Denition of a Vector Space Linear Algebra (part 1) : Vector Spaces (by Evan Dummit, 2017, v. 1.07) Contents 1 Vector Spaces 1 1.1 The Formal Denition of a Vector Space.................................. 1 1.2 Subspaces...................................................

More information

October 7, :8 WSPC/WS-IJWMIP paper. Polynomial functions are renable

October 7, :8 WSPC/WS-IJWMIP paper. Polynomial functions are renable International Journal of Wavelets, Multiresolution and Information Processing c World Scientic Publishing Company Polynomial functions are renable Henning Thielemann Institut für Informatik Martin-Luther-Universität

More information

APPENDIX A. Background Mathematics. A.1 Linear Algebra. Vector algebra. Let x denote the n-dimensional column vector with components x 1 x 2.

APPENDIX A. Background Mathematics. A.1 Linear Algebra. Vector algebra. Let x denote the n-dimensional column vector with components x 1 x 2. APPENDIX A Background Mathematics A. Linear Algebra A.. Vector algebra Let x denote the n-dimensional column vector with components 0 x x 2 B C @. A x n Definition 6 (scalar product). The scalar product

More information

PROOF OF TWO MATRIX THEOREMS VIA TRIANGULAR FACTORIZATIONS ROY MATHIAS

PROOF OF TWO MATRIX THEOREMS VIA TRIANGULAR FACTORIZATIONS ROY MATHIAS PROOF OF TWO MATRIX THEOREMS VIA TRIANGULAR FACTORIZATIONS ROY MATHIAS Abstract. We present elementary proofs of the Cauchy-Binet Theorem on determinants and of the fact that the eigenvalues of a matrix

More information

2 Garrett: `A Good Spectral Theorem' 1. von Neumann algebras, density theorem The commutant of a subring S of a ring R is S 0 = fr 2 R : rs = sr; 8s 2

2 Garrett: `A Good Spectral Theorem' 1. von Neumann algebras, density theorem The commutant of a subring S of a ring R is S 0 = fr 2 R : rs = sr; 8s 2 1 A Good Spectral Theorem c1996, Paul Garrett, garrett@math.umn.edu version February 12, 1996 1 Measurable Hilbert bundles Measurable Banach bundles Direct integrals of Hilbert spaces Trivializing Hilbert

More information

On GMW designs and a conjecture of Assmus and Key Thomas E. Norwood and Qing Xiang Dept. of Mathematics, California Institute of Technology, Pasadena,

On GMW designs and a conjecture of Assmus and Key Thomas E. Norwood and Qing Xiang Dept. of Mathematics, California Institute of Technology, Pasadena, On GMW designs and a conjecture of Assmus and Key Thomas E. Norwood and Qing iang Dept. of Mathematics, California Institute of Technology, Pasadena, CA 91125 June 24, 1998 Abstract We show that a family

More information

On the Local Convergence of an Iterative Approach for Inverse Singular Value Problems

On the Local Convergence of an Iterative Approach for Inverse Singular Value Problems On the Local Convergence of an Iterative Approach for Inverse Singular Value Problems Zheng-jian Bai Benedetta Morini Shu-fang Xu Abstract The purpose of this paper is to provide the convergence theory

More information

The subject of this paper is nding small sample spaces for joint distributions of

The subject of this paper is nding small sample spaces for joint distributions of Constructing Small Sample Spaces for De-Randomization of Algorithms Daphne Koller Nimrod Megiddo y September 1993 The subject of this paper is nding small sample spaces for joint distributions of n Bernoulli

More information

Institute for Advanced Computer Studies. Department of Computer Science. Iterative methods for solving Ax = b. GMRES/FOM versus QMR/BiCG

Institute for Advanced Computer Studies. Department of Computer Science. Iterative methods for solving Ax = b. GMRES/FOM versus QMR/BiCG University of Maryland Institute for Advanced Computer Studies Department of Computer Science College Park TR{96{2 TR{3587 Iterative methods for solving Ax = b GMRES/FOM versus QMR/BiCG Jane K. Cullum

More information

On the decay of elements of inverse triangular Toeplitz matrix

On the decay of elements of inverse triangular Toeplitz matrix On the decay of elements of inverse triangular Toeplitz matrix Neville Ford, D. V. Savostyanov, N. L. Zamarashkin August 03, 203 arxiv:308.0724v [math.na] 3 Aug 203 Abstract We consider half-infinite triangular

More information

j=1 u 1jv 1j. 1/ 2 Lemma 1. An orthogonal set of vectors must be linearly independent.

j=1 u 1jv 1j. 1/ 2 Lemma 1. An orthogonal set of vectors must be linearly independent. Lecture Notes: Orthogonal and Symmetric Matrices Yufei Tao Department of Computer Science and Engineering Chinese University of Hong Kong taoyf@cse.cuhk.edu.hk Orthogonal Matrix Definition. Let u = [u

More information

A MULTIGRID ALGORITHM FOR. Richard E. Ewing and Jian Shen. Institute for Scientic Computation. Texas A&M University. College Station, Texas SUMMARY

A MULTIGRID ALGORITHM FOR. Richard E. Ewing and Jian Shen. Institute for Scientic Computation. Texas A&M University. College Station, Texas SUMMARY A MULTIGRID ALGORITHM FOR THE CELL-CENTERED FINITE DIFFERENCE SCHEME Richard E. Ewing and Jian Shen Institute for Scientic Computation Texas A&M University College Station, Texas SUMMARY In this article,

More information

Congurations of periodic orbits for equations with delayed positive feedback

Congurations of periodic orbits for equations with delayed positive feedback Congurations of periodic orbits for equations with delayed positive feedback Dedicated to Professor Tibor Krisztin on the occasion of his 60th birthday Gabriella Vas 1 MTA-SZTE Analysis and Stochastics

More information

Preconditioning for Nonsymmetry and Time-dependence

Preconditioning for Nonsymmetry and Time-dependence Preconditioning for Nonsymmetry and Time-dependence Andy Wathen Oxford University, UK joint work with Jen Pestana and Elle McDonald Jeju, Korea, 2015 p.1/24 Iterative methods For self-adjoint problems/symmetric

More information

An exploration of matrix equilibration

An exploration of matrix equilibration An exploration of matrix equilibration Paul Liu Abstract We review three algorithms that scale the innity-norm of each row and column in a matrix to. The rst algorithm applies to unsymmetric matrices,

More information

linearly indepedent eigenvectors as the multiplicity of the root, but in general there may be no more than one. For further discussion, assume matrice

linearly indepedent eigenvectors as the multiplicity of the root, but in general there may be no more than one. For further discussion, assume matrice 3. Eigenvalues and Eigenvectors, Spectral Representation 3.. Eigenvalues and Eigenvectors A vector ' is eigenvector of a matrix K, if K' is parallel to ' and ' 6, i.e., K' k' k is the eigenvalue. If is

More information

2 W. LAWTON, S. L. LEE AND ZUOWEI SHEN is called the fundamental condition, and a sequence which satises the fundamental condition will be called a fu

2 W. LAWTON, S. L. LEE AND ZUOWEI SHEN is called the fundamental condition, and a sequence which satises the fundamental condition will be called a fu CONVERGENCE OF MULTIDIMENSIONAL CASCADE ALGORITHM W. LAWTON, S. L. LEE AND ZUOWEI SHEN Abstract. Necessary and sucient conditions on the spectrum of the restricted transition operators are given for the

More information

Approximation Algorithms for Maximum. Coverage and Max Cut with Given Sizes of. Parts? A. A. Ageev and M. I. Sviridenko

Approximation Algorithms for Maximum. Coverage and Max Cut with Given Sizes of. Parts? A. A. Ageev and M. I. Sviridenko Approximation Algorithms for Maximum Coverage and Max Cut with Given Sizes of Parts? A. A. Ageev and M. I. Sviridenko Sobolev Institute of Mathematics pr. Koptyuga 4, 630090, Novosibirsk, Russia fageev,svirg@math.nsc.ru

More information

1 Linear Algebra Problems

1 Linear Algebra Problems Linear Algebra Problems. Let A be the conjugate transpose of the complex matrix A; i.e., A = A t : A is said to be Hermitian if A = A; real symmetric if A is real and A t = A; skew-hermitian if A = A and

More information

The Solvability Conditions for the Inverse Eigenvalue Problem of Hermitian and Generalized Skew-Hamiltonian Matrices and Its Approximation

The Solvability Conditions for the Inverse Eigenvalue Problem of Hermitian and Generalized Skew-Hamiltonian Matrices and Its Approximation The Solvability Conditions for the Inverse Eigenvalue Problem of Hermitian and Generalized Skew-Hamiltonian Matrices and Its Approximation Zheng-jian Bai Abstract In this paper, we first consider the inverse

More information

Toeplitz matrices: spectral properties and preconditioning in the CG method

Toeplitz matrices: spectral properties and preconditioning in the CG method Toeplitz matrices: spectral properties and preconditioning in the CG method Stefano Serra-Capizzano December 5, 2014 Abstract We consider multilevel Toeplitz matrices T n (f) generated by Lebesgue integrable

More information

A NOTE ON MATRIX REFINEMENT EQUATIONS. Abstract. Renement equations involving matrix masks are receiving a lot of attention these days.

A NOTE ON MATRIX REFINEMENT EQUATIONS. Abstract. Renement equations involving matrix masks are receiving a lot of attention these days. A NOTE ON MATRI REFINEMENT EQUATIONS THOMAS A. HOGAN y Abstract. Renement equations involving matrix masks are receiving a lot of attention these days. They can play a central role in the study of renable

More information

THE PERTURBATION BOUND FOR THE SPECTRAL RADIUS OF A NON-NEGATIVE TENSOR

THE PERTURBATION BOUND FOR THE SPECTRAL RADIUS OF A NON-NEGATIVE TENSOR THE PERTURBATION BOUND FOR THE SPECTRAL RADIUS OF A NON-NEGATIVE TENSOR WEN LI AND MICHAEL K. NG Abstract. In this paper, we study the perturbation bound for the spectral radius of an m th - order n-dimensional

More information

Mathematics Department Stanford University Math 61CM/DM Inner products

Mathematics Department Stanford University Math 61CM/DM Inner products Mathematics Department Stanford University Math 61CM/DM Inner products Recall the definition of an inner product space; see Appendix A.8 of the textbook. Definition 1 An inner product space V is a vector

More information

Moment Computation in Shift Invariant Spaces. Abstract. An algorithm is given for the computation of moments of f 2 S, where S is either

Moment Computation in Shift Invariant Spaces. Abstract. An algorithm is given for the computation of moments of f 2 S, where S is either Moment Computation in Shift Invariant Spaces David A. Eubanks Patrick J.Van Fleet y Jianzhong Wang ẓ Abstract An algorithm is given for the computation of moments of f 2 S, where S is either a principal

More information

CS137 Introduction to Scientific Computing Winter Quarter 2004 Solutions to Homework #3

CS137 Introduction to Scientific Computing Winter Quarter 2004 Solutions to Homework #3 CS137 Introduction to Scientific Computing Winter Quarter 2004 Solutions to Homework #3 Felix Kwok February 27, 2004 Written Problems 1. (Heath E3.10) Let B be an n n matrix, and assume that B is both

More information

Linear Algebra, 4th day, Thursday 7/1/04 REU Info:

Linear Algebra, 4th day, Thursday 7/1/04 REU Info: Linear Algebra, 4th day, Thursday 7/1/04 REU 004. Info http//people.cs.uchicago.edu/laci/reu04. Instructor Laszlo Babai Scribe Nick Gurski 1 Linear maps We shall study the notion of maps between vector

More information

An idea how to solve some of the problems. diverges the same must hold for the original series. T 1 p T 1 p + 1 p 1 = 1. dt = lim

An idea how to solve some of the problems. diverges the same must hold for the original series. T 1 p T 1 p + 1 p 1 = 1. dt = lim An idea how to solve some of the problems 5.2-2. (a) Does not converge: By multiplying across we get Hence 2k 2k 2 /2 k 2k2 k 2 /2 k 2 /2 2k 2k 2 /2 k. As the series diverges the same must hold for the

More information

290 J.M. Carnicer, J.M. Pe~na basis (u 1 ; : : : ; u n ) consisting of minimally supported elements, yet also has a basis (v 1 ; : : : ; v n ) which f

290 J.M. Carnicer, J.M. Pe~na basis (u 1 ; : : : ; u n ) consisting of minimally supported elements, yet also has a basis (v 1 ; : : : ; v n ) which f Numer. Math. 67: 289{301 (1994) Numerische Mathematik c Springer-Verlag 1994 Electronic Edition Least supported bases and local linear independence J.M. Carnicer, J.M. Pe~na? Departamento de Matematica

More information

r( = f 2 L 2 (1.2) iku)! 0 as r!1: (1.3) It was shown in book [7] that if f is assumed to be the restriction of a function in C

r(  = f 2 L 2 (1.2) iku)! 0 as r!1: (1.3) It was shown in book [7] that if f is assumed to be the restriction of a function in C Inverse Obstacle Problem: Local Uniqueness for Rougher Obstacles and the Identication of A Ball Changmei Liu Department of Mathematics University of North Carolina Chapel Hill, NC 27599 December, 1995

More information

Finite-dimensional spaces. C n is the space of n-tuples x = (x 1,..., x n ) of complex numbers. It is a Hilbert space with the inner product

Finite-dimensional spaces. C n is the space of n-tuples x = (x 1,..., x n ) of complex numbers. It is a Hilbert space with the inner product Chapter 4 Hilbert Spaces 4.1 Inner Product Spaces Inner Product Space. A complex vector space E is called an inner product space (or a pre-hilbert space, or a unitary space) if there is a mapping (, )

More information

Stability, Queue Length and Delay of Deterministic and Stochastic Queueing Networks Cheng-Shang Chang IBM Research Division T.J. Watson Research Cente

Stability, Queue Length and Delay of Deterministic and Stochastic Queueing Networks Cheng-Shang Chang IBM Research Division T.J. Watson Research Cente Stability, Queue Length and Delay of Deterministic and Stochastic Queueing Networks Cheng-Shang Chang IBM Research Division T.J. Watson Research Center P.O. Box 704 Yorktown Heights, NY 10598 cschang@watson.ibm.com

More information

Polynomial functions over nite commutative rings

Polynomial functions over nite commutative rings Polynomial functions over nite commutative rings Balázs Bulyovszky a, Gábor Horváth a, a Institute of Mathematics, University of Debrecen, Pf. 400, Debrecen, 4002, Hungary Abstract We prove a necessary

More information

Continuity. Chapter 4

Continuity. Chapter 4 Chapter 4 Continuity Throughout this chapter D is a nonempty subset of the real numbers. We recall the definition of a function. Definition 4.1. A function from D into R, denoted f : D R, is a subset of

More information

Notes on the matrix exponential

Notes on the matrix exponential Notes on the matrix exponential Erik Wahlén erik.wahlen@math.lu.se February 14, 212 1 Introduction The purpose of these notes is to describe how one can compute the matrix exponential e A when A is not

More information

On the Preconditioning of the Block Tridiagonal Linear System of Equations

On the Preconditioning of the Block Tridiagonal Linear System of Equations On the Preconditioning of the Block Tridiagonal Linear System of Equations Davod Khojasteh Salkuyeh Department of Mathematics, University of Mohaghegh Ardabili, PO Box 179, Ardabil, Iran E-mail: khojaste@umaacir

More information

IMC 2015, Blagoevgrad, Bulgaria

IMC 2015, Blagoevgrad, Bulgaria IMC 05, Blagoevgrad, Bulgaria Day, July 9, 05 Problem. For any integer n and two n n matrices with real entries, B that satisfy the equation + B ( + B prove that det( det(b. Does the same conclusion follow

More information

Factoring Polynomials with Rational Coecients. Kenneth Giuliani

Factoring Polynomials with Rational Coecients. Kenneth Giuliani Factoring Polynomials with Rational Coecients Kenneth Giuliani 17 April 1998 1 Introduction Factorization is a problem well-studied in mathematics. Of particular focus is factorization within unique factorization

More information

Coins with arbitrary weights. Abstract. Given a set of m coins out of a collection of coins of k unknown distinct weights, we wish to

Coins with arbitrary weights. Abstract. Given a set of m coins out of a collection of coins of k unknown distinct weights, we wish to Coins with arbitrary weights Noga Alon Dmitry N. Kozlov y Abstract Given a set of m coins out of a collection of coins of k unknown distinct weights, we wish to decide if all the m given coins have the

More information

DEN: Linear algebra numerical view (GEM: Gauss elimination method for reducing a full rank matrix to upper-triangular

DEN: Linear algebra numerical view (GEM: Gauss elimination method for reducing a full rank matrix to upper-triangular form) Given: matrix C = (c i,j ) n,m i,j=1 ODE and num math: Linear algebra (N) [lectures] c phabala 2016 DEN: Linear algebra numerical view (GEM: Gauss elimination method for reducing a full rank matrix

More information

In particular, if A is a square matrix and λ is one of its eigenvalues, then we can find a non-zero column vector X with

In particular, if A is a square matrix and λ is one of its eigenvalues, then we can find a non-zero column vector X with Appendix: Matrix Estimates and the Perron-Frobenius Theorem. This Appendix will first present some well known estimates. For any m n matrix A = [a ij ] over the real or complex numbers, it will be convenient

More information

The following can also be obtained from this WWW address: the papers [8, 9], more examples, comments on the implementation and a short description of

The following can also be obtained from this WWW address: the papers [8, 9], more examples, comments on the implementation and a short description of An algorithm for computing the Weierstrass normal form Mark van Hoeij Department of mathematics University of Nijmegen 6525 ED Nijmegen The Netherlands e-mail: hoeij@sci.kun.nl April 9, 1995 Abstract This

More information

Back circulant Latin squares and the inuence of a set. L F Fitina, Jennifer Seberry and Ghulam R Chaudhry. Centre for Computer Security Research

Back circulant Latin squares and the inuence of a set. L F Fitina, Jennifer Seberry and Ghulam R Chaudhry. Centre for Computer Security Research Back circulant Latin squares and the inuence of a set L F Fitina, Jennifer Seberry and Ghulam R Chaudhry Centre for Computer Security Research School of Information Technology and Computer Science University

More information

Jurgen Garlo. the inequality sign in all components having odd index sum. For these intervals in

Jurgen Garlo. the inequality sign in all components having odd index sum. For these intervals in Intervals of Almost Totally Positive Matrices Jurgen Garlo University of Applied Sciences / FH Konstanz, Fachbereich Informatik, Postfach 100543, D-78405 Konstanz, Germany Abstract We consider the class

More information

Linear Algebra: Characteristic Value Problem

Linear Algebra: Characteristic Value Problem Linear Algebra: Characteristic Value Problem . The Characteristic Value Problem Let < be the set of real numbers and { be the set of complex numbers. Given an n n real matrix A; does there exist a number

More information

The average dimension of the hull of cyclic codes

The average dimension of the hull of cyclic codes Discrete Applied Mathematics 128 (2003) 275 292 www.elsevier.com/locate/dam The average dimension of the hull of cyclic codes Gintaras Skersys Matematikos ir Informatikos Fakultetas, Vilniaus Universitetas,

More information

Analysis on Graphs. Alexander Grigoryan Lecture Notes. University of Bielefeld, WS 2011/12

Analysis on Graphs. Alexander Grigoryan Lecture Notes. University of Bielefeld, WS 2011/12 Analysis on Graphs Alexander Grigoryan Lecture Notes University of Bielefeld, WS 0/ Contents The Laplace operator on graphs 5. The notion of a graph............................. 5. Cayley graphs..................................

More information

Linear estimation in models based on a graph

Linear estimation in models based on a graph Linear Algebra and its Applications 302±303 (1999) 223±230 www.elsevier.com/locate/laa Linear estimation in models based on a graph R.B. Bapat * Indian Statistical Institute, New Delhi 110 016, India Received

More information

Waveform Relaxation Method with Toeplitz. Operator Splitting. Sigitas Keras. August Department of Applied Mathematics and Theoretical Physics

Waveform Relaxation Method with Toeplitz. Operator Splitting. Sigitas Keras. August Department of Applied Mathematics and Theoretical Physics UNIVERSITY OF CAMBRIDGE Numerical Analysis Reports Waveform Relaxation Method with Toeplitz Operator Splitting Sigitas Keras DAMTP 1995/NA4 August 1995 Department of Applied Mathematics and Theoretical

More information

QUASI-UNIFORMLY POSITIVE OPERATORS IN KREIN SPACE. Denitizable operators in Krein spaces have spectral properties similar to those

QUASI-UNIFORMLY POSITIVE OPERATORS IN KREIN SPACE. Denitizable operators in Krein spaces have spectral properties similar to those QUASI-UNIFORMLY POSITIVE OPERATORS IN KREIN SPACE BRANKO CURGUS and BRANKO NAJMAN Denitizable operators in Krein spaces have spectral properties similar to those of selfadjoint operators in Hilbert spaces.

More information

On Riesz-Fischer sequences and lower frame bounds

On Riesz-Fischer sequences and lower frame bounds On Riesz-Fischer sequences and lower frame bounds P. Casazza, O. Christensen, S. Li, A. Lindner Abstract We investigate the consequences of the lower frame condition and the lower Riesz basis condition

More information

Numerical Integration for Multivariable. October Abstract. We consider the numerical integration of functions with point singularities over

Numerical Integration for Multivariable. October Abstract. We consider the numerical integration of functions with point singularities over Numerical Integration for Multivariable Functions with Point Singularities Yaun Yang and Kendall E. Atkinson y October 199 Abstract We consider the numerical integration of functions with point singularities

More information

Real Analysis Math 131AH Rudin, Chapter #1. Dominique Abdi

Real Analysis Math 131AH Rudin, Chapter #1. Dominique Abdi Real Analysis Math 3AH Rudin, Chapter # Dominique Abdi.. If r is rational (r 0) and x is irrational, prove that r + x and rx are irrational. Solution. Assume the contrary, that r+x and rx are rational.

More information

Fast Angular Synchronization for Phase Retrieval via Incomplete Information

Fast Angular Synchronization for Phase Retrieval via Incomplete Information Fast Angular Synchronization for Phase Retrieval via Incomplete Information Aditya Viswanathan a and Mark Iwen b a Department of Mathematics, Michigan State University; b Department of Mathematics & Department

More information

Markov Sonin Gaussian rule for singular functions

Markov Sonin Gaussian rule for singular functions Journal of Computational and Applied Mathematics 169 (2004) 197 212 www.elsevier.com/locate/cam Markov Sonin Gaussian rule for singular functions G.Mastroianni, D.Occorsio Dipartimento di Matematica, Universita

More information

Using q-calculus to study LDL T factorization of a certain Vandermonde matrix

Using q-calculus to study LDL T factorization of a certain Vandermonde matrix Using q-calculus to study LDL T factorization of a certain Vandermonde matrix Alexey Kuznetsov May 2, 207 Abstract We use tools from q-calculus to study LDL T decomposition of the Vandermonde matrix V

More information

Jordan Journal of Mathematics and Statistics (JJMS) 5(3), 2012, pp A NEW ITERATIVE METHOD FOR SOLVING LINEAR SYSTEMS OF EQUATIONS

Jordan Journal of Mathematics and Statistics (JJMS) 5(3), 2012, pp A NEW ITERATIVE METHOD FOR SOLVING LINEAR SYSTEMS OF EQUATIONS Jordan Journal of Mathematics and Statistics JJMS) 53), 2012, pp.169-184 A NEW ITERATIVE METHOD FOR SOLVING LINEAR SYSTEMS OF EQUATIONS ADEL H. AL-RABTAH Abstract. The Jacobi and Gauss-Seidel iterative

More information

A Method for Constructing Diagonally Dominant Preconditioners based on Jacobi Rotations

A Method for Constructing Diagonally Dominant Preconditioners based on Jacobi Rotations A Method for Constructing Diagonally Dominant Preconditioners based on Jacobi Rotations Jin Yun Yuan Plamen Y. Yalamov Abstract A method is presented to make a given matrix strictly diagonally dominant

More information

Means of unitaries, conjugations, and the Friedrichs operator

Means of unitaries, conjugations, and the Friedrichs operator J. Math. Anal. Appl. 335 (2007) 941 947 www.elsevier.com/locate/jmaa Means of unitaries, conjugations, and the Friedrichs operator Stephan Ramon Garcia Department of Mathematics, Pomona College, Claremont,

More information

K. NAGY Now, introduce an orthonormal system on G m called Vilenkin-like system (see [G]). The complex valued functions rk n G m! C are called general

K. NAGY Now, introduce an orthonormal system on G m called Vilenkin-like system (see [G]). The complex valued functions rk n G m! C are called general A HARDY{LITTLEWOOD{LIKE INEQUALITY ON TWO{DIMENSIONAL COMPACT TOTALLY DISCONNECTED SPACES K. Nagy Abstract. We prove a Hardy-Littlewood type ineuality with respect to a system called Vilenkin-like system

More information

arxiv: v1 [math.na] 1 Sep 2018

arxiv: v1 [math.na] 1 Sep 2018 On the perturbation of an L -orthogonal projection Xuefeng Xu arxiv:18090000v1 [mathna] 1 Sep 018 September 5 018 Abstract The L -orthogonal projection is an important mathematical tool in scientific computing

More information

Detailed Proof of The PerronFrobenius Theorem

Detailed Proof of The PerronFrobenius Theorem Detailed Proof of The PerronFrobenius Theorem Arseny M Shur Ural Federal University October 30, 2016 1 Introduction This famous theorem has numerous applications, but to apply it you should understand

More information

9.1 Preconditioned Krylov Subspace Methods

9.1 Preconditioned Krylov Subspace Methods Chapter 9 PRECONDITIONING 9.1 Preconditioned Krylov Subspace Methods 9.2 Preconditioned Conjugate Gradient 9.3 Preconditioned Generalized Minimal Residual 9.4 Relaxation Method Preconditioners 9.5 Incomplete

More information

1. Introduction Let the least value of an objective function F (x), x2r n, be required, where F (x) can be calculated for any vector of variables x2r

1. Introduction Let the least value of an objective function F (x), x2r n, be required, where F (x) can be calculated for any vector of variables x2r DAMTP 2002/NA08 Least Frobenius norm updating of quadratic models that satisfy interpolation conditions 1 M.J.D. Powell Abstract: Quadratic models of objective functions are highly useful in many optimization

More information

. Consider the linear system dx= =! = " a b # x y! : (a) For what values of a and b do solutions oscillate (i.e., do both x(t) and y(t) pass through z

. Consider the linear system dx= =! =  a b # x y! : (a) For what values of a and b do solutions oscillate (i.e., do both x(t) and y(t) pass through z Preliminary Exam { 1999 Morning Part Instructions: No calculators or crib sheets are allowed. Do as many problems as you can. Justify your answers as much as you can but very briey. 1. For positive real

More information

SPRING 2006 PRELIMINARY EXAMINATION SOLUTIONS

SPRING 2006 PRELIMINARY EXAMINATION SOLUTIONS SPRING 006 PRELIMINARY EXAMINATION SOLUTIONS 1A. Let G be the subgroup of the free abelian group Z 4 consisting of all integer vectors (x, y, z, w) such that x + 3y + 5z + 7w = 0. (a) Determine a linearly

More information

Maximizing the numerical radii of matrices by permuting their entries

Maximizing the numerical radii of matrices by permuting their entries Maximizing the numerical radii of matrices by permuting their entries Wai-Shun Cheung and Chi-Kwong Li Dedicated to Professor Pei Yuan Wu. Abstract Let A be an n n complex matrix such that every row and

More information

A Note on Spectra of Optimal and Superoptimal Preconditioned Matrices

A Note on Spectra of Optimal and Superoptimal Preconditioned Matrices A Note on Spectra of Optimal and Superoptimal Preconditioned Matrices Che-ManCHENG Xiao-QingJIN Seak-WengVONG WeiWANG March 2006 Revised, November 2006 Abstract In this short note, it is proved that given

More information

The restarted QR-algorithm for eigenvalue computation of structured matrices

The restarted QR-algorithm for eigenvalue computation of structured matrices Journal of Computational and Applied Mathematics 149 (2002) 415 422 www.elsevier.com/locate/cam The restarted QR-algorithm for eigenvalue computation of structured matrices Daniela Calvetti a; 1, Sun-Mi

More information

2 Section 2 However, in order to apply the above idea, we will need to allow non standard intervals ('; ) in the proof. More precisely, ' and may gene

2 Section 2 However, in order to apply the above idea, we will need to allow non standard intervals ('; ) in the proof. More precisely, ' and may gene Introduction 1 A dierential intermediate value theorem by Joris van der Hoeven D pt. de Math matiques (B t. 425) Universit Paris-Sud 91405 Orsay Cedex France June 2000 Abstract Let T be the eld of grid-based

More information

Convexity and Star-shapedness of Real Linear Images of Special Orthogonal Orbits

Convexity and Star-shapedness of Real Linear Images of Special Orthogonal Orbits Convexity and Star-shapedness of Real Linear Images of Special Orthogonal Orbits Pan-Shun Lau 1, Tuen-Wai Ng 1, and Nam-Kiu Tsing 1 1 Department of Mathematics, The University of Hong Kong, Pokfulam, Hong

More information

In [7], Vogel introduced the \lagged diusivity xed point iteration", which we denote by FP, to solve the system (6). If A u k, H and L u k denote resp

In [7], Vogel introduced the \lagged diusivity xed point iteration, which we denote by FP, to solve the system (6). If A u k, H and L u k denote resp Cosine Transform Based Preconditioners for Total Variation Deblurring Raymond H. Chan, Tony F. Chan, Chiu-Kwong Wong Abstract Image reconstruction is a mathematically illposed problem and regularization

More information

AIMS Exercise Set # 1

AIMS Exercise Set # 1 AIMS Exercise Set #. Determine the form of the single precision floating point arithmetic used in the computers at AIMS. What is the largest number that can be accurately represented? What is the smallest

More information

ON THE HÖLDER CONTINUITY OF MATRIX FUNCTIONS FOR NORMAL MATRICES

ON THE HÖLDER CONTINUITY OF MATRIX FUNCTIONS FOR NORMAL MATRICES Volume 10 (2009), Issue 4, Article 91, 5 pp. ON THE HÖLDER CONTINUITY O MATRIX UNCTIONS OR NORMAL MATRICES THOMAS P. WIHLER MATHEMATICS INSTITUTE UNIVERSITY O BERN SIDLERSTRASSE 5, CH-3012 BERN SWITZERLAND.

More information

DS-GA 1002 Lecture notes 0 Fall Linear Algebra. These notes provide a review of basic concepts in linear algebra.

DS-GA 1002 Lecture notes 0 Fall Linear Algebra. These notes provide a review of basic concepts in linear algebra. DS-GA 1002 Lecture notes 0 Fall 2016 Linear Algebra These notes provide a review of basic concepts in linear algebra. 1 Vector spaces You are no doubt familiar with vectors in R 2 or R 3, i.e. [ ] 1.1

More information

Lecture 8 : Eigenvalues and Eigenvectors

Lecture 8 : Eigenvalues and Eigenvectors CPS290: Algorithmic Foundations of Data Science February 24, 2017 Lecture 8 : Eigenvalues and Eigenvectors Lecturer: Kamesh Munagala Scribe: Kamesh Munagala Hermitian Matrices It is simpler to begin with

More information

Rearrangements and polar factorisation of countably degenerate functions G.R. Burton, School of Mathematical Sciences, University of Bath, Claverton D

Rearrangements and polar factorisation of countably degenerate functions G.R. Burton, School of Mathematical Sciences, University of Bath, Claverton D Rearrangements and polar factorisation of countably degenerate functions G.R. Burton, School of Mathematical Sciences, University of Bath, Claverton Down, Bath BA2 7AY, U.K. R.J. Douglas, Isaac Newton

More information

Numerical tensor methods and their applications

Numerical tensor methods and their applications Numerical tensor methods and their applications 8 May 2013 All lectures 4 lectures, 2 May, 08:00-10:00: Introduction: ideas, matrix results, history. 7 May, 08:00-10:00: Novel tensor formats (TT, HT, QTT).

More information

Jae Heon Yun and Yu Du Han

Jae Heon Yun and Yu Du Han Bull. Korean Math. Soc. 39 (2002), No. 3, pp. 495 509 MODIFIED INCOMPLETE CHOLESKY FACTORIZATION PRECONDITIONERS FOR A SYMMETRIC POSITIVE DEFINITE MATRIX Jae Heon Yun and Yu Du Han Abstract. We propose

More information

Stanford Mathematics Department Math 205A Lecture Supplement #4 Borel Regular & Radon Measures

Stanford Mathematics Department Math 205A Lecture Supplement #4 Borel Regular & Radon Measures 2 1 Borel Regular Measures We now state and prove an important regularity property of Borel regular outer measures: Stanford Mathematics Department Math 205A Lecture Supplement #4 Borel Regular & Radon

More information

Iyad T. Abu-Jeib and Thomas S. Shores

Iyad T. Abu-Jeib and Thomas S. Shores NEW ZEALAND JOURNAL OF MATHEMATICS Volume 3 (003, 9 ON PROPERTIES OF MATRIX I ( OF SINC METHODS Iyad T. Abu-Jeib and Thomas S. Shores (Received February 00 Abstract. In this paper, we study the determinant

More information

Minimum and maximum values *

Minimum and maximum values * OpenStax-CNX module: m17417 1 Minimum and maximum values * Sunil Kumar Singh This work is produced by OpenStax-CNX and licensed under the Creative Commons Attribution License 2.0 In general context, a

More information