Research article

Sturm's Theorem for Min matrices

  • Received: 17 February 2023 Revised: 29 April 2023 Accepted: 11 May 2023 Published: 18 May 2023
  • MSC : 15A18, 15B99

  • In the present paper, we study Min matrix Amin=[amin(i,j)]ni,j=1, where as's are the elements of a real sequence {as}. We first obtain a recurrence relation for the characteristic polynomial for matrix Amin, and some relations between the coefficients of its characteristic polynomial. Next, we show that the sequence of the characteristic polynomials of the i×i(in) Min matrices satisfies the Sturm sequence properties according to different required conditions of the sequence {as}. Using Sturm's Theorem, we get some results about the eigenvalues, such as the number of eigenvalues in an interval. Thus, we obtain the number of positive and negative eigenvalues of Min matrix Amin. Finally, we give an example to illustrate our results.

    Citation: Efruz Özlem Mersin. Sturm's Theorem for Min matrices[J]. AIMS Mathematics, 2023, 8(7): 17229-17245. doi: 10.3934/math.2023880

    Related Papers:

    [1] Helena Myšková, Ján Plavka . Optimizing the max-min function with a constraint on a two-sided linear system. AIMS Mathematics, 2024, 9(4): 7791-7809. doi: 10.3934/math.2024378
    [2] Haifa Bin Jebreen, Beatriz Hernández-Jiménez . Pseudospectral method for fourth-order fractional Sturm-Liouville problems. AIMS Mathematics, 2024, 9(9): 26077-26091. doi: 10.3934/math.20241274
    [3] Zhongqian Wang, Xuejun Zhang, Mingliang Song . Three nonnegative solutions for Sturm-Liouville BVP and application to the complete Sturm-Liouville equations. AIMS Mathematics, 2023, 8(3): 6543-6558. doi: 10.3934/math.2023330
    [4] Ahmed M.A. El-Sayed, Eman M.A. Hamdallah, Hameda M. A. Alama . Multiple solutions of a Sturm-Liouville boundary value problem of nonlinear differential inclusion with nonlocal integral conditions. AIMS Mathematics, 2022, 7(6): 11150-11164. doi: 10.3934/math.2022624
    [5] Jinming Cai, Shuang Li, Kun Li . Matrix representations of Atkinson-type Sturm-Liouville problems with coupled eigenparameter-dependent conditions. AIMS Mathematics, 2024, 9(9): 25297-25318. doi: 10.3934/math.20241235
    [6] Ebrahim. A. Youness, Abd El-Monem. A. Megahed, Elsayed. E. Eladdad, Hanem. F. A. Madkour . Min-max differential game with partial differential equation. AIMS Mathematics, 2022, 7(8): 13777-13789. doi: 10.3934/math.2022759
    [7] Xiao Wang, D. D. Hai . On a class of one-dimensional superlinear semipositone $ (p, q) $ -Laplacian problem. AIMS Mathematics, 2023, 8(11): 25740-25753. doi: 10.3934/math.20231313
    [8] Abdeljabbar Ghanmi, Abdelhakim Sahbani . Existence results for $ p(x) $-biharmonic problems involving a singular and a Hardy type nonlinearities. AIMS Mathematics, 2023, 8(12): 29892-29909. doi: 10.3934/math.20231528
    [9] Emrah Polatlı . On some properties of a generalized min matrix. AIMS Mathematics, 2023, 8(11): 26199-26212. doi: 10.3934/math.20231336
    [10] Yinlan Chen, Yawen Lan . The best approximation problems between the least-squares solution manifolds of two matrix equations. AIMS Mathematics, 2024, 9(8): 20939-20955. doi: 10.3934/math.20241019
  • In the present paper, we study Min matrix Amin=[amin(i,j)]ni,j=1, where as's are the elements of a real sequence {as}. We first obtain a recurrence relation for the characteristic polynomial for matrix Amin, and some relations between the coefficients of its characteristic polynomial. Next, we show that the sequence of the characteristic polynomials of the i×i(in) Min matrices satisfies the Sturm sequence properties according to different required conditions of the sequence {as}. Using Sturm's Theorem, we get some results about the eigenvalues, such as the number of eigenvalues in an interval. Thus, we obtain the number of positive and negative eigenvalues of Min matrix Amin. Finally, we give an example to illustrate our results.



    Matrices have wide use in a variety of problems in mathematics and many other sciences, such as physics and engineering. Considering that the matrix representation of a particular problem can yield significant results, some concepts, such as eigenvalue, singular value, norm and determinant, are useful for these results. There are some special matrices that attract the attention of researchers; Min and Max matrices are such matrices. Min and Max matrices with minimum and maximum entries were first introduced by Pólya and Szegö [1] as

    Amin=[111112221233123n]andAmax=[123n223n333nnnnn], (1.1)

    respectively. These matrices are expressed as Amin=[min(i,j)]ni,j=1 and Amax=[max(i,j)]ni,j=1. Catalani [2] gave some relations between the principal minors of the matrix Amin and the Fibonacci numbers. Bhatia [3] showed that the matrix Amin is infinitely divisible, and [4] studied this and related matrices. Eigenvalues and inverse of the matrix C=[min{aib,ajb}]ni,j=1 were studied by Fonseca [5] for a>0 and ab. Moyé [6] studied the covariance matrix of Brownian motion, which appears to be a certain Min matrix. By the motivation of the Moyé's paper, Neudecker et al. [7] posed some problems on the determinant, inverse and positive definiteness of more general type of Min matrices with real number entries, then Chu et al. [8] answered these problems. The general forms of Min and Max matrices given in Eq (1.1) are

    Amin=[a1a1a1a1a1a2a2a2a1a2a3a3a1a2a3an]andAmax=[a1a2a3ana2a2a3ana3a3a3ananananan], (1.2)

    respectively [7,8,9]. These Min and Max matrices are expressed as Amin=[amin(i,j)]ni,j=1 and Amax=[amax(i,j)]ni,j=1, where as's are the elements of a real sequence {as}. The determinants of the matrices Amin and Amax are [9]

    det(Amin)=a1(a2a1)(a3a2)(anan1)

    and

    det(Amax)=(a1a2)(a2a3)(an1an)an.

    Bahşi and Solak [10,11,12] characterized the matrices Ak=[k+min(i,j)1]ni,j=1 and Bk=[k+max(i,j)1]ni,j=1 for kR, and studied some of their properties, such as the determinants, inverses and characteristic polynomials. The general form of Min matrix was called a nested symmetric matrix and some of its properties, such as the determinant, inverse, principle minors, LU and QR-decompositions were studied by Stuart [13]. Petroudi and Pirouz [14] defined the exponential form of Min matrix as A=[amin(i,j)1]ni,j=1, where a>1 is a positive real constant. They investigated some properties of this matrix, such as the determinant, inverse, Hadamard inverse and norm. Petroudi and Pirouz [15,16,17] examined the matrices Fmin=[Fmin(i,j)]ni,j=1, Fmax=[Fmax(i,j)]ni,j=1F=[Fmin(i,j)+1]ni,j=1 and eF=[eFmin(i,j)+1]ni,j=1, where Fn is the nth Fibonacci numbers, for some properties as mentioned above. Some relations between the general forms of Min, Max matrices and meet, join matrices were examined by Mattila and Haukkanen [9]. They used meet and join matrices as a tool to obtain their results. Petroudi and Pirouz [18] studied the particular symmetric matrix H=[Hmin(i,j)]ni,j=1, where Hn is the nth Harmonic number. The authors investigated its Hadamard exponential matrix, along with some of its properties. They also derived the Euclidean norms and some bounds for the spectral norms of these matrices. Kılıç and Arıkan [19] studied the matrices Amin, Amax, and their Hadamard inverses as the generalizations of Min and Max matrices. The authors obtained the LU-decompositions, inverse, Cholesky decompositions and LU-decompositions of the inverses of these matrices. The characteristic polynomials, determinants, inverses and Hadamard inverses of Max and Min matrices whose entries consist of the hyper-Fibonacci and hyper-Lucas numbers were examined in [20,21]. Kızılateş and Terzioğlu [22] defined the matrices r-Min, r-Max and their Hadamard inverses. They investigated some properties of these matrices, such as the determinant, inverse, norm and factorizations.

    It is well known that the eigenvalues of a matrix are equivalent to the roots of the matrix's characteristic polynomial. Since the eigenvalues give some important information about matrices, the problem of finding the zeros of a polynomial is important for many sciences. There are some iterative methods, such as Newton's formula [23], and some bounds, such as Cauchy's bound [24], for this need. Also, Descartes' rule of sign [24] and Budan Fourier Theorem [24] give the upper bounds for the number of zeros in an interval. These results do not give the exact number of zeros of a polynomial in an interval. Sturm's Theorem is a very useful tool for just this purpose for any polynomial without multiple zeros. Sturm's Theorem uses the number of sign changes of the consecutive members of the Sturm sequence to get the exact number of zeros in an interval. Sturm's Theorem has been known by means of Sturm's studies [25,26,27] first appeared in 1829. There are many versions and analogies of the Sturm sequence properties and Sturm's Theorem in the literature [28,29,30,31].

    Now, we give the Sturm analogy, which we use for this paper.

    Definition 1.1. [28] Let P0(x),P1(x),,Pn(x) be continuous functions on an interval (a,b) (with the possibilities a=, b=). If

    (1) P0(x) has no zeros in (a,b),

    (2) The set of zeros of Pi(x) is discrete for 1in,

    (3) If Pi(x0)=0, then Pi1(x0)Pi+1(x0)<0 for 1in1,

    (4) If Pi(x0)=0, then Pi1(x0)[Pi(x0+ε2)Pi(x0ε1)]<0 for 1in and sufficiently small ε1,ε2>0,

    then the sequence P0(x),P1(x),,Pn(x) has the Sturm sequence properties.

    Theorem 1.1. [28] Suppose that the sequence P0(x),P1(x),,Pn(x) has the Sturm sequence properties on (a,b). Let α<β be any numbers in (a,b). Then Pn(x) has exactly c(β)c(α) distinct zeros in interval (α,β), where c(α) denotes the number of changes in sign of consecutive members of the sequence P0(α), P1(α),,Pn(α).

    Theorem 1.2. [28] If P0(x),P1(x),,Pn(x) has the Sturm sequence properties, then the zeros of Pi(x) and Pi1(x) are interlaced for 1in.

    Sturm's Theorem was applied to symmetric tridiagonal matrices by Greenberg [28] to solve some nonlinear eigenvalue problems. Mersin and Bahşi [32] applied Sturm's Theorem to the generalized Frank matrices, and examined their eigenvalues by using the Sturm sequence properties.

    In the present paper, we examine the matrix Amin=[amin(i,j)]ni,j=1 given in the Eq (1.2), considering different required conditions for the sequence {as}, such as positive, and either strictly increasing or strictly decreasing. We seek to answer the following questions: What is the recurrence relation for the characteristic polynomial of the matrix Amin? Are there any relations between the coefficients of the characteristic polynomials of this matrix? Does the sequence of the characteristic polynomials of the i×i(in) Min matrices has the Sturm sequence properties? Can we determine the number of the eigenvalues of Min matrix Amin in an interval?

    Theorem 2.1. Let Pn(λ) be the characteristic polynomial of the matrix Amin=[amin(i,j)]ni,j=1 for any real sequence {as}. Then,

    Pn(λ)=(anan12λ)Pn1(λ)λ2Pn2(λ), (2.1)

    with the initial conditions P0(λ)=1 and P1(λ)=a1λ.

    Proof. The characteristic polynomial of the matrix Amin is

    Pn(λ)=det(AminλI)=|a1λa1a1a1a1a1a2λa2a2a2a1a2a3λa3a3a1a2a3an1λan1a1a2a3an1anλ|.

    Subtracting ith column from the (i+1)th column and then ith row from the (i+1)th row for i=n1,n2,,1, respectively, we get

    Pn(λ)=|a1λλ000λa2a12λλ000λa3a22λ00000an1an22λλ000λanan12λ|.

    Thus, we have

    Pn(λ)=(anan12λ)Pn1(λ)λ2Pn2(λ),

    with initials P0(λ)=1 and P1(λ)=a1λ.

    Theorem 2.2. Let Pn(λ)=λn+γ(n)n1λn1++γ(n)1λ+γ(n)0 be as in Theorem 2.1. Then,

    γ(n)0=(an+an1)γ(n1)0=(1)ndet(Amin),
    γ(n)1=(an+an1)γ(n1)1+2γ(n1)0,
    γ(n)i=(an+an1)γ(n1)i+2γ(n1)i1γ(n2)i2,2in2,

    and

    γ(n)n1=an+an1+2γ(n1)n2γ(n2)n3=tr(Amin).

    Proof. The following recurrence relation

    Pn(λ)=(2λan+an1)Pn1(λ)λ2Pn2(λ),  (2.2)

    is equivalent to the recurrence relation in (2.1). Then, considering Eq (2.2) and the coefficients of Pn(λ),Pn1(λ) and Pn2(λ), we have

    λn+γ(n)n1λn1++γ(n)1λ+γ(n)0=(2λan+an1)(λn1+γ(n1)n2λn2++γ(n1)1λ+γ(n1)0)λ2(λn2+γ(n2)n3λn3++γ(n2)1λ+γ(n2)0).

    Thus, desired formulas are obtained by comparison of the coefficients. Also,

    γ(n)0=(an+an1)γ(n1)0=(an+an1)(an1+an2)γ(n2)0=(an+an1)(an1+an2)(an2+an3)(a2+a1)γ(1)0=(an+an1)(an1+an2)(an2+an3)(a2+a1)(a1)=(1)ndet(Amin).

    To prove the equality

    γ(n)n1=an+an1+2γ(n1)n2γ(n2)n3=tr(Amin),

    we must show that

    2γ(n1)n2γ(n2)n3=a1a2an22an1 (2.3)

    is valid for n3. We use the induction method on n. Since

    2γ(2)1γ(1)0=2((a2+a1)γ(1)1+2γ(1)0)γ(1)0=2(a2+a1)+3γ(1)0=2(a2+a1)+3(a1+a0)γ(0)0=a12a2,

    the result is true for n=3. Assume that the result is true for n=k. That is, the equality

    2γ(k1)k2γ(k2)k3=a1a2ak22ak1 (2.4)

    is true. Considering Eq (2.4), we get

    2γ(k)k1γ(k1)k2=2((ak+ak1)γ(k1)k1+2γ(k1)k2γ(k2)k3)γ(k1)k2=2(ak+ak1)+3γ(k1)k22γ(k2)k3+γ(k1)k2γ(k1)k2=2(ak+ak1)+2(2γ(k1)k2γ(k2)k3)γ(k1)k2=2(ak+ak1)+2(a1a2ak22ak1)γ(k1)k2,

    for n=k+1. Since

    γ(k1)k2=(ak1+ak2)γ(k2)k2+2γ(k2)k3γ(k3)k4=(ak1+ak2)+(a1a2ak32ak2),

    we have

    2γ(k)k1γ(k1)k2=2(ak1+ak2)+2(a1a2ak22ak1)(ak1+ak2)(a1a2ak32ak2)=a1a2ak12ak.

    This completes the proof of Eq (2.3). Hence, we get

    γ(n)n1=an+an1+2γ(n1)n2γ(n2)n3=an+an1+(a1a2an22an1)=a1a2an=tr(Amin).

    Remark 2.1. We encounter the term a0 for n=1 in the proof of Theorem 2.2. We should specify that the reader should take a0=0 when required.

    Now, we demonstrate that the sequence of the characteristic polynomials of the i×i(in) Min matrices Amin

    P0(λ)=1,P1(λ),P2(λ),,Pn1(λ),Pn(λ)

    has the Sturm sequence properties according to different required conditions for {as} such as positive, and either strictly increasing or strictly decreasing. First, we give some Lemmas to use for this purpose.

    Lemma 2.1. If the real sequence {as} is positive, and either strictly increasing or strictly decreasing, then

    (i) Zero is not a root of Pi(λ) for 1in (or equivalently zero is not an eigenvalue of the i×i matrix Amin for 1in),

    (ii) Two consecutive terms Pi(λ), Pi+1(λ) do not have a common zero for 1in1.

    Proof. Let the real sequence {as} be positive, and either strictly increasing or strictly decreasing. Then,

    (i) From the recurrence relation (2.1) and the equality P1(0)=a1, we get

    Pi(0)=(aiai1)Pi1(0)=(aiai1)(ai1ai2)Pi2(0)=(aiai1)(ai1ai2)(a2a1)a10

    for 1in.

    (ii) Suppose that Pi+1(λ0)=Pi(λ0)=0 for some i with 1in1, then considering the recurrence relation (2.1), we get

    Pi1(λ0)=Pi2(λ0)==P0(λ0)=0.

    Since this result contradicts P0(λ)=1, two consecutive terms Pi(λ), Pi+1(λ) can not have a common zero for 1in1.

    Lemma 2.2. Suppose that the real sequence {as} is positive, and either strictly increasing or strictly decreasing.

    (i) If the sequence {as} is strictly increasing, and J(0,) is an interval that contains no zeros of Pi1(λ) for 1in, then Pi(λ)Pi1(λ) is strictly decreasing on interval J.

    (ii) If the sequence {as} is strictly decreasing, I1(,0) and I2(0,) are any intervals that contain no zeros of Pi1(λ) for 1in, then Pi(λ)Pi1(λ) is strictly increasing on interval I1, and strictly decreasing on I2.

    Proof. We use the induction method on i for the proofs.

    (i) Since

    P1(λ)P0(λ)=a1λ1=a1λ

    is strictly decreasing on interval (0,), the result is true for i=1. Let the result be true for ik, and K(0,) be an interval that contains no zeros of Pk(λ) and Pk1(λ). Then, from the recurrence relation (2.1), we have

    Pk+1(λ)Pk(λ)=(ak+1ak2λ)λ2Pk1(λ)Pk(λ)

    for k+1n. It is clear that ak+1ak2λ is strictly decreasing on interval (0,). Also, considering the assumption (for ik), we can say that λ2Pk1(λ)Pk(λ) is strictly decreasing on K. Then, we have Pk+1(λ)Pk(λ) is strictly decreasing on K.

    If Pk1(y)=0 and (x,z)(0,) is an interval that contains y, but no zeros of Pk(λ), then Pk+1(λ)Pk(λ) is strictly decreasing on intervals (x,y) and (y,z). From the continuity, we have Pk+1(λ)Pk(λ) is strictly decreasing on interval (x,z).

    (ii) Since

    P1(λ)P0(λ)=a1λ1=a1λ

    is strictly increasing on interval (,0), and strictly decreasing on interval (0,), the result is true for i=1. Let the result be true for ik, and K1(,0), K2(0,) be two intervals that have no zeros of Pk(λ) and Pk1(λ). Then for k+1n, considering the recurrence relation (2.1), we have

    Pk+1(λ)Pk(λ)=(ak+1ak2λ)λ2Pk1(λ)Pk(λ).

    ak+1ak2λ is strictly increasing on interval (,0), and strictly decreasing on (0,). From the assumption (for ik), λ2Pk1(λ)Pk(λ) is strictly increasing on K1, and strictly decreasing on K2. Then, we have Pk+1(λ)Pk(λ) is strictly increasing on K1, and strictly decreasing on K2.

    Suppose that Pk1(y1)=0, and (x1,z1)(,0) is an interval that contains y1, but no zeros of Pk(λ). Then, Pk+1(λ)Pk(λ) is strictly increasing on intervals (x1,y1) and (y1,z1). Thus, considering the continuity, we have Pk+1(λ)Pk(λ) is strictly increasing on interval (x1,z1). Similarly, if Pk1(y2)=0 and (x2,z2)(0,) is an interval that contains y2, but no zeros of Pk(λ), then Pk+1(λ)Pk(λ) is strictly decreasing on intervals (x2,y2) and (y2,z2). From the continuity, we have Pk+1(λ)Pk(λ) is strictly decreasing on interval (x2,z2).

    Theorem 2.3. Suppose that the real sequence {as} is positive, either strictly increasing or strictly decreasing and the sequence

    P0(λ)=1,P1(λ),P2(λ),,Pn1(λ),Pn(λ) (2.5)

    consists of the characteristic polynomials of the i×i(in) matrices Amin.

    (i) If the sequence {as} is strictly increasing, then the sequence given in Eq (2.5) has the Sturm sequence properties on interval (0,),

    (ii) If the sequence {as} is strictly decreasing, then the sequence given in Eq (2.5) has the Sturm sequence properties on interval (,).

    Proof. (i) Let the sequence {as} be strictly increasing. We must show that four conditions (1)(4) in Definition 1.1 are satisfied by the sequence of the characteristic polynomials of the i×i(in) matrices Amin.

    (1) It is clear that P0(λ)=1 has no zeros.

    (2)P1(λ)=a1λ has only one zero as λ0=a1. Thus, (2) is true for i=1. Suppose that (2) is true for ik, then the set of zeros of Pk(λ) is discrete. Considering the recurrence relation (2.1), we have

    Pk+1(λ)=Pk(λ)[(ak+1ak2λ)λ2Pk1(λ)Pk(λ)].

    By using Lemma 2.1(ii), Pk+1(λ) and Pk(λ) have no common zero, and by using Lemma 2.2(i), Pk+1(λ)Pk(λ) is strictly decreasing between any two consecutive zeros of Pk(λ). Hence, Pk+1(λ) has at most one zero between any two consecutive zeros of Pk(λ). That is, (2) is true for k+1n.

    (3) Considering the recurrence relation (2.1) we have

    Pi+1(λ)=(ai+1ai2λ)Pi(λ)λ2Pi1(λ),

    for 1in1. If Pi(λ)=0, then we get Pi+1(λ)=λ2Pi1(λ) for 1in1. Since λ2>0, the inequality Pi+1(λ)Pi1(λ)<0 is true for λ(0,).

    (4) Suppose that Pi(λ0)=0 for 1in, and [λ0ε1,λ0+ε2] is an interval that contains no zeros of Pi1(λ) for sufficiently small ε1,ε2>0. Then, the sign of Pi1(λ) does not change. By using Lemma 2.2 (i) Pi(λ)Pi1(λ) is strictly decreasing on interval J(0,). Thus, the sign of Pi(λ)Pi1(λ) (or equivalently the sign of Pi1(λ)Pi(λ)) is (+) and () in intervals [λ0ε1,λ0) and (λ0,λ0+ε2], respectively. That is,

    Pi1(λ0ε1)Pi(λ0ε1)>0>Pi1(λ0+ε2)Pi(λ0+ε2).

    Since the sign of Pi1(λ) does not change in interval [λ0ε1,λ0+ε2], we have

    Pi1(λ0)Pi(λ0ε1)>0>Pi1(λ0)Pi(λ0+ε2)

    and

    Pi1(λ0)[Pi(λ0+ε2)Pi(λ0ε1)]<0.

    This completes the proof.

    (ii) The proof is similar to the proof of (i).

    Theorem 2.4. Suppose that the real sequence {as} is positive, and either strictly increasing or strictly decreasing.

    (i) If {as} is strictly increasing, then all eigenvalues of the n×n matrix Amin are distinct and positive,

    (ii) If {as} is strictly decreasing, then one of the eigenvalues of the n×n matrix Amin is positive, and the remaining n1 eigenvalues are distinct and negative.

    Proof. We must compute the numbers of the eigenvalues in intervals (0,) and (,0) for the proofs. Assume that λı and λıı are the minimum and maximum zeros of Pi(λ) for 1in, respectively. By Theorems 1.1 and 2.3, the number of distinct zeros of Pi(λ) in interval (x,y) is equal to ci(y)ci(x), where ci(α) is the number of sign changes of the sequence

    P0(α),P1(α),P2(α),,Pi1(α),Pi(α),

    for 1in. Because 0 is not a zero of Pi(λ), the sign of Pi(λ) does not change in interval [0,λı). Then, the sign of Pi(x) is equal to sign of Pi(0) for x(0,λı). Thus, we get ci(x)=ci(0). The sign of Pi(λ) does not change in interval (λıı,). Then, we have ci(y)=ci() for y(λıı,). Since, the degree of Pi(λ) is i, the form of Pi(λ) is

    Pi(λ)=(1)iλi+. (2.6)

    Then, the sign of Pi() is (1)i. Hence, we have ci()=i. Since ci(y)ci(x)=ci()ci(0) for x(0,λı) and y(λıı,), we must also calculate ci(0), to evaluate the number of eigenvalues in interval (0,). Considering as is a positive real number, we have

    (i) For the strictly increasing sequence {as}, it is clear that

    Pi(0)=(aiai1)(ai1ai2)(a2a1)a1>0.

    Then, we have ci(0)=0. Thus, the number of distinct zeros of Pi(λ) in interval (x,y) for x(0,λı) and y(λıı,) is

    ci(y)ci(x)=ci()ci(0)=i.

    Because the number of zeros of Pi(λ) is i, we can say that all the zeros of Pi(λ) are in interval (x,y) for x(0,λı) and y1(λıı,). Thus, all the zeros of Pi(λ) are distinct and positive for 1in. In other words, all eigenvalues of the n×n matrix Amin are distinct and positive.

    (ii) For the strictly decreasing sequence {as}, the sign of

    Pi(0)=(aiai1)(ai1ai2)(a2a1)a1

    is (1)i1. Hence, we have ci(0)=i1. Thus, the number of zeros of Pi(λ) in interval (x,y) for x(0,λı) and y(λıı,) is

    ci(y)ci(x)=ci()ci(0)=i(i1)=1.

    That is, Pi(λ) has one eigenvalue in interval (0,).

    Now, we show that the n×n matrix Amin has n1 distinct eigenvalues in interval (,0). The sign of Pi(λ) does not change in interval (,λı) and ci(x)=ci() for x(,λı). Since the number of distinct zeros of Pi(λ) in interval (x,y) is ci(y)ci(x)=ci(0)ci() for x(,λı) and y(λıı,0), we must compute the value ci(). Considering Eq (2.6), it is clear that

    Pi()>0.

    Then, the number of sign change of Pi() is zero. Thus, ci()=0, and we have

    ci(y)ci(x)=ci(0)ci()=(i1)0=i1.

    Since the number of zeros of Pi(λ) is i, and one of the zeros is positive, remaining i1 zeros of Pi(λ) are in interval (x,y) for x(,λı) and y(λıı,0). Hence, Pn(λ) has n1 eigenvalues in interval (,0). This shows that one of the eigenvalues of the n×n matrix Amin is positive, and the remaining n1 eigenvalues are distinct and negative.

    Remark 2.2. We note that, we use the notations Pi() and ci() in the proof of Theorem 2.4, rather than limλPi(λ) and limλci(λ), respectively.

    Theorem 2.5. If the real sequence {as} is positive, and either strictly increasing or strictly decreasing, then the eigenvalues of i×i and (i1)×(i1) matrices Amin are interlaced for 2in. That is,

    λ(i)1>λ(i1)1>λ(i)2>λ(i1)2>>λ(i1)i1>λ(i)i,

    where λ(i)s's are the eigenvalues of the i×i matrices Amin for s=1,2,,i.

    Proof. Theorems 1.2 and 2.3 give the desired result immediately.

    Remark 2.3. Our results work even if the real sequence {as} is negative and strictly decreasing (or strictly increasing). If bs=as, then bs is a positive real number, the sequence {bs} is strictly increasing (or strictly decreasing). Since Amin=Bmin, all eigenvalues of Bmin have opposite sign with all eigenvalues of Amin. For example, for the negative, either strictly increasing or strictly decreasing real sequence {as}:

    (i) If {as} is strictly decreasing, then all eigenvalues of the n×n matrix Amin are distinct and negative,

    (ii) If {as} is strictly increasing, then one of the eigenvalues of the n×n matrix Amin is negative, and the remaining n1 eigenvalues are distinct and positive.

    In this section, we illustrate our results with the following example.

    Consider the real sequence {as} with the elements as=2s1. Then, the 5×5 Min matrix corresponding to this sequence is

    Amin=[amin(i,j)]5i,j=1=[1111113333135551357713579],

    and its Hadamard inverse is

    A(1)min=[1amin(i,j)]5i,j=1=[11111113131313113151515113151717113151719].

    Theorem 2.1 yields the characteristic polynomials of the i×i(2i5) matrices Amin and A(1)min as

    Pi(λ)=2(1λ)Pi1(λ)λ2Pi2(λ)

    and

    Qi(μ)=2(1aiai1+μ)Qi1(μ)μ2Qi2(μ).

    Thus,

    P0(λ)=1,P1(λ)=1λ,P2(λ)=2(1λ)(1λ)λ2=λ24λ+2,P3(λ)=2(1λ)(λ24λ+2)λ2(1λ)=λ3+9λ212λ+4,P4(λ)=2(1λ)(λ3+9λ212λ+4)λ2(λ24λ+2)=λ416λ3+40λ232λ+8,P5(λ)=2(1λ)(λ416λ3+40λ232λ+8)λ2(λ3+9λ212λ+4)=λ5+25λ4100λ3+140λ280λ+16, (3.1)

    and

    Q0(μ)=1,Q1(μ)=1μ,Q2(μ)=2(13+μ)(1μ)μ2=μ243μ23,Q3(μ)=2(115+μ)(μ243μ23)μ2(1μ)=μ3+2315μ2+6845μ+445,Q4(μ)=2(135+μ)(μ3+2315μ2+6845μ+445)μ2(μ243μ23)=μ4176105μ338481575μ24161575μ81575,Q5(μ)=2(163+μ)(μ4176105μ338481575μ24161575μ81575)μ2(μ3+2315μ2+6845μ+445)=μ5+563315μ4+11339633075μ3+5129299225μ2+36819845μ+1699225. (3.2)

    If we compute Pi5(λ) and Qi5(μ) using det(AminλI) and det(A(1)minμI) for i5 respectively, we obtain the same results as above.

    There are the following relations between the coefficients of the characteristic polynomials given in Eq (3.1) considering the form of Pn(λ)=λn+γ(n)n1λn1++γ(n)1λ+γ(n)0 as mentioned in Theorem 2.2, we have the coefficients as

    γ(1)0=(a1+a0)γ(0)0=(1+0)1=1,γ(2)0=(a2+a1)γ(1)0=(3+1)(1)=2,γ(3)0=(a3+a2)γ(2)0=(5+3)(2)=4,γ(4)0=(a4+a3)γ(3)0=(7+5)(4)=8,γ(5)0=(a5+a4)γ(4)0=(9+7)(8)=16,γ(2)1=(a2+a1)+2γ(1)0=(3+1)+2γ(1)0=4,γ(3)1=(a3+a2)γ(2)1+2γ(2)0=(5+3)(4)+2(2)=12,γ(4)1=(a4+a3)γ(3)1+2γ(3)0=(7+5)(12)+2(4)=32,γ(5)1=(a5+a4)γ(4)1+2γ(4)0=(9+7)(32)+2(8)=80,γ(3)2=(a3+a2)+2γ(2)1γ(1)0=(3+1)+2(4)(1)=9,γ(4)2=(a4+a3)γ(3)2+2γ(3)1γ(2)0=(5+3)(9)+2(12)(2)=40,γ(5)2=(a5+a4)γ(4)2+2γ(4)1γ(3)0=(7+5)(40)+2(32)(4)=140,γ(4)3=(a4+a3)+2γ(3)2γ(2)1=(7+5)+2(9)(4)=16,γ(5)3=(a5+a4)γ(4)3+2γ(4)2γ(3)1=(7+5)(16)+2(40)(12)=100,γ(5)4=(a5+a4)+2γ(4)3γ(3)2=(9+7)+2(16)(9)=25.

    Considering the values

    det(Amin)=1(31)(53)(75)(97)=16,
    tr(Amin)=1+3+5+7+9=25,

    we observe that the equalities given in Theorem 2.2 are provided for the 5×5 matrix Amin. For example

    γ(5)0=16(1)5det(Amin)

    and

    γ(5)4=25=tr(Amin).

    The relations for the coefficients of the characteristic polynomials of the matrix A(1)min given in Eq (3.2) can be obtained similarly.

    The roots of Pi5(λ) and Qi5(μ) (or the eigenvalues of i×i(i5) matrices Amin and A(1)min, respectively) are

    λ(1)1=1,λ(2)1=3.41,λ(2)2=0.59,λ(3)1=7.46,λ(3)2=1,λ(3)3=0.54,λ(4)1=13.1,λ(4)2=1.62,λ(4)3=0.723,λ(4)4=0.520,λ(5)1=20.4,λ(5)2=2.42,λ(5)3=1,λ(5)4=0.630,λ(5)5=0.512,

    and

    μ(1)1=1,μ(2)1=1.72,μ(2)2=0.383,μ(3)1=2.229,μ(3)2=0.063,μ(3)3=0.63,μ(4)1=2.64,μ(4)2=0.025,μ(4)3=0.091,μ(4)4=0.847,μ(5)1=2.991,μ(5)2=0.013,μ(5)3=0.034,μ(5)4=0.114,μ(5)5=1.042,

    where λ(i)s and μ(i)s denote the roots of Pi5(λ) and Qi5(μ), respectively for s=1,2,,i. Hence, we have

    Pi5(λ) and Qi5(μ) do not vanish for λ=μ=0.

    Pi4(λ) and Pi+1(λ) (or Qi4(μ) and Qi+1(μ)) have not a common zero.

    ● The sets of zeros of both Pi5(λ) and Qi5(μ) are discrete.

    ● If Pi4(λ)=0, then Pi1(λ)Pi+1(λ)<0. For example, since P2(1)=1, P3(1)=0, P4(1)=1, we have P2(1)P4(1)=1<0. Similarly, if Qi4(μ)=0, then Qi1(λ)Qi+1(λ)<0. For example, since Q2(0.063)=0.579, Q3(0.063)=0, Q4(0.063)=0.002, we have Q2(0.063)Q4(0.063)=0.001<0.

    ● If Pi(λ0)=0 and Qi(μ0)=0, then for sufficiently small ε1,ε2>0,

    Pi1(λ0)[Pi(λ0+ε2)Pi(λ0ε1)]<0,

    and

    Qi1(μ0)[Qi(μ0+ε2)Qi(μ0ε1)]<0,

    where 1i5. For example, since

    P3(1)=0,P2(1)=1,P3(1+1100)=0.031,P3(111000)=0.003,

    we have

    P2(1)[P3(1+1100)P3(111000)]=0.034<0,

    where ε1=11000,ε2=1100. Similarly, since

    Q4(2.64)=0,Q3(2.64)=3.635,Q4(2.64+110000)=0.01,Q4(2.641100)=0.248,

    we have

    Q3(2.64)[Q4(2.64+110000)Q4(2.641100)]=0.938<0,

    where ε1=1100,ε2=110000.

    ● The sequences P0(λ), P1(λ), P2(λ), P3(λ), P4(λ), P5(λ), and Q0(μ), Q1(μ), Q2(μ), Q3(μ), Q4(μ), Q5(μ) have the Sturm sequence properties.

    ● All of eigenvalues of the i×i(i5) matrices Amin (or the zeros of Pi5(λ)) are distinct and positive. Also one of the eigenvalues of the i×i(i5) matrices A(1)min (or the zeros of Qi5(μ)) is positive, and the remaining i1 eigenvalues are distinct and negative.

    ● The eigenvalues of the i×i and (i1)×(i1) matrices Amin are interlaced for 2i5. For example,

    λ(5)1=20.4>λ(4)1=13.1>λ(5)2=2.42>λ(4)2=1.62>λ(5)3=1>λ(4)3=0.723>λ(5)4=0.630>λ(4)4=0.520>λ(5)5=0.512.

    Similarly, the eigenvalues of the i×i and (i1)×(i1) matrices A(1)min are interlaced for 2i5. For example,

    μ(5)1=2.991>μ(4)1=2.64>μ(5)2=0.013>μ(4)2=0.025>μ(5)3=0.034>μ(4)3=0.091>μ(5)4=0.114>μ(4)4=0.847>μ(5)5=1.042.

    Finally, we compute the number of eigenvalues of the 5×5 matrix Amin in intervals (0,2) and (2,25). So then, we need the number of sign changes of Pi5(λ) for λ=0, λ=2, and λ=25. Table 1 serves this need.

    Table 1.  The number of sign changes of Pi5(λ) for λ=0, λ=2, and λ=25.
    Characteristic polynomials of the i×i(i5) matrices Amin Sign of Pi(λ) for λ=0 Sign of Pi(λ) for λ=2 Sign of Pi(λ) for λ=25
    P0(λ)=1 + + +
    P1(λ)=1λ +
    P2(λ)=λ24λ+2 + +
    P3(λ)=λ3+9λ212λ+4 + +
    P4(λ)=λ416λ3+40λ232λ+8 + +
    P5(λ)=λ5+25λ4100λ3+140λ280λ+16 +
    Number of sign changes c5(0)=0 c5(2)=3 c5(25)=5

     | Show Table
    DownLoad: CSV

    From Table 1, we have c5(0)=0, c5(2)=3, and c5(25)=5, where c5(α) denotes the number of changes in sign of Pi5(α). Thus, the number of eigenvalues of the 5×5 matrix Amin in interval (0,2) is c5(2)c5(0)=30=3, and the number of eigenvalues in interval (2,25) is c5(25)c5(2)=53=2. Really, the eigenvalues of the 5×5 matrix Amin are 20.4,2.42,1,0.630, and 0.512.

    Similarly, we compute the number of eigenvalues of the 5×5 matrix A(1)min in intervals (2,0) and (0,3). Table 2 includes the number of sign changes of Qi5(μ) for μ=2, μ=0, and μ=3.

    Table 2.  The number of sign changes of Qi5(μ) for μ=2, μ=0, and μ=3.
    Characteristic polynomials of the i×i(i5) matrices A(1)min Sign of Qi(μ) for μ=2 Sign of Qi(μ) for μ=0 Sign of Qi(μ) for μ=3
    Q0(μ)=1 + + +
    Q1(μ)=1μ + +
    Q2(μ)=μ243μ23 + +
    Q3(μ)=μ3+2315μ2+6845μ+445 + +
    Q4(μ)=μ4176105μ338481575μ24161575μ81575 + +
    Q5(μ)=μ5+563315μ4+11339633075μ3+5129299225μ2+36819845μ+1699225 + +
    Number of sign changes c5(2)=0 c5(0)=4 c5(3)=5

     | Show Table
    DownLoad: CSV

    According to Table 2, we have c5(2)=0, c5(0)=4, and c5(3)=5. Thus, the number of eigenvalues of the 5×5 matrix A(1)min in interval (2,0) is c5(0)c5(2)=40=4, and the number of eigenvalues in interval (0,3) is c5(3)c5(0)=54=1. Really, the eigenvalues of the 5×5 matrix A(1)min are 2.991,0.013,0.034,0.114, and 1.042.

    In this paper, we obtained a recurrence relation for the characteristic polynomials of the real symmetric Min matrix Amin=[amin(i,j)]ni,j=1, where as's are the elements of a real sequence {as}. We also gave some relations between the coefficients of the characteristic polynomials of this matrix. Additionally, we obtained that the sequence of the characteristic polynomials of the i×i(in) Min matrices satisfies the Sturm sequence properties considering different required conditions for the real sequence {as}. We showed that the eigenvalues of the i×i and (i1)×(i1) Min matrices are interlaced as a consequence of Sturm's Theorem, where 2in. It is well known that the eigenvalues of real symmetric matrices are real; we specified how many of the real eigenvalues are positive, and how many are negative of the n×n matrices Amin with the help of Sturm's Theorem.

    The author declares no conflict of interest in this paper.



    [1] G. Pólya, G. Szegö, Problems and theorems in analysis II, Berlin: Springer, 1998. http://dx.doi.org/10.1007/978-3-642-61905-2
    [2] M. Catalani, A particular matrix and its relationships with Fibonacci numbers, arXiv: math/0209249.
    [3] R. Bhatia, Infinitely divisible matrices, Am. Math. Mon., 113 (2006), 221–235. http://dx.doi.org/10.2307/27641890 doi: 10.2307/27641890
    [4] R. Bhatia, Min matrices and mean matrices, Math. Intelligencer, 33 (2011), 22–28. http://dx.doi.org/10.1007/s00283-010-9194-z doi: 10.1007/s00283-010-9194-z
    [5] C. Da Fonseca, On the eigenvalues of some tridiagonal matrices, J. Comput. Appl. Math., 200 (2007), 283–286. http://dx.doi.org/10.1016/j.cam.2005.08.047 doi: 10.1016/j.cam.2005.08.047
    [6] L. Moyé, Statistical monitoring of clinical trials: fundamentals for investigators, New York: Springer, 2006. http://dx.doi.org/10.1007/0-387-27782-X
    [7] H. Neudecker, G. Trenkler, S. Liu, Problem section, Stat. Papers, 50 (2009), 221–223. http://dx.doi.org/10.1007/s00362-008-0174-8
    [8] K. Chu, S. Puntanen, G. Styan, Problem section, Stat. Papers, 52 (2011), 257–262. http://dx.doi.org/10.1007/s00362-010-0363-0
    [9] M. Mattila, P. Haukkanen, Studying the various properties of Min and Max matrices-elemantary vs. more advanced methods, Spec. Matrices, 4 (2016), 101–109. http://dx.doi.org/10.1515/spma-2016-0010 doi: 10.1515/spma-2016-0010
    [10] M. Bahşi, S. Solak, A particular matrix and its properties, Int. J. Math. Sci. Appl., 1 (2011), 971–974.
    [11] S. Solak, M. Bahşi, A particular matrix and its some properties, Sci. Res. Essays, 8 (2013), 1–5. http://dx.doi.org/10.5897/SRE11.410 doi: 10.5897/SRE11.410
    [12] M. Bahşi, S. Solak, Some particular matrices and their characteristic polynomials, Linear Multilinear A., 63 (2015), 2071–2078. http://dx.doi.org/10.1080/03081087.2014.940940 doi: 10.1080/03081087.2014.940940
    [13] J. Stuart, Nested matrices and inverse M-matrices, Czech. Math. J., 65 (2015), 537–544. http://dx.doi.org/10.1007/s10587-015-0192-3 doi: 10.1007/s10587-015-0192-3
    [14] S. Jafari-Petroudi, B. Pirouz, A particular matrix, its inversion and some norms, Appl. Comput. Math., 4 (2015), 47–52. http://dx.doi.org/10.11648/j.acm.20150402.13
    [15] S. Jafari-Petroudi, B. Pirouz, A note on Hadamard inverse and Hadamard exponential of a matrix with Fibonacci numbers, Proceedings of The 7th National Conference on Mathematics, 2015, 28–29.
    [16] S. Jafari-Petroudi, M. Pirouz, A note on tribonacci numbers with particular matrices, Proceedings of The 28th International Conference of The Jangjeon Mathematical Society, 2015, 71.
    [17] S. Jafari-Petroudi, M. Pirouz, On the bounds for the spectral norm of particular matrices with Fibonacci and Lucas numbers, Int. J. Adv. Appl. Math. Mech., 3 (2016), 82–90.
    [18] S. Jafari-Petroudi, M. Pirouz, Toward special symmetric matrices with harmonic numbers, Proceedings of The 8th National Conference on Mathematics, 2016, 11–12.
    [19] E. Kılıç, T. Arıkan, Studying new generalizations of Max-Min matrices with a novel approach, Turk. J. Math., 43 (2019), 2010–2024.
    [20] T. Solmaz, M. Bahşi, Max and Min matrices with hyper-Fibonacci numbers, Asian-Eur. J. Math., 15 (2022), 2250084. http://dx.doi.org/10.1142/S179355712250084X doi: 10.1142/S179355712250084X
    [21] D. Özgul, M. Bahşi, Min matrices with hyper-Lucas numbers, J. Sci. Arts, 4 (2020), 855–864. http://dx.doi.org/10.46939/J.Sci.Arts-20.4-a07 doi: 10.46939/J.Sci.Arts-20.4-a07
    [22] C. Kızılateş, N. Terzioğlu, On r-min and r-max matrices, J. Appl. Math. Comput., 68 (2022), 4559–4588. http://dx.doi.org/10.1007/s12190-022-01717-y doi: 10.1007/s12190-022-01717-y
    [23] M. Jain, S. Iyengar, R. Jain, Numerical methods: problems and solutionsz, Bangalore: New Age International, 2007.
    [24] P. Ciarlet, J. Lions, Solution of equations in Rn, In: Handbook of numerical analysis, Amsterdam: Elsevier, (1994), 625–778.
    [25] M. Sturm, Analyse d' un Mémoire sur la résolution des équations numériques, In: Collected works of Charles François Sturm, Basel: Birkhäuser, 2009,323–326. http://dx.doi.org/10.1007/978-3-7643-7990-2_24
    [26] M. Sturm, Extrait d'un Mémoire sur L'intécration d'un système d'équations différentielles linéaires, présenté à l'Académie des sciences, In: Collected works of Charles François Sturm, Basel: Birkhäuser, 2009,334–342. http://dx.doi.org/10.1007/978-3-7643-7990-2_27
    [27] P. Sturm, Mémoire sur la résolution des équations numériques, In: Collected works of Charles François Sturm, Basel: Birkhäuser, 2009,345–390. http://dx.doi.org/10.1007/978-3-7643-7990-2_29
    [28] L. Greenberg, Sturm sequences for nonlinear eigenvalue problems, SIAM J. Math. Anal., 20 (1989), 182–199. http://dx.doi.org/10.1137/0520015 doi: 10.1137/0520015
    [29] E. Isaacson, H. Keller, Analysis of numerical methods, 2 Eds., New York: John Wiley-Sons, 1966.
    [30] J. Stoer, R. Bulirsch, Introduction to numerical analysis, New York: Springer-Verlag, 2002. http://dx.doi.org/10.1007/978-0-387-21738-3
    [31] A. Mostowski, M. Stark, Introduction to higher algebra, Pergamon: Elsevier, 1964. http://dx.doi.org/10.1016/C2013-0-10019-0
    [32] E. Mersin, M. Bahşi, Sturm theorem for the generalized Frank matrix, Hacet. J. Math. Stat., 50 (2021), 1002–1011. http://dx.doi.org/10.15672/hujms.773281 doi: 10.15672/hujms.773281
  • This article has been cited by:

    1. Wei Xie, Characteristic polynomials of some min and max matrices, 2024, 03, 2811-0072, 10.1142/S2811007224500019
    2. Hasan Gökbaş, Some properties of the generalized max Frank matrices, 2024, 9, 2473-6988, 26826, 10.3934/math.20241305
    3. Emrah Polatlı, On some properties of a generalized min matrix, 2023, 8, 2473-6988, 26199, 10.3934/math.20231336
    4. Baijuan Shi, Can Kızılateş, A new generalization of the Frank matrix and its some properties, 2024, 43, 2238-3603, 10.1007/s40314-023-02524-2
    5. Efruz Özlem Mersin, Mustafa Bahşi, Bounds for the maximum eigenvalues of the Fibonacci-Frank and Lucas-Frank matrices, 2024, 73, 1303-5991, 420, 10.31801/cfsuasmas.1299736
  • Reader Comments
  • © 2023 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(1591) PDF downloads(62) Cited by(5)

Figures and Tables

Tables(2)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog