Kyungpook Mathematical Journal 2018; 58(4): 637-650  
On [m,C]-symmetric Operators
Department of Mathematics, Kanagawa University, Hiratsuka 259-1293, Japan, e-mail : chiyom01@kanagawa-u.ac.jp, Department of Mathematics and Statistics, Sejong University, Seoul 05006, Korea, e-mail : jieunlee7@sejong.ac.kr and jieun7@ewhain.net, Department of Mathematics, Tohoku Medical and Pharmaceutical University, Sendai 981-8558, Japan, e-mail : tanahasi@tohoku-mpu.ac.jp, Meguro-ku Nakane 11-10-201, Tokyo 152-0031, Japan, e-mail : juntomi@med.email.ne.jp
*Corresponding Author.
Received: August 23, 2017; Revised: January 12, 2018; Accepted: October 23, 2018; Published online: December 23, 2018.
© Kyungpook Mathematical Journal. All rights reserved.

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Abstract

In this paper first we show properties of isosymmetric operators given by M. Stankus [13]. Next we introduce an [m,C]-symmetric operator T on a complex Hilbert space ℋ. We investigate properties of the spectrum of an [m,C]-symmetric operator and prove that if T is an [m,C]-symmetric operator and Q is an n-nilpotent operator, respectively, then T +Q is an [m+2n − 2, C]-symmetric operator. Finally, we show that if T is [m,C]-symmetric and S is [n,D]-symmetric, then TS is [m+n−1, CD]-symmetric.

Keywords: Hilbert space, linear operator, conjugation, 𝑚-isometry, 𝑚-symmetric operator.
1. Introduction

Let ℋ be a complex Hilbert space with the inner product 〈,〉 and B(ℋ) be the set of bounded linear operators on ℋ. Let ℕ be the set of all natural numbers. For the study of Jordan operators, J.W. Helton ([9] and [10]) introduced an operator TB(ℋ) which satisfies

αm(T):=j=0m(-1)j(mj)T*m-jTj=0(m).

In particular, if T is normal, then αm(T) = (T* − T)m. An operator TB(ℋ) is said to be an m-symmetric operator if αm(T) = 0. Hence T is 1-symmetric if and only if T is Hermitian. It is well known that if T is m-symmetric, then T is n-symmetric for all nm. The concept of m-symmetric operators is little strong. For example, if T is m-symmetric, then σ(T) ⊂ ℝ (cf.[10]). And T is Hermitian even if T is 2-symmetric. Also if T is normal and m-symmetric, then T is Hermitian due to the fact that T* − T is normal and nilpotent, that is, T* − T = 0.

Recently, C. Gu and M. Stankus ([8]) showed interesting properties of m-symmetric operators. On the other hand, for m ∈ ℕ, an operator TB(ℋ) is said to be an m-isometric operator if

βm(T):=j=0m(-1)j(mj)T*m-jTm-j=0.

It is well known that if T is m-isometric, then T is n-isometric for all nm. In 1995, J. Agler and M. Stankus [1] introduced an m-isometric operator and showed many important results of such an operator. If T is an invertible m-isometric operator and m is even, then T is (m–1)-isometric. But if T is m-symmetric and m is even, then T is always (m–1)-symmetric by Theorem 3.4 of [12]. For every odd number m, there exists an invertible m-isometric operator T which is not (m–1)-isometric (see Theorem 1 in [5]).

Throughout this paper, let I be the identity operator on ℋ and m, n be natural numbers. An operator QB(ℋ) is said to be a nilpotent operator of order n if Qn = 0 and Qn–1 ≠ 0. For a subset A ⊂ ℂ, let A* = { : zA }. Let σ(T) and σp(T) be the spectrum and the point spectrum of TB(ℋ), respectively. The approximate point spectrum of T is defined by σa(T) := { z ∈ ℂ : T − zI is not bounded below}, and the surjective spectrum of T is defined by σs(T) := { z ∈ ℂ : T − zI is not surjective}. It is known that σ(T) = σa(T)∪ σs(T), σa(T)* = σs(T*), and σs(T)* = σa(T*).

2. Isosymmetric Operators

First we show the following result of m-symmetric operators.

Proposition 2.1

Let TB(ℋ). Then the following statements hold;

T is a 2-symmetric operator if and only if T is Hermitian.

Let T be an m-symmetric operator. For ab and non-zero vectors x, y ∈ ℋ, if Tx = ax and Ty = by, thenx, y〉 = 0.

Let T be an m-symmetric operator. For ab and sequences {xk}, {yk} of unit vectors of ℋ, if (T − a)xk → 0 and (T − b)yk → 0, thenlimkxk,yk=0.

Proof

(a) If T is Hermitian, then it is obvious that T is 2-symmetric. If T is 2-symmetric, then T is 1-symmetric from [12, Theorem 3.4] and so it is Hermitian. (b) Since a, bσ(T), it follows from [10] that a, b are real numbers. Hence it holds

0=αm(T)x,y=(b-a)m·x,y.

Since ab, we have 〈x, y〉 = 0.

(c) By similar arguments of the proof of (b), a, b are real numbers and it holds

0=limkαm(T)xk,yk=(b-a)m·limkxk,yk.

Since ab, we have limkxk,yk=0.

Definition 1

For an operator TB(ℋ), we define γm,n(T) by

γm,n(T)=j=0m(-1)j(mj)T*m-jαn(T)Tm-j=k=0n(-1)k(nk)T*n-kβm(T)Tk.

Then T is said to be (m, n)-isosymmetric if γm,n(T) = 0.

It is easy to see that

γm+1,n(T)=T*γm,n(T)T-γm,n(T)and γm,n+1(T)=T*γm,n(T)-γm,n(T)T.

Hence if T is (m, n)-isosymmetric, then T is (m′, n′)-isosymmetric for all n′n and n′n. M. Stankus proved the following properties.

Proposition 2.2.([13, Corollary 30])

Let T be (m, n)-isosymmetric.

If σ(T) ⊂ {x ∈ ℝ : |x| > 1} or σ(T) ⊂ {x ∈ ℝ : |x| < 1}, then T is n-symmetric.

If σ(T) ⊂ {e : 0 < θ < π} or σ(T) ⊂ {e : π < θ < 2π}, then T is m-isometric.

For a, b ∈ ℂ and non-zero vectors x, y ∈ ℋ, if Tx = ax, Ty = by, then it holds that

γm,n(T)x,y=(j=0m(-1)j(mj)T*m-jαn(T)Tm-j)x,y=(ab¯-1)m(a-b¯)nx,y.

Hence we have the following theorem.

Theorem 2.3

Let T be (m, n)-isosymmetric and x, y be unit vectors and xk, yk be sequences of unit vectors of ℋ.

If Tx = ax, Ty = by, ab and ab̄, thenx, y〈 = 0.

If (T − a)xk → 0, (T − b)yk → 0 (k → ∞), ab and ab̄, thenlimkxk,yk=0.

Theorem 2.4

Let T be (m, n)-isosymmetric.

Then Tk is (m, n)-isosymmetric for any k ∈ ℕ.

If T is invertible, then T−1is (m, n)-isosymmetric.

Proof

(1) Note that for k ∈ ℕ, the following equation holds;

(ykxk-1)m(yk-xk)n=((yx-1)(yk-1xk-1+yk-2xk-2++1))m·((y-x)(yk-1+yk-2x++xk-1))n==0m(k-1)j=0n(k-1)λμjym(k-1)-yn(k-1)-j(yx-1)m(y-x)nxjxm(k-1)-

where λ and μj are some constants. From this, we have

γm,n(Tk)==0m(k-1)j=0n(k-1)λμjT*m(k-1)-+n(k-1)-jγm,n(T)Tj+m(k-1)-.

Hence Tk is (m, n)-isosymmetric.

(2) Assume that T is invertible. Since

0=T*-m-nγm,n(T)T-m-n=j=0m(-1)j(mj)T*-m-nT*m-jαn(T)Tm-jT-m-n=j=0m(-1)j(mj)T*-n-jαn(T)T-n-j=j=0m(-1)j(mj)T*-j(T*-nαn(T)T-n)T-j={j=0m(-1)j(mj)T*-j·(αn(T-1))·T-j=γm,n(T-1)(mis even)j=0m(-1)j(mj)T*-j·(-αn(T-1))·T-j=-γm,n(T-1)(mis odd),

it follows that T−1 is (m, n)-isosymmetric.

Operators T and S are said to be doubly commuting if TS = ST and TS* = S*T. From the equation

((y1+y2)(x1+x2)-1)m((y1+y2)-(x1+x2))n=j=0ni+l+h=m(nj)(mi,l,h)(y1+y2)iy2l(y1x1-1)h(y1-x1)n-j(y2-x2)jx1lx2i,

if T and S are doubly commuting, then it holds

γm,n(T+S)=j=0ni+l+h=m(nj)(mi,l,h)·(T*+S*)iS*lγh,n-j(T)αj(S)TlSi.

Theorem 2.5

Let T be (m, n)-isosymmetric and let Q be a nilpotent operator of order k. If T and Q are doubly commuting, then T +Q is (m+2k −2, n+2k −1)- isosymmetric.

Proof

From equation (2.1), it holds

γm+2k-2,n+2k-1(T+Q)=j=0n+2k-1i+l+h=m+2k-2(n+2k-1j)(m+2k-2i,l,h)·(T*+Q*)iQ*lγh,n+2k-1-j(T)αj(Q)TlQi.

(1) If j ≥ 2k or ik or lk, then αj(Q) = 0 or Qi = 0 or Q*l = 0, respectively.

(2) If j ≤ 2k −1 and ik −1 and lk − 1, then h = m+2k − 2 − i − lm and n + 2k − 1 − jn + 2k − 1 (2k − 1) = n, i.e., γh,n+2k−1−j(T) = 0.

By (1) and (2) we have γm+2k−2,n+2k−1(T + Q) = 0. Therefore T + Q is (m + 2k − 2, n + 2k − 1)-isosymmetric.

Note that the equation

(y1y2x1x2-1)m·(y1y2-x1x2)n=k=0mj=0n(mk)(nj)y1j+k(y1x1-1)m-k(y1-x1)n-j(y2x2-1)k(y2-x2)jx1kx2n-j.

From this, if T and S are doubly commuting, then it holds

γm,n(TS)=k=0mj=0n(mk)(nj)T*j+kγm-k,n-j(T)·γk,j(S)TkSn-j.

Theorem 2.6

Let T be (m, n)-isosymmetric and let S be m′-isometric and n′-symmetric. If T and S are doubly commuting, then TS is (m+m′ −1, n+n′ −1)- isosymmetric.

Proof

From equation (2.2), it holds

γm+m-1,n+n-1(TS)=k=0m+m-1j=0n+n-1(n+n-1j)(m+m-1k)T*j+k·γm+m-1-k,n+n-1-j(T)·γk,j(S)·TkSn-j.

(1) If km′ or jn′, then γk,j(S) = 0.

(2) If km′ −1 and jn′ − 1, then m+m′ − 1 − km and n + n′ − 1 − jn, i.e., γm+m′−1−k,n+n′−1−j(T) = 0.

By (1) and (2) we have γm+m′−1,n+n′−1(TS) = 0. Hence it completes the proof.

For a complex Hilbert space ℋ, let ℋ⊗ℋ denote the completion of the algebraic tensor product of ℋ and ℋ endowed a reasonable uniform cross-norm. For operators TB(ℋ) and SB(ℋ), TSB(ℋ ⊗ ℋ) denote the tensor product operator defined by T and S. Note that TS = (TI)(IS) = (IS)(TI).

Theorem 2.7

Let T be (m, n)-isosymmetric and let S be m′-isometric and n′-symmetric. Then TS is (m + m′ − 1, n + n′ − 1)-isosymmetric.

Proof

It is clear that if T is (m, n)-isosymmetric, then TI is (m, n)-isosymmetric and if S is m′-isometric and n′-symmetric, then IS is m′-isometric and n′-symmetric. Since TI and IS are doubly commuting, it follows from Theorem 2.6 that TS is (m + m′ − 1, n + n′ − 1)-isosymmetric. Hence it completes the proof.

3. Conjugation and Example

In this section, we introduce [m,C]-symmetric operators and provide several examples. An antilinear operator C on ℋ is said to be a conjugation if C satisfies C2 = I and 〈Cx,Cy〉= 〈y, x〉 for all x, y ∈ ℋ. An operator TB(ℋ) is said to be complex symmetric if CTC = T* for some conjugation C.

Definition 2

For an operator TB(ℋ) and a conjugation C, we define the operator αm(T;C) by

αm(T;C)=j=0m(-1)j(mj)CTm-jC·Tj.

An operator TB(ℋ) is said to be an [m,C]-symmetric operator if αm(T;C) = 0.

Hence if T is complex symmetric and [m,C]-symmetric, then T is m-symmetric. It holds that

CTC·αm(T;C)-αm(T;C)·T=αm+1(T;C).

Moreover, if T is [m,C]-symmetric, then T is [n,C]-symmetric for every natural number n (≥ m) and ker(αm−1(T;C)) (m ≥ 2) is an invariant subspace for T.

Example 3.1

Let ℋ = ℂ2 and let C be a conjugation on ℋ given by C(xy)=(y¯x¯) for x, y ∈ ℂ.

If T=(i11-i) on C2, then T is not Hermitian and CTC=(i11-i)=T.

Hence T is [1, C]-symmetric.

Hence, in this case, σ(T) = {0} due to the fact that T is nilpotent.

Let S=(i22-i) on C2. Then S is not Hermitian and CSC=(i22-i) and CSC = SS*. Therefore, S is [1, C]-symmetric. Furthermore, σ(S) = {1,−1}.

If R=(112i12i2) on C2, then

CR2C-2CRC·R+R2=(154-32i-32i34)-2(940094)+(3432i32i154)=0.

Hence R is [2, C]-symmetric. It is easy to see that R is not [1, C]-symmetric. Moreover, σ(R)={32}.

Let W=(2i11-2i) on C2. Then it is easy to see that CWC = WW*.

Hence W is [1, C]-symmetric and σ(W)={3i,-3i}.

Example 3.2

Let ℋ = ℓ2, let {en}n=1 be the natural basis of ℋ and let C : ℋ → ℋ be a conjugation given by

C(n=1xnen)=n=1xn¯en

where {xn} is a sequence in C with n=1|xn|2< and Cen = en.

If UB(ℋ) is the unilateral shift on ℓ2, then it is easy to compute U = CUC and so U is a [1, C]-symmetric operator with (unit disk).

Let W be the weighted shift given by Wen = αnen+1, where αn={2i         (n=1)n+1ni(n2).

Then

(CW2C-2CWCW+W2)en=[(αn¯-αn)αn+1¯-αn(αn+1¯-αn+1)]en

for all n ≥ 1. Hence W is [2, C]-symmetric operator.

An operator A ∈ ℒ(ℋ) is n-Jordan if A = T + N where T is self-adjoint, N is nilpotent of order [n+12], and TN = NT where [k] denotes the integer part of k.

Example 3.3

Let C be a conjugation C on ℋ. Suppose that A = T + N is an n-Jordan operator where T = T* = CTC, N is nilpotent of order [n+12], TN = NT, and CN = NC. Then A is [n,C]-symmetric for the conjugation C. Indeed, since T = T* = CTC, TN = NT, and CN = NC, it follows that

j=0n(-1)j(nj)CAn-jC·Aj=j=0n(-1)j(nj)(T+N)n-j·(T+N)j=0

which means that A is an n-symmetric operator from [12, Theorem 3.2]. Hence A is [n,C]-symmetric.

4. [m,C]-symmetric Operators

Let C be a conjugation on ℋ. Then C satisfies ||Cx|| = ||x|| and C(αx) = ᾱ·Cx for all x ∈ ℋ and all α ∈ ℂ. Moreover, since C2 = I, it follows that (CTC)* = CT*C and (CTC)n = CTnC for every positive integer n (see [7] for more details).

We now provide properties of [m,C]-symmetric operators.

Theorem 4.1

Let TB(ℋ) and let C be a conjugation on. Then the following assertions hold;

T is an [m,C]-symmetric operator if and only if so is T*.

If T is an [m,C]-symmetric operator, then Tk is [m,C]-symmetric for any k ∈ ℕ.

If T is an [m,C]-symmetric operator and invertible, then T−1is [m,C]- symmetric.

If T is a [2, C]-symmetric operator, then ker(T) ⊂ ker(T2)∩C(ker(T2)).

Proof

(a) Since T is [m,C]-symmetric, it follows that αm(T;C) = 0. Therefore,

0=C(αm(T;C))*C={αm(T*;C)         (mis even)-αm(T*;C)         (mis odd).

Hence T* is [m,C]-symmetric. The converse implication holds in a similar way. (b) Note that

(ak-bk)m=((a-b)(ak-1+ak-2b++bk-1))m=j=0m(k-1)λjam(k-1)-j(a-b)mbj,

where λj are some coefficients (j = 0, ...,m(k − 1)). This implies that

αm(Tk;C)=j=0m(k-1)λjCTm(k-1)-jC·αm(T;C)·Tj=0.

Hence Tk is [m,C]-symmetric.

(c) Since T is [m,C]-symmetric, it follows that αm(T;C) = 0 and therefore

0=CT-mC·αm(T;C)·T-m={αm(T-1;C)         (mis even)-αm(T-1;C)         (mis odd).

Hence T−1 is [m,C]-symmetric.

(d) It is clear ker(T) ⊂ ker(T2). If T is [2, C]-symmetric and x ∈ ker(T), then

CT2Cx=2CTCTx-T2x=0

and hence T2Cx = 0. Thus Cx ∈ ker(T2) and so xC(ker(T2)). Hence we get ker(T) ⊂ ker(T2)∩C(ker(T2)).

Lemma 4.2

For TB(ℋ), a conjugation C, and two complex numbers λ, μ, it holds

αm(T;C)=n1+n2+n3=m(-1)n2(mn1,n2,n3)(CTC-λI)n1(T-μI)n2(λ-μ)n3.

In particular, for λ ∈ ℂ we have

αm(T;C)=j=0m(-1)j(mj)(CTC-λI)m-j(T-λI)j.
Proof

Using the multinomial formula, it holds

αm(T;C)=(y-x)m|y=CTC,x=T=([y-λ]-[x-μ]+[λ-μ])m|y=CTC,x=T=n1+n2+n3=m(-1)n2(mn1,n2,n3)(y-λ)n1(x-μ)n2(λ-μ)n3|y=CTC,x=T=n1+n2+n3=m(-1)n2(mn1,n2,n3)(CTC-λI)n1(T-μI)n2(λ-μ)n3.

Equation (4.1) follows from λ = μ in the first formula.

By Lemma 2.7 of [3], for TB(ℋ) and two complex numbers λ, μ, it holds

βm(T)=n1+n2+n3=m(-1)n2(mn1,n2,n3)(T*-μ¯I)n1Tn1μ¯n2(T-λI)n2(λμ¯-1)n3.

We investigate properties of spectra of [m,C]-symmetric operators. In [11], S. Jung, E. Ko and J. E. Lee proved the following result.

Proposition 4.3.([11, Lemma 3.21])

If C is a conjugation onand TB(ℋ), then σ(CTC) = σ(T)*, σp(CTC) = σp(T)*, σa(CTC) = σa(T)* and σs(CTC) = σs(T)*.

Theorem 4.4

Let TB(ℋ) be an [m,C]-symmetric operator where C is a conjugation on. Then σp(T) = σp(T)*, σa(T) = σa(T)*, σs(T) = σs(T)* and σ(T) = σ(T)*.

Proof

Let zσa(T). Then there exists a sequence {xn} of unit vectors such that (Tz)xn → 0 (n→∞). By equation (4.1) it holds

αm(T;C)=j=0m(-1)j(mj)(CTC-zI)m-j(T-zI)j.

Hence we have 0 = limn→∞αm(T;C)xn = limn→∞ (CTCz)m xn. Therefore, it is easy to see zσa(CTC). Hence σa(T) ⊂ σa(CTC). Since σa(CTC) = σa(T)* by Proposition 4.3, this means σa(T) ⊂ σa(T)*. Hence σa(T)*σa(T)** = σa(T) and so σa(T) = σa(T)*.

Since T* is also an [m,C]-symmetric operator by Theorem 4.1, we have σa(T*) = σa(T*)*. Hence σs(T) = σs(T)* and σ(T) = σa(T) ∪ σs(T) = σa(T)*σs(T)* = σ(T)*. From the above proof it is clear that σp(T) = σp(T)*.

For T,SB(ℋ), a pair (T,S) of operators is said to be a C-doubly commuting pair if TS = ST and CSC · T = T · CSC for a conjugation C.

Lemma 4.5

Let (T,S) be a C-doubly commuting pair where C is a conjugation on. Then it holds

αm(T+S;C)=j=0m(mj)αj(T;C)·αm-j(S;C).
Proof

From the assumption, it holds T ·CSjC = CSjC ·T and S·CTjC = CTjC ·S for every j ∈ ℕ. It is clear that equation (4.2) holds for m = 1. Assume that equation (4.2) holds for m. Then by (3.1) we have

αm+1(T+S;C)=C(T+S)C·αm(T+S;C)-αm(T+S;C)·(T+S)=j=0m(mj)(CTC+CSC)·αj(T;C)·αm-j(S;C)-j=0m(mj)αj(T;C)·αm-j(S;C)·(T+S)=j=0m(mj)(CTC·αj(T;C)-αj(T;C)·T)αm-j(S;C)+j=0m(mj)αj(T;C)(CSC·αm-j(S;C)-αm-j(S;C)·S)=j=0m(mj)αj+1(T;C)·αm-j(S;C)+j=0m(mj)αj(T;C)·αm+1-j(S;C)=j=0m+1(m+1j)αj(T;C)·αm+1-j(S;C).

Hence equation (4.2) holds for any m ∈ ℕ.

Therefore we have the following theorem.

Theorem 4.6

Let TB(ℋ) be an [m,C]-symmetric operator and let SB(ℋ) be an [n,C]-symmetric operator where C is a conjugation on. If (T,S) is a C-doubly commuting pair, then T + S is an [m + n − 1, C]-symmetric operator.

Proof

By Lemma 4.5, it holds

αm+n-1(T+S;C)=j=0m+n-1(m+n-1j)αj(T;C)·αm+n-1-j(S;C).

(i) If 0 ≤ jm − 1, then m+ n − 1 − jm+ n − 1 − (m − 1) = n. Therefore we have

αm+n−1−j(S;C) = 0.

(ii) If jm, then αj(T;C) = 0.

Hence we get αm+n−1(T + S;C) = 0 and so T + S is [m+ n − 1, C]-symmetric.

Theorem 4.7

Let C be a conjugation on. If Q is a nilpotent operator of order n, then Q is a [2n − 1, C]-symmetric operator.

Proof

It holds

α2n-1(Q;C)=j=02n-1(-1)j(2n-1j)CQ2n-1-jC·Qj.

(i) If 0 ≤ jn − 1, then 2n − 1 − j ≥ 2n − 1 − (n − 1) = n. Hence Q2n−1−j = 0.

(ii) If jn, then Qj = 0.

Therefore α2n−1(Q;C) = 0 and hence Q is [2n − 1, C]-symmetric.

Corollary 4.8

Let TB(ℋ) be an [m,C]-symmetric operator and let QB(ℋ) be a nilpotent operator of order n where C is a conjugation on. If (T,Q) is a C-doubly commuting pair, then T + Q is an [m+2n − 2, C]-symmetric operator.

Proof

The proof follows from Theorems 4.6 and 4.7.

Example 4.9

Let Cn be the conjugation on Cn defined by

Cn(z1,z2,,zn):=(z1¯,z2¯,,zn¯).

Assume that Rn is an n × n matrix as follows;

Rn=aIn+Qn=(a0000a000000000a)+(0b0000b0000b0000)

for a, bC. Since Qn is nilpotent of order n, it follows that Corollary 4.8 that Rn is a [2n − 1, Cn]-symmetric operator.

Lemma 4.10

If (T,S) is a C-doubly commuting pair where C is a conjugation on ℋ, then it holds

αm(TS;C)=j=0m(mj)αj(T;C)·Tm-j·CSjC·αm-j(S;C).
Proof

It is easy to see that equation (4.3) holds for m = 1. Assume that equation (4.3) holds for m. Then by (3.1) we obtain

αm+1(TS;C)=(CTSC)·αm(TS;C)-αm(TS;C)·TS=CTC·CSCj=0m(mj)αj(T;C)·Tm-j·CSjC·αm-j(S;C)-j=0m(mj)αj(T;C)·Tm+1-j·CSjC·αm-j(S;C)·S=j=0m(mj)(CTC·αj(T;C)-αj(T;C)·T)Tm-j·CSj+1C·αm-j(S;C)+j=0m(mj)αj(T;C)·Tm+1-j·CSjC·(CSC·αm-j(S;C)-αm-j(S;C)·S)=j=0m(mj)αj+1(T;C)·Tm-j·CSj+1C·αm-j(S;C)+j=0m(mj)αj(T;C)·Tm+1-j·CSjC·αm+1-j(S;C)=j=0m+1(m+1j)αj(T;C)·Tm+1-j·CSjC·αm+1-j(S;C).

Hence equation (4.3) holds for any m ∈ ℕ.

Theorem 4.11

Let T be an [m,C]-symmetric operator and let S be an [n,C]- symmetric operator where C is a conjugation on. If (T,S) is a C-doubly commuting pair, then TS is an [m + n − 1, C]-symmetric operator.

Proof

Since (T,S) is a C-doubly commuting pair, it follows from equation (4.3) that

αm+n-1(TS;C)=j=0m+n-1(m+n-1j)αj(T;C)·Tm+n-1-j·CSjC·αm+n-1-j(S;C).

(i) If 0 ≤ jm − 1, then m+ n − 1 − jm+ n − 1 − (m − 1) = n. Therefore we get αm+n−1−j(S;C) = 0.

(ii) If mj, then αj(T;C) = 0.

Therefore αm+n−1(TS;C) = 0. Hence TS is [m + n − 1, C]-symmetric.

Corollary 4.12

Let C be a conjugtion on. If T = T1I and S = IS1where T1and S1are [m,C]-symmetric, then TS is [2m − 1, C]-symmetric.

Proof

Since T1 and S1 are [m,C]-symmetric, it follows that T = T1I and S = IS1 are [m,C]-symmetric. In addition, we know that (T,S) is a C-doubly commuting pair. Therefore, TS is [2m − 1, C]-symmetric by Theorem 4.11.

In [6], B. Duggal proved the following proposition.

Proposition 4.13

Let T and S be an m-isometric operator and an n-isometric operator, respectively. Then TS is an (m + n − 1)-isometric operator.

Similarly, we show the following result.

Theorem 4.14

Let T be an [m,C]-symmetric operator and let S be an [n,D]- symmetric operator where C and D are conjugations on. Then TS is an [m + n − 1, CD]-symmetric operator on ℋ ⊗ ℋ.

Proof

Since C and D are conjugations on ℋ, it follows from [4] that CD is a conjugation on ℋ ⊗ ℋ. If T is [m,C]-symmetric and S is [n,D]-symmetric, it is easy to see that TI is [m,CD]-symmetric and IS is [n,CD]-symmetric on ℋ ⊗ ℋ, respectively. Also it is clear that (TI, IS) is a CD-doubly commuting pair. Since TS = (TI)(IS), it follows from Theorem 4.11 that (TI)(IS) = TS is [m + n − 1, CD]-symmetric.

Acknowledgements

This paper is dedicated to the memory of Professor Takayuki Furuta with cordial appreciation. This is partially supported by Grant-in-Aid Scientific Research No.15K04910. The second author was supported by Basic Science Research Program through the National Research Foundation of Korea(NRF) funded by the Ministry of Education, Science and Technology(2016R1A2B4007035).

References
  1. Agler, J, and Stankus, M (1995). m-isometric transformations of Hilbert space I. Intgr Equat Oper Theory. 21, 383-429.
    CrossRef
  2. Bermúdes, T, Martinón, A, Müller, V, and Noda, AJ (2014). Perturbation of m-isometries and nilpotent operators. Abstr Appl Anal.
    CrossRef
  3. Chō, M, Gu, C, and Lee, WY (2016). Elementary properties of ∞-isometries on a Hilbert space. Linear Algebra Appl. 511, 378-402.
    CrossRef
  4. Chō, M, Lee, JE, and Motoyoshi, H (2017). On [m,C]-isometric operators. Filomat. 31, 2073-2080.
    CrossRef
  5. Chō, M, Ôta, S, and Tanahashi, K (2013). Invertible weighted shift operators which are misometries. Proc Amer Math Soc. 141, 4241-4247.
    CrossRef
  6. Duggal, B (2012). Tensor product of n-isometries. Linear Algebra Appl. 437, 307-318.
    CrossRef
  7. Garcia, SR, and Putinar, M (2006). Complex symmetric operators and applications. Trans Amer Math Soc. 358, 1285-1315.
    CrossRef
  8. Gu, C, and Stankus, M (2015). Some results on higher order isometries and symmetries: Products and sums a nilpotent operator. Linear Algebra Appl. 469, 500-509.
    CrossRef
  9. Helton, JW . Operators with a representation as multiplication by x on a Sobolev space., Proc. Internat. Conference on Operator Theory, 1970, Hungary.
  10. Helton, JW (1972). Infinite dimensional Jordan operators and Sturm-Liouville conjugate point theory. Trans Amer Math Soc. 170, 305-331.
    CrossRef
  11. Jung, S, Ko, E, and Lee, JE (2013). On complex symmetric operator matrices. J Math Anal Appl. 406, 373-385.
    CrossRef
  12. McCullough, SA, and Rodman, L (1997). Hereditary classes of operators and matrices. Amer Math Monthly. 104, 415-430.
    CrossRef
  13. Stankus, M (2013). m-Isometries, n-symmetries and other linear transformations which are hereditary roots. Integr Equat Oper Theory. 75, 301-321.
    CrossRef


This Article

Social Network Service

e-submission

Archives

Indexed/Covered by