Next Article in Journal
Edge Irregular Reflexive Labeling for Disjoint Union of Generalized Petersen Graph

Article

# New Sufficient Condition for the Positive Definiteness of Fourth Order Tensors

School of mathematics, Zunyi Normal College, Zunyi, Guizhou 563006, China
*
Author to whom correspondence should be addressed.
Mathematics 2018, 6(12), 303; https://doi.org/10.3390/math6120303
Received: 30 September 2018 / Revised: 22 November 2018 / Accepted: 30 November 2018 / Published: 5 December 2018

## Abstract

In this paper, we give a new Z-eigenvalue localization set for Z-eigenvalues of structured fourth order tensors. As applications, a sharper upper bound for the Z-spectral radius of weakly symmetric nonnegative fourth order tensors is obtained and a new Z-eigenvalue based sufficient condition for the positive definiteness of fourth order tensors is also presented. Finally, numerical examples are given to verify the efficiency of our results.

## 1. Introduction

Let $A = ( a i 1 i 2 ⋯ i m ) ∈ R [ m , n ]$ be an m-th order n dimensional real square tensor, x be a real n-vector. Then, let $N = { 1 , 2 , … , n }$, we define the following real n-vector:
$A x m − 1 = ∑ i 2 , ⋯ , i m = 1 n a i i 2 … i m x i 2 … x i m i ∈ N , x [ m − 1 ] = ( x i m − 1 ) i ∈ N .$
If there exists a real vector x and a real number $λ$ such that
$A x m − 1 = λ x [ m − 1 ] ,$
then $λ$ is called H-eigenvalue of $A$ and x is called H-eigenvector of $A$ associated with $λ$. If there exists a real vector x and a real number $λ$ such that
$A x m − 1 = λ x , x T x = 1 .$
then $λ$ is called Z-eigenvalue of $A$ and x is called Z-eigenvector of $A$ associated with $λ$ [1,2].
An mth-degree homogeneous polynomial form of n variables
$f ( x ) = A x m = ∑ i 2 , ⋯ , i m = 1 n a i 1 … i m x i 1 … x i m$
is positive definite, i.e., $f ( x ) > 0$, if and only if the real symmetric tensor $A$ is positive definite [2]. When m is even, an eigenvalue method is given to verify the positive definiteness of $A$.
Theorem 1
([2]). Let $A$ be an even-order real symmetric tensor. Then
(1) $A$ is positive definite if and only if all of its H-eigenvalues are positive;
(2) $A$ is positive definite if and only if all of its Z-eigenvalues are positive.
From Theorem 1, we can verify the positive definiteness of $A$ by the H-eigenvalues or the Z-eigenvalues of $A$. But when m and n are large, it is difficult to compute all the H-eigenvalues (or Z-eigenvalues) or the smallest H-eigenvalue (or Z-eigenvalue) of an order m dimension n real tensor $A$. Based on the Geršhgorin-type theorem for H-eigenvalues, which is introduced in [2], Li et al. provided some sufficient conditions for the positive definiteness of an even-order real symmetric tensor [3], and some improved results are obtained in [4,5,6,7,8].
First, let us recall the definitions of strictly diagonally dominant (SDD) tensors and quasi-doubly SDD (QDSDD) tensors [7].
Definition 1.
A tensor $A = ( a i 1 … i m ) ∈ R [ m , n ]$ is called a strictly diagonally dominant (SDD) tensor if for $i ∈ N$,
$| a i … i | > R i ( A ) .$
Definition 2.
A tensor $A = ( a i 1 … i m ) ∈ R [ m , n ]$ is called a quasi-doubly strictly diagonally dominant (QDSDD) tensor if for $i , j ∈ N , i ≠ j$,
$( | a i … i | − R i j ( A ) ) | a j … j | > | a i j … j | R j ( A ) .$
The following useful theorem is given in [7].
Theorem 2
([7]). Let $A$ be an even-order real symmetric tensor with all positive diagonal entries.
(1) If $A$ is strictly diagonally dominant, then $A$ is positive definite;
(2) If $A$ is quasi-doubly strictly diagonally dominant, then $A$ is positive definite.
Positive definiteness of fourth order tensors has important applications in signal processing, automatic control, and magnetic resonance imaging [9,10,11,12]. Recently, in order to preserve positive definiteness for a fourth order tensor, a ternary quartics approach is proposed in [13]. Extending the Riemannian framework from 2nd order tensors to the space of 4th order tensors, a riemannian approach is given to guarantee positive definiteness for a fourth order tensor [14]. In [11], the authors explain the definition of the smallest Z-eigenvalue and present a computational method for calculating it. Very recently, much literature has focused on the properties of Z-eigenvalues of tensors [15,16,17,18,19,20,21,22,23,24], but there are no Z-eigenvalues based sufficient conditions for the positive definiteness of an even-order real symmetric tensor.
In this paper, based on the Z-eigenvalue localization sets of structured fourth order tensors, a new sufficient condition for the positive definiteness of fourth order tensors is given.

## 2. New Z-Eigenvalue Localization Set for Structured Fourth Order Tensors

In this section, a Geršhgorin-type theorem for Z-eigenvalues of structured fourth order tensors is obtained. For any $k ∈ N$, let
$Δ k = { ( i 2 i 3 i 4 ) : there are at least two i h = k for h = 2 , 3 , 4 } ,$
$Δ k ¯ = { ( i 2 , i 3 , i 4 ) : there are at most one i h = k for h = 2 , 3 , 4 } ,$
then,
$R i ( A ) = ∑ i 2 , i 3 , i 4 ∈ N | a i i 2 i 3 i 4 | = r i Δ k ¯ ( A ) + r i Δ k ( A ) ,$
where
$r i Δ k ( A ) = ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k | a i i 2 i 3 i 4 | , r i Δ k ¯ ( A ) = ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k ¯ | a i i 2 i 3 i 4 | ,$
where we assume that
$Δ k t = { ( i 2 i 3 i 4 ) : ( i 2 i 3 i 4 ) ∈ Δ k and i h = t for some h = 2 , 3 , 4 } ,$
$Δ k t ¯ = { ( i 2 i 3 i 4 ) : ( i 2 i 3 i 4 ) ∈ Δ k and i h ≠ t for any h = 2 , 3 , 4 } ,$
$Δ k ¯ t = { ( i 2 , i 3 , i 4 ) : ( i 2 i 3 i 4 ) ∈ Δ k ¯ and i h = t for some h = 2 , 3 , 4 } ,$
and
$Δ k ¯ t ¯ = { ( i 2 , i 3 , i 4 ) : ( i 2 i 3 i 4 ) ∈ Δ k ¯ and i h ≠ t for any h = 2 , 3 , 4 } .$
We give our main results in this section as follows.
Theorem 3.
Let $A = ( a i 1 ⋯ i m ) ∈ R [ 4 , n ]$ with
$β i Δ 1 i ( A ) = … = β i Δ n i ( A ) = C i ( c o n s t a n t ) , i ∈ N .$
Then
$σ ( A ) ⊆ Υ ( A ) = ⋃ i , j ∈ N , i ≠ j Υ i j ( A ) ,$
where
$Υ i j ( A ) = { z ∈ R : | z − C i | − r i Δ k ¯ i ( A ) | z − C j | ≤ β i Δ k i ¯ ( A ) + r i Δ k ¯ i ¯ ( A ) β j Δ k j ¯ ( A ) + r j Δ k ¯ j ¯ ( A ) + r j Δ k ¯ j ( A ) } ,$
and
$β i Δ k i ( A ) = ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k i | a i i 2 i 3 i 4 | = C i , r i Δ k ¯ i ( A ) = ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k ¯ i | a i i 2 i 3 i 4 | ,$
$β i Δ k i ¯ ( A ) = max k ∈ N ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k i ¯ | a i i 2 i 3 i 4 | , r i Δ k ¯ i ¯ ( A ) = ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k ¯ i ¯ | a i i 2 i 3 i 4 | ,$
$β j Δ k j ( A ) = ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k j | a j i 2 i 3 i 4 | = C j , r j Δ k ¯ j ( A ) = ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k ¯ j | a j i 2 i 3 i 4 | ,$
$β j Δ k j ¯ ( A ) = max k ∈ N ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k j ¯ | a j i 2 i 3 i 4 | , r j Δ k ¯ j ¯ ( A ) = ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k ¯ j ¯ | a j i 2 i 3 i 4 | .$
Proof.
Let $λ$ be a Z-eigenvalue of $A$ with corresponding Z-eigenvector $x = ( x 1 , ⋯ , x n ) T ∈ R n \ { 0 }$, i.e.,
$A x 3 = λ x , and x T x = 1 .$
Let $| x t | ≥ | x s | ≥ max i ∈ N , i ≠ t , s | x i |$, then for any $k ∈ N$, we have
$λ − ( ∑ ( i 2 , i 3 , i 4 ) ∈ Δ 1 t a t i 2 i 3 i 4 x 1 2 + … + ∑ ( i 2 , i 3 , i 4 ) ∈ Δ n t a t i 2 i 3 i 4 x n 2 ) x t = ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k t ¯ a t i 2 i 3 i 4 x i 2 x i 3 x i 4 + ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k ¯ t a t i 2 i 3 i 4 x i 2 x i 3 x i 4 + ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k ¯ t ¯ a t i 2 i 3 i 4 x i 2 x i 3 x i 4 .$
Taking modulus in the above equation, and using the triangle inequality and $x T x = 1$, we get
$| λ − C t | | x t | ≤ ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k t ¯ | a t i 2 i 3 i 4 | | x i 2 | | x i 3 | | x i 4 | + ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k ¯ | a t i 2 i 3 i 4 | | x i 2 | | x i 3 | | x i 4 | ≤ ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k t ¯ | a t i 2 i 3 i 4 | | x k | 2 | x t | + ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k ¯ t | a t i 2 i 3 i 4 | | x s | + ∑ ( i 2 , i 3 , i 4 ) ∈ Δ k ¯ t ¯ | a t i 2 i 3 i 4 | | x s | .$
Therefore,
$| λ − C t | − r t Δ k ¯ t ( A ) | x t | ≤ β t Δ k t ¯ ( A ) + r t Δ k ¯ t ¯ ( A ) | x s | .$
If $| x s | = 0$, then $| λ − C t | − r t Δ k ¯ t ( A ) ≤ 0$, and it is obvious that $λ ∈ Υ ( A )$.
If $| x s | > 0$, from equality (2), we similarly get
$| λ − C s | | x s | ≤ β s Δ k s ¯ ( A ) + r s Δ k ¯ s ¯ ( A ) + r s Δ k ¯ s ( A ) | x t | .$
Multiplying inequalities (3) with (4), we have
$| λ − C t | − r t Δ k ¯ t ( A ) | λ − C s | ≤ β t Δ k t ¯ ( A ) + r t Δ k ¯ t ¯ ( A ) β s Δ k s ¯ ( A ) + r s Δ k ¯ s ¯ ( A ) + r s Δ k ¯ s ( A ) .$
Thus, we complete the proof. □

## 3. Upper Bound for the Z-Spectral Radius of Weakly Symmetric Nonnegative Tensors

In this section, we obtain a sharp upper bound for weakly symmetric nonnegative tensors. Firstly, let us recall the definition of the Z-spectral radius of tensor $A$.
Definition 3
([2]). Let $A ∈ R [ m , n ]$. The Z-spectral radius $ρ ( A )$ of $A$ is defined as
$ρ ( A ) : = sup { | λ | : λ ∈ σ ( A ) } ,$
where $σ ( A )$, called the Z-spectrum of $A$, is the set of all Z-eigenvalues of $A$.
A tensor $A$ is called weakly symmetric if the associated homogeneous polynomial $A x m$ satisfies
$∇ A x m = m A x m − 1 .$
We need the following Perron-Frobenius Theorem for the Z-eigenvalue of nonnegative tensors [23].
Lemma 1.
Suppose that the m-order n-dimensional tensor $A$ is weakly symmetric, nonnegative and irreducible. Then $ρ ( A )$ is a positive Z-eigenvalue with a positive Z-eigenvector.
Based on the above Lemma, we give the main result of this section.
Theorem 4.
Let $A = ( a i 1 ⋯ i m ) ∈ R [ 4 , n ]$ be weakly symmetric, nonnegative and irreducible and
$max i ∈ N { β i Δ 1 i ( A ) , … , β i Δ n i ( A ) } = C i ( c o n s t a n t ) , i ∈ N .$
Then
$ρ ( A ) ≤ max { max i , j ∈ N , i ≠ j 1 2 C i + r i Δ k ¯ i ( A ) + C j + Λ i j 1 2 ( A ) , max i ∈ N C i } ,$
where
$Λ i j ( A ) = C i + r i Δ k ¯ i ( A ) − C j 2 + 4 β i Δ k i ¯ ( A ) + r i Δ k ¯ i ¯ ( A ) β j Δ k j ¯ ( A ) + r j Δ k ¯ j ¯ ( A ) + r j Δ k ¯ j ( A ) .$
Proof.
Let $x > 0$ be a Z-eigenvector of $A$ corresponding to $ρ ( A )$, if $ρ ( A ) ≥ max i ∈ N C i$, then from the proof of Theorem 3, there exist $t , s ∈ N$, $s ≠ t$ such that
$λ − C t − r t Δ k ¯ t ( A ) ( λ − C s ) ≤ β t Δ k t ¯ ( A ) + r t Δ k ¯ t ¯ ( A ) β s Δ k s ¯ ( A ) + r s Δ k ¯ s ¯ ( A ) + r s Δ k ¯ s ( A ) .$
Then, solving for $ρ ( A )$ we get
$ρ ( A ) ≤ 1 2 C t + r t Δ k ¯ t ( A ) + C s + Λ t s 1 2 ( A ) .$
Thus, we complete the proof. □

## 4. Z-Eigenvalue Based Sufficient Condition for the Positive Definiteness of Fourth Order Tensors

In this section, we provide a new checkable sufficient condition for the positive definiteness of fourth order tensors, which is based on the inclusion set for Z-eigenvalues of structured fourth order tensors.
Theorem 5.
Let $A ∈ R [ 4 , n ]$ with $β i Δ 1 i ( A ) = … = β i Δ n i ( A ) = C i > 0$ be a symmetric tensor. If for all $i , j ∈ N$, $j ≠ i$,
$( C i − r i Δ k ¯ i ( A ) ) C j > ( β i Δ k i ¯ ( A ) + r i Δ k ¯ i ¯ ( A ) ) ( β j Δ k j ¯ ( A ) + r j Δ k ¯ j ¯ ( A ) + r j Δ k ¯ j ( A ) ) ,$
then $A$ is positive definite.
Proof.
Assume that $λ ≤ 0$ is a Z-eigenvalue of $A$. From Theorem 3, we have $λ ∈ Υ ( A )$, hence, there are $i 0 , j 0 ∈ N$ such that
$| λ − C i 0 | − r i 0 Δ k ¯ i 0 ( A ) | λ − C j 0 |$
$≤ β i 0 Δ k i 0 ¯ ( A ) + r i 0 Δ k ¯ i 0 ¯ ( A ) β j 0 Δ k j 0 ¯ ( A ) + r j 0 Δ k ¯ j 0 ¯ ( A ) + r j 0 Δ k ¯ j 0 ( A ) .$
From $C i 0 > 0$ for all $i 0 ∈ N$, we get
$| λ − C i 0 | − r i 0 Δ k ¯ i 0 ( A ) | λ − C j 0 | ≥ C i 0 − r i 0 Δ k ¯ i 0 ( A ) C j 0 > β i 0 Δ k i 0 ¯ ( A ) + r i 0 Δ k ¯ i 0 ¯ ( A ) β j 0 Δ k j 0 ¯ ( A ) + r j 0 Δ k ¯ j 0 ¯ ( A ) + r j 0 Δ k ¯ j 0 ( A ) .$
This is a contradiction. Hence, $λ > 0$. Then, the symmetric tensor $A$ is positive definite. □
Theorem 6.
Let $A ∈ R [ 4 , n ]$ with $β i Δ 1 i ( A ) > 0 , … , β i Δ n i ( A ) > 0$ be a symmetric tensor,
$min i ∈ N { β i Δ 1 i ( A ) , … , β i Δ n i ( A ) } = C i .$
Assume $B$ is a symmetric tensor whose $( i 1 i 2 i 3 i 4 )$-th entry is respectively defined as follows:
$B i 1 i 2 i 3 i 4 = b i i j j = b i j i j = b i j j i = 1 3 C i , i ≠ j , b i i i i = C i , i ∈ N , a i 1 i 2 i 3 i 4 , otherwise ,$
If $B$ is positive definite, then $A$ is positive definite.
Proof.
Let $x ∈ R n$ be a nonzero vector. Since $B$ is positive definite, from the definition of the positive definiteness of symmetric tensors, we have
$B x 4 > 0 .$
Then, we have
$0 < A x 4 = B x 4 + ∑ i , j ∈ N ( a i i j j − b i i j j ) x i 2 x j 2 + ∑ i , j ∈ N ( a i j i j − b i j i j ) x i 2 x j 2 + ∑ i , j ∈ N ( a i j j i − b i j j i ) x i 2 x j 2 .$
Thus $A$ is positive definite. □
By Theorems 5 and 6, we have the following sufficient condition for the positive definiteness of symmetric fourth order tensors.
Theorem 7.
Let $A ∈ R [ 4 , n ]$ with $β i Δ 1 i ( A ) > 0 , … , β i Δ n i ( A ) > 0$ be a symmetric tensor,
$min i ∈ N { β i Δ 1 i ( A ) , … , β i Δ n i ( A ) } = C i .$
If for all $i , j ∈ N$, $j ≠ i$,
$( C i − r i Δ k ¯ i ( A ) ) C j > ( β i Δ k i ¯ ( A ) + r i Δ k ¯ i ¯ ( A ) ) ( β j Δ k j ¯ ( A ) + r j Δ k ¯ j ¯ ( A ) + r j Δ k ¯ j ( A ) ) ,$
then $A$ is positive definite.
Based on the above theorem, we introduce the definition of Z-eigenvalue based quasi-doubly strictly diagonally dominated(Z-QDSDD) symmetric fourth order tensors.
Definition 4.
Let $A ∈ R [ 4 , n ]$ with $β i Δ 1 i ( A ) > 0 , … , β i Δ n i ( A ) > 0$ be a symmetric tensor,
$min i ∈ N { β i Δ 1 i ( A ) , … , β i Δ n i ( A ) } = C i .$
Then, the fourth order tensor $A$ is called Z-eigenvalue based quasi-doubly strictly diagonally dominated(Z-QDSDD), if for all $i , j ∈ N$, $j ≠ i$,
$( C i − r i Δ k ¯ i ( A ) ) C j > ( β i Δ k i ¯ ( A ) + r i Δ k ¯ i ¯ ( A ) ) ( β j Δ k j ¯ ( A ) + r j Δ k ¯ j ¯ ( A ) + r j Δ k ¯ j ( A ) ) ,$

## 5. Numerical Examples

In this section, some examples are given to show the efficiency of our results. First, an example is given to show the efficiency of the result in Theorem 3.
Example 1.
Consider the tensor $A = ( a i 1 i 2 i 3 i 4 )$ of order 4 dimension 2 with entries defined as follows:
$a 1111 = a 1122 = 1 , a 1211 = a 1222 = − 1 , a 2211 = a 2222 = 2 , a 2111 = a 2122 = − 2 ,$
and other $a i 1 i 2 i 3 i 4 = 0$. By computation, we get that, $σ ( A ) = { 0 , 3 }$.
By Theorem 3.3 of [15], we have
$L ( A ) = { z ∈ R : | z | ≤ 5 } .$
By Theorem 5 of [24], we have
$K ( A ) = { z ∈ R : | z | ≤ 6.5615 } .$
By Theorem 3,
$β 1 Δ k 1 ( A ) = C 1 = 1 ,$
$β 1 Δ k 1 ¯ ( A ) = max { | a 1211 | , | a 1222 | } = 1 ,$
$β 2 Δ k 2 ( A ) = C 2 = 2 ,$
$β 2 Δ k 2 ¯ ( A ) = max { | a 2111 | , | a 2122 | } = 2 ,$
$r 1 Δ k ¯ 1 ( A ) = r 2 Δ k ¯ 1 ( A ) = 0 ,$
$r 1 Δ k ¯ 1 ¯ ( A ) = r 2 Δ k ¯ 1 ¯ ( A ) = 0 ,$
then we have
$Υ ( A ) = { z ∈ R : | z − 1 | | z − 2 | ≤ 2 } .$
The Z-eigenvalue inclusion sets $Υ ( A )$ and the exact Z-eigenvalues are drawn in Figure 1. We can see that, $Υ ( A )$ can capture all Z-eigenvalues of $A$, and the Z-eigenvalue inclusion set $Υ ( A )$ is located on the right side of the coordinate axis, which is better than the Z-eigenvalue inclusion sets $K ( A )$ and $L ( A )$.
We now show the efficiency of the new upper bound in Theorem 4 by the following example.
Example 2.
Let $A = ( a i 1 i 2 i 3 i 4 ) ∈ R [ 4 , 2 ]$ be a symmetric tensor defined as follows:
$a 1111 = 0 , a 1211 = a 1121 = a 1112 = a 2111 = 1 , a 2211 = a 2121 = a 2112 = a 1122 = a 1212 = a 1221 = 1 , a 2222 = 4 ,$
while the other $a i 1 i 2 i 3 i 4 = 0$. By Theorem 4.6 of [15], we have
$ρ ( A ) ≤ 7 .$
By Theorem 7 of [24], we have
$ρ ( A ) ≤ 7 .$
By Theorem 4,
$C 1 = 3 , C 2 = 4$
$r 1 Δ k ¯ 1 ( A ) = β 1 Δ k 1 ¯ ( A ) = r 1 Δ k ¯ 1 ¯ ( A ) = 0 ,$
and
$r 2 Δ k ¯ 2 ( A ) = β 1 Δ k 1 ¯ ( A ) = r 1 Δ k ¯ 1 ¯ ( A ) = r 1 Δ k ¯ 1 ( A ) = 0 ,$
then we have
$ρ ( A ) ≤ 4 .$
In fact, $ρ ( A ) = 4$. Hence, the bound in Theorem 4 is sharper and could reach the true value of $ρ ( A )$ in some cases.
Finally, we now show the efficiency of result in Theorem 7 by the following example.
Example 3.
Let $A = ( a i 1 i 2 i 3 i 4 ) ∈ R [ 4 , 2 ]$ be a symmetric tensor defined as follows:
$a 1111 = 1 , a 2222 = 2 , a 1112 = a 1121 = a 1211 = a 2111 = 0.6 , a 2221 = a 2212 = a 2122 = a 1222 = 1 , a 1122 = a 1221 = a 1212 = 10 , a 2211 = a 2121 = a 2112 = 20 .$
By computation, we get that,
$a 1111 = 1 < R 1 ( A ) = 32.8 , a 2222 = 2 < R 2 ( A ) = 63.6 .$
Hence, $A$ is not a SDD tensor. Then, we cannot use Theorem 2 (1) to determine the positiveness of $A$.
We can get
$( | a 1111 | − R 1 2 ( A ) ) | a 2222 | = − 63.6 < | a 1222 | R 2 ( A ) = 63.6 ,$
$( | a 2222 | − R 2 1 ( A ) ) | a 1111 | = − 61 < | a 2111 | R 1 ( A ) = 19.68 .$
Hence, $A$ is not a QSDD tensor. Then, we cannot use Theorem 2 (2) to determine the positiveness of $A$.
However, it is easy to find
$C 1 = 1 , C 2 = 2 ,$
and
$( C 1 − r 1 Δ k ¯ 1 ( A ) ) C 2 = 2 > ( β 1 Δ k 1 ¯ ( A ) + r 1 Δ k ¯ 1 ¯ ( A ) ) ( β 2 Δ k 2 ¯ ( A ) + r 2 Δ k ¯ 2 ¯ ( A ) + r 2 Δ k ¯ 2 ( A ) ) = 1.8 .$
In other words, $A$ satisfies all the conditions of Theorem 7, i.e., $A$ is a Z-QDSDD tensor. Hence, from Theorem 7, $A$ is a positive definite tensor. In fact,
$σ ( A ) = { 0.9908 , 1.9669 , 19.1249 , 22.2080 } .$
From the definition of positive definite tensors, $A$ is positive definite.

## 6. Conclusions

In this paper, focused the fourth order tensors, a new Z-eigenvalue localization set for Z-eigenvalues of structured fourth order tensors is given. As an application, a sharper upper bound for the Z-spectral radius of weakly symmetric nonnegative fourth order tensors is obtained and a Z-eigenvalue based sufficient condition for the positive definiteness of structured fourth order tensors is also given. A positive definite diffusion tensor is a convex optimization problem with a convex quadratic objective function constrained by the nonnegativity requirement on the smallest Z-eigenvalue of the diffusivity function [11], but it is difficult to compute all the Z-eigenvalues or the smallest Z-eigenvalue of a fourth order tensor when n is large. Finally, we introduce the definition of Z-eigenvalue based doubly strictly diagonally dominated(Z-QDSDD) symmetric fourth order tensors and show that, if a tensor $A$ is Z-QDSDD, then $A$ is positive definite.

## Author Contributions

Conceptualization, J.H. and Y.L.; software, J.T.; writing—original draft preparation, J.H.; writing—review and editing, Z.Z.; funding acquisition, J.H. and Y.L. and J.K. and Z.Z.

## Funding

This research was supported by National Natural Science Foundations of China (11661084, 71461027); Science and Technology Foundation of Guizhou province (Qian Ke He Ji Chu [2016]1161, [2017]1201, [2015]2147); Guizhou Province Natural Science Foundation in China (Qian Jiao He KY [2016]255, [2015]451, [2017]256); Innovative talent team in Guizhou Province (Qian Ke He Pingtai Rencai[2016]5619); High-level innovative talents of Guizhou Province (Zun Ke He Ren Cai[2017]8); The doctoral scientific research foundation of Zunyi Normal College (BS[2015]09).

## Conflicts of Interest

The authors declare no conflict of interest.

## References

1. Lim, L.H. Singular values and eigenvalues of tensors: A variational approach. In Proceedings of the IEEE International Workshop on Computational Advances in MultiSensor Adaptive Processing, Puerto Vallarta, Mexico, 13–15 December 2005; pp. 129–132. [Google Scholar]
2. Qi, L. Eigenvalues of a real supersymmetric tensor. J. Symb. Comput. 2005, 40, 1302–1324. [Google Scholar] [CrossRef]
3. Li, C.; Wang, F.; Zhao, J.; Zhu, Y.; Li, Y. Criterions for the positive definiteness of real supersymmetric tensors. J. Comput. Appl. Math. 2014, 255, 1–14. [Google Scholar] [CrossRef]
4. Bu, C.; Wei, Y.P.; Sun, L.; Zhou, J. Brualdi-type eigenvalue inclusion sets of tensors. Linear Algebra Appl. 2015, 480, 168–175. [Google Scholar] [CrossRef]
5. Bu, C.; Jin, X.; Li, H.; Deng, C. Brauer-type eigenvalue inclusion sets and the spectral radius of tensors. Linear Algebra Appl. 2017, 512, 234–248. [Google Scholar] [CrossRef]
6. Li, C.; Li, Y.; Kong, X. New eigenvalue inclusion sets for tensors. Numer. Linear Algebra Appl. 2014, 21, 39–50. [Google Scholar] [CrossRef]
7. Li, C.; Jiao, A.; Li, Y. An S-type eigenvalue localization set for tensors. Linear Algebra Appl. 2016, 493, 469–483. [Google Scholar] [CrossRef]
8. Li, Y.; Liu, Q.; Qi, L. Programmable criteria for strong H-tensors. Numer. Algorithms 2017, 1, 1–23. [Google Scholar]
9. Chen, Y.; Qi, L.; Wang, Q. Positive semi-definiteness and sum-of-squares property of fourth order four dimensional Hankel tensors. J. Comput. Appl. Math. 2016, 302, 356–368. [Google Scholar] [CrossRef][Green Version]
10. Barmpoutis, A.; Hwang, M.S.; Howland, D.; Forder, J.R.; Vemuri, B.C. Regularized positive-definite fourth order tensor filed estimation from DW-MRI. Neuroimage 2009, 45, S153–S162. [Google Scholar] [CrossRef]
11. Qi, L.; Yu, G.; Wu, E.X. Higher order positive semidefinite diffusion tensor imaging. SIAM J. Imaging Sci. 2010, 3, 416–433. [Google Scholar] [CrossRef][Green Version]
12. Buscarino, A.; Fortuna, L.; Frasca, M.; Xibilia, M.G. Invariance of characteristic values and L norm under lossless positive real transformations. J. Franklin Inst. 2016, 353, 2057–2073. [Google Scholar] [CrossRef]
13. Barmpoutis, A.; Jian, B.; Vemuri, B.C.; Shepherd, T.M. Symmetric positive 4th order tensors and their estimation from diffusion weighted MRI. In Information Processing and Medical Imaging; Karssemeijer, M., Lelieveldt, B., Eds.; Springer: Berlin, Germany, 2007; pp. 308–319. [Google Scholar]
14. Ghosh, A.; Descoteaux, M.; Deriche, R. Riemannian framework for estimating symmetric positive definite 4th order diffusion tensors. In Medical Image Com- puting and Computer-Assisted Intervention MICCAI 2008; Metaxas, D., Axel, L., Fichtinger, G., Szekeley, G., Eds.; Springer-Verlag: Berlin, Germany, 2008; pp. 858–865. [Google Scholar]
15. Wang, G.; Zhou, G.; Caccetta, L. Z-eigenvalue inclusion theorems for tensors. Discret. Contin. Dyn. Syst. Ser. B 2017, 22, 187–198. [Google Scholar] [CrossRef]
16. Song, Y.; Qi, L. Spectral properties of positively homogeneous operators induced by higher order tensors. SIAM J. Matrix Anal. Appl. 2013, 34, 1581–1595. [Google Scholar] [CrossRef]
17. Li, W.; Liu, D.; Vong, S.-W. Z-eigenpair bounds for an irreducible nonnegative tensor. Linear Algebra Appl. 2015, 483, 182–199. [Google Scholar] [CrossRef]
18. He, J. Bounds for the largest eigenvalue of nonnegative tensors. J. Comput. Anal. Appl. 2016, 20, 1290–1301. [Google Scholar]
19. He, J.; Liu, Y.-M.; Ke, H.; Tian, J.-K.; Li, X. Bounds for the Z-spectral radius of nonnegative tensors. Springerplus 2016, 5, 1727. [Google Scholar] [CrossRef] [PubMed]
20. Liu, Q.; Li, Y. Bounds for the Z-eigenpair of general nonnegative tensors. Open Math. 2016, 14, 181–194. [Google Scholar] [CrossRef]
21. He, J.; Huang, T.-Z. Upper bound for the largest Z-eigenvalue of positive tensors. Appl. Math. Lett. 2014, 38, 110–114. [Google Scholar] [CrossRef]
22. Zhao, J. A new Z-eigenvalue localization set for tensors. J. Inequal. Appl. 2017, 2017, 85. [Google Scholar] [CrossRef]
23. Chang, K.C.; Pearson, K.J.; Zhang, T. Some variational principles for Z-eigenvalues of nonnegative tensors. Linear Algebra Appl. 2013, 438, 4166–4182. [Google Scholar] [CrossRef]
24. Sang, C. A new Brauer-type Z -eigenvalue inclusion set for tensors. Numer. Algorithms 2018, 1, 1–14. [Google Scholar] [CrossRef]
Figure 1. Comparisons of Z-eigenvalue inclusion sets.
Figure 1. Comparisons of Z-eigenvalue inclusion sets.