Research article

Limit theorems for negatively superadditive-dependent random variables with infinite or finite means

  • The author studies the laws of large numbers for weighted sums of negatively superadditive-dependent random variables. The obtained results in this paper extend and improve the corresponding theorems of Yang et al. [Commun. Stat. Theor. M., 48 (2019), 3044-3054]. Moreover, the author obtains a new theorem of mean convergence for weighted sums of negatively superadditive-dependent random variables, which was not considered in Yang et al. (2019).

    Citation: Yongfeng Wu. Limit theorems for negatively superadditive-dependent random variables with infinite or finite means[J]. AIMS Mathematics, 2023, 8(11): 25311-25324. doi: 10.3934/math.20231291

    Related Papers:

    [1] Muhammad Tariq, Hijaz Ahmad, Soubhagya Kumar Sahoo, Artion Kashuri, Taher A. Nofal, Ching-Hsien Hsu . Inequalities of Simpson-Mercer-type including Atangana-Baleanu fractional operators and their applications. AIMS Mathematics, 2022, 7(8): 15159-15181. doi: 10.3934/math.2022831
    [2] Soubhagya Kumar Sahoo, Fahd Jarad, Bibhakar Kodamasingh, Artion Kashuri . Hermite-Hadamard type inclusions via generalized Atangana-Baleanu fractional operator with application. AIMS Mathematics, 2022, 7(7): 12303-12321. doi: 10.3934/math.2022683
    [3] Saad Ihsan Butt, Erhan Set, Saba Yousaf, Thabet Abdeljawad, Wasfi Shatanawi . Generalized integral inequalities for ABK-fractional integral operators. AIMS Mathematics, 2021, 6(9): 10164-10191. doi: 10.3934/math.2021589
    [4] Tekin Toplu, Mahir Kadakal, İmdat İşcan . On n-Polynomial convexity and some related inequalities. AIMS Mathematics, 2020, 5(2): 1304-1318. doi: 10.3934/math.2020089
    [5] M. Emin Özdemir, Saad I. Butt, Bahtiyar Bayraktar, Jamshed Nasir . Several integral inequalities for (α, s,m)-convex functions. AIMS Mathematics, 2020, 5(4): 3906-3921. doi: 10.3934/math.2020253
    [6] Hong Yang, Shahid Qaisar, Arslan Munir, Muhammad Naeem . New inequalities via Caputo-Fabrizio integral operator with applications. AIMS Mathematics, 2023, 8(8): 19391-19412. doi: 10.3934/math.2023989
    [7] Haoliang Fu, Muhammad Shoaib Saleem, Waqas Nazeer, Mamoona Ghafoor, Peigen Li . On Hermite-Hadamard type inequalities for n-polynomial convex stochastic processes. AIMS Mathematics, 2021, 6(6): 6322-6339. doi: 10.3934/math.2021371
    [8] Jia-Bao Liu, Saad Ihsan Butt, Jamshed Nasir, Adnan Aslam, Asfand Fahad, Jarunee Soontharanon . Jensen-Mercer variant of Hermite-Hadamard type inequalities via Atangana-Baleanu fractional operator. AIMS Mathematics, 2022, 7(2): 2123-2141. doi: 10.3934/math.2022121
    [9] Shuang-Shuang Zhou, Saima Rashid, Muhammad Aslam Noor, Khalida Inayat Noor, Farhat Safdar, Yu-Ming Chu . New Hermite-Hadamard type inequalities for exponentially convex functions and applications. AIMS Mathematics, 2020, 5(6): 6874-6901. doi: 10.3934/math.2020441
    [10] Jamshed Nasir, Saber Mansour, Shahid Qaisar, Hassen Aydi . Some variants on Mercer's Hermite-Hadamard like inclusions of interval-valued functions for strong Kernel. AIMS Mathematics, 2023, 8(5): 10001-10020. doi: 10.3934/math.2023506
  • The author studies the laws of large numbers for weighted sums of negatively superadditive-dependent random variables. The obtained results in this paper extend and improve the corresponding theorems of Yang et al. [Commun. Stat. Theor. M., 48 (2019), 3044-3054]. Moreover, the author obtains a new theorem of mean convergence for weighted sums of negatively superadditive-dependent random variables, which was not considered in Yang et al. (2019).



    A set KR is said to be convex, if Θ,ΩK and τ[0,1], we have

    (1τ)Θ+τΩK.

    A function Φ:KR is said to be convex, if Θ,ΩK and τ[0,1], we get

    Φ((1τ)Θ+τΩ)(1τ)Φ(Θ)+τΦ(Ω).

    Theory of convexity has also played an important part in the development of theory of inequalities. Several results in theory of inequalities are direct consequences of the applications of convexity. Among these results one of the most extensively as well as intensively studied result is the Hermite-Hadamard inequality. This result reads as:

    Let Φ:I=[a1,a2]RR be a convex function, then

    Φ(a1+a22)1a2a1a2a1Φ(Θ)dΘΦ(a1)+Φ(a2)2.

    In recent years, different approaches have been used in obtaining new versions of Hermite-Hadamard inequality. Mohammed et al. [1] found new discrete inequalities of Hermite-Hadamard type for convex functions. Mohammed et al. [2] established generalized Hermite-Hadamard inequalities via the tempered fractional integrals. Mohammed et al. [3] obtained new fractional inequalities of Hermite-Hadamard type involving the incomplete gamma functions. Rahman et al. [4] derived certain fractional proportional integral inequalities via convex functions. Rahman et al. [5] give new bounds of generalized proportional fractional integrals in general form via convex functions and presented their applications. Sarikaya et al. [6] have used the approach of fractional calculus and obtained fractional analogues of Hermite-Hadamard inequality. Since then a variety of different approaches from fractional calculus have been used in obtaining fractional analogues of classical inequalities. For instance, Budak and Agarwal [7] obtained new generalized fractional midpoint type inequalities. Chu et al. [8] obtained new Simpson type of inequalities using Katugampola fractional integrals. Chu et al. [9] new generalized fractional Hermite-Hadamard inequality using χk-Hilfer fractional integrals. Kashuri et al. [10] obtained new generalized fractional integral identities and obtained new inequalities. Liu et al. [11] obtained new Hermite-Hadamard type of inequalities via ψ-fractional integrals. Onalan et al. [12] obtained fractional analogues of Hermite-Hadamard type integral inequalities via fractional integral operators with Mittag-Leffler kernel. Talib and Awan [13] obtained some new estimates of upper bounds for n-th order differentiable functions involving χ-Riemann-Liouville integrals via γ-preinvex functions. Wu et al. [14] obtained estimates of upper bound for a k-th order differentiable functions involving Riemann-Liouville integrals via higher order strongly h-preinvex functions. Wu et al. [15] established some integral inequalities for n-polynomial ζ-preinvex functions. Zhang et al. [16] obtained new k-fractional integral inequalities containing multi parameters via generalized (s,m)-preinvexity property of the functions. Huang et al. [17] derived some inequalities of the Hermite-Hadamard type via k-fractional conformable integrals. Rahman et al. [18] established certain inequalities via generalized proportional Hadamard fractional integral operators. Rahman et al. [19] obtained the Minkowski inequalities via generalized proportional fractional integral operators. For some more details, see [20,21,22,23,24,25].

    In recent years, several new extensions and generalizations of classical convex functions have been defined in the literature. Chu et al. [8] introduced the notion of higher order strongly n-polynomial convex function as follows:

    Definition 1.1. [[8]] A function Φ:KR is said to be higher order strongly n-polynomial convex, if Θ,ΩK,τ[0,1], and u,σ>0, we have

    Φ(τΘ+(1τ)Ω)1nns=1[1(1τ)s]Φ(Θ)+1nns=1[1τs]Φ(Ω)u(τσ(1τ)+τ(1τ)σ)ΩΘσ.

    Remark 1.1. Note that, if we take u=0 in Definition 1, then we have the class of n-polynomial convex functions introduced and studied by Toplu et al. [26]. If we take σ=2 in Definition 1, then we get the class of strongly n-polynomial convex functions. If we take n=1 and σ=2, then we obtain the class of strongly convex functions [27].

    Fractional calculus is an effective tool to explain physical phenomenas and also real world problems. The concept of fractional order derivative and integrals that will shed light on some unknown points about differential equations and solutions of some fractional order differential equations, which proved to be useless for their solution, is a novelty in applied sciences as well as in mathematics. New derivatives and integrals contribute to the solution of differential equations that are expressed and solved in classical analysis, as well as fractional order derivatives and integrals. In addition, it has increased its contribution to the literature with its applications in areas such as engineering, biostatistics and mathematical biology. Fractional derivative and integral operators not only differed from each other in terms of singularity, locality and kernels, but also brought innovations to fractional analysis in terms of their usage areas and spaces. Baleanu et al. [28], investigated existence results for solutions of a coupled system of hybrid boundary value problems with hybrid conditions. Khan et al. [29], analyzed positive solution and Hyers-Ulam stability for a class of singular fractional differential equations with p-Laplacian in Banach space. Khan et al. [30], give the existence and Hyers-Ulam stability for a nonlinear singular fractional differential equations with Mittag-Leffler kernel. For more details regarding fractional calculus and their applications, see [31,32,33,34,35].

    Let us recall some basic concepts from fractional calculus which will be helpful in obtaining our main results.

    Definition 1.2. [36] Let ΦL1[a1,a2]. The Riemann-Liouville integrals Jνa1+Φ and Jνa2Φ of order ν>0 are defined by

    Jνa1+Φ(Θ)=1Γ(ν)Θa1(Θτ)ν1Φ(τ)dτ,Θ>a1,

    and

    Jνa2Φ(Θ)=1Γ(ν)a2Θ(τΘ)ν1Φ(τ)dτ,Θ<a2.

    In 2015 Caputo and Fabrizio suggested a new operator with fractional order, this derivative is based on the exponential kernel. Earlier this year 2016 Atangana and Baleanu developed another version which used the generalized Mittag-Leffler function as non-local and non-singular kernel. Both operators show some properties of filter. However, the Atangana and Baleanu version has in addition to this, all properties of fractional derivative. This shown effectiveness and advantages of the Atangana-Baleanu integral operators.

    So, let see the following definitions about Atangana-Baleanu fractional derivatives and fractional integrals that are given in [37,38,39], respectively.

    Definition 1.3. Let ΦH1(a1,a2) and not necessarily differentiable then, the definition of the new fractional derivative (Atangana-Baleanu fractional derivative in Riemann-Liouville sense) is given as follows:

    ABRa1DατΦ(τ)=B(α)1αddττa1Φ(Θ)Eα[α(τΘ)α1α]dΘ,a2>a1,α[0,1],

    where Eα(Θ) is the well-known Mittag-Leffler function.

    Definition 1.4. Let ΦH1(a1,a2) then, the definition of the new fractional derivative (Atangana-Baleanu derivative in Caputo sense) is given as:

    ABCa1DατΦ(τ)=B(α)1ατa1Φ(Θ)Eα[α(τΘ)α1α]dΘ,a2>a1,α[0,1].

    Here, B(α)>0 is the normalization function which satisfies the condition B(0)=B(1)=1. They suggested that B(α) has the same properties as in Caputo and Fabrizio case. The above definitions are very helpful to real world problem and also they have great advantages when Laplace transform is apply to solve some physical problems. Since the normalization function B(α) is positive, it immediately follows that the fractional Atangana-Baleanu integral of a positive function is positive. It should be noted that, when the order α1, we recover the classical integral. Also, the initial function is recovered whenever the fractional order α0.

    Definition 1.5. The left hand side fractional integral related to the new fractional derivative with nonlocal kernel of a function ΦL1[a1,a2] is defined as follows:

    ABa1IατΦ(τ)=1αB(α)Φ(τ)+αB(α)Γ(α)τa1Φ(Θ)(τΘ)α1dΘ,a2>a1,α[0,1].

    The right hand side of Atangana-Baleanu fractional integral is given as:

    ABIαa2Φ(τ)=1αB(α)Φ(τ)+αB(α)Γ(α)a2τΦ(Θ)(Θτ)α1dΘ.

    Here, Γ(α) is the gamma function.

    For more details about Atangana-Baleanu fractional integral operators and their applications, see [40,41,42,43,44,45,46,47,48].

    Inspired by above results, the main motivation of this paper is to establish two new fractional integral identities involving Atangana-Baleanu fractional integrals. Using these identities as auxiliary results, we will derive new fractional counterparts of classical inequalities essentially using first and second order differentiable higher order strongly n-polynomial convex functions. Furthermore, in order to show the efficiency of our main results, we will offer applications for special means and for differentiable functions of first and second order that are in absolute value bounded. Finally, some conclusions and future research will be given.

    In this section, we derive two new fractional integral identities using Atangana-Baleanu fractional integrals.

    Lemma 2.1. Let Φ:[a1,a2]R be a differentiable function on (a1,a2) with a1<a2. If ΦL1[a1,a2], then for all Θ[a1,a2], we have

    ABIαa2{Φ(a1+a2Θ)}+ABa1Iατ{Φ(a1+a2Θ)}Φ(a1+a2Θ)B(α)Γ(α)[(Θa1)α+(a2Θ)α+2(1α)Γ(α)]=1B(α)Γ(α)[(Θa1)α+110(1τα)Φ(τa2+(1τ)(a1+a2Θ))dτ+(a2Θ)α+110(τα1)Φ(τa1+(1τ)(a1+a2Θ))dτ].

    Proof. Integrating by parts, we have

    10(1τα)Φ(τa2+(1τ)(a1+a2Θ))dτ=Φ(a1+a2Θ)Θa1+αΘa110τα1Φ(τa2+(1τ)(a1+a2Θ))dτ=Φ(a1+a2Θ)Θa1+α(Θa1)α+1a2a1+a2Θ(u(a1+a2Θ))α1Φ(u)du.

    Multiplying both sides of the last inequality by 1B(α)Γ(α), and then adding the term 1αB(α)Φ(a1+a2Θ), we get

    (Θa1)α+1B(α)Γ(α)10(1τα)Φ(τa2+(1τ)(a1+a2Θ))dτ+1αB(α)Φ(a1+a2Θ)=(Θa1)αB(α)Γ(α)Φ(a1+a2Θ)+αB(α)Γ(α)a2a1+a2Θ(u(a1+a2Θ))α1Φ(u)du+1αB(α)Φ(a1+a2Θ)=(Θa1)αB(α)Γ(α)Φ(a1+a2Θ)+ABIαa2{Φ(a1+a2Θ)}. (2.1)

    Similarly, we have

    (a2Θ)α+1B(α)Γ(α)10(τα1)Φ(τa1+(1τ)(a1+a2Θ))dτ+1αB(α)Φ(a1+a2Θ)=(a2Θ)αB(α)Γ(α)Φ(a1+a2Θ)+ABa1Iαa1+a2Θ{Φ(a1+a2Θ)}. (2.2)

    By the identities (2.1) and (2.2), we obtain the required result.

    Remark 2.1. Taking α=1 in Lemma 2.1, we get the following identity:

    1a2a1a2a1Φ(Θ)dΘΦ(a1+a2Θ)=1a2a1[(Θa1)210(1τ)Φ(τa2+(1τ)(a1+a2Θ))dτ+(a2Θ)210(τ1)Φ(τa1+(1τ)(a1+a2Θ))dτ].

    Lemma 2.2. Let Φ:[a1,a2]R be a twice differentiable function on (a1,a2) with a1<a2. If ΦL1[a1,a2], then for all Θ[a1,a2], we have

    Φ(a1+a2Θ)(α+1)B(α)Γ(α)[(a2Θ)α+1(Θa1)α+1][ABIαa2{Φ(a1+a2Θ)}+ABa1Iαa1+a2Θ{Φ(a1+a2Θ)}]+2(1α)B(α)Φ(a1+a2Θ)+1B(α)Γ(α)[(Θa1)αΦ(a2)+(a2Θ)αΦ(a1)]=1(α+1)B(α)Γ(α)[(Θa1)α+210(1τα+1)Φ(τa2+(1τ)(a1+a2Θ))dτ(a2Θ)α+210(τα+11)Φ(τa1+(1τ)(a1+a2Θ))dτ]. (2.3)

    Proof. Consider the right hand side of 2.3, we have

    I:=1(α+1)B(α)Γ(α)[(Θa1)α+210(1τα+1)Φ(τa2+(1τ)(a1+a2Θ))dτ(a2Θ)α+210(τα+11)Φ(τa1+(1τ)(a1+a2Θ))dτ]=1(α+1)B(α)Γ(α)[(Θa1)α+2I1(a2Θ)α+2I2], (2.4)

    where

    I1:=10(1τα+1)Φ(τa2+(1τ)(a1+a2Θ))dτ=Φ(a1+a2Θ)Θa1+α+1Θa110ταΦ(τa2+(1τ)(a1+a2Θ))dτ=Φ(a1+a2Θ)Θa1+α+1(Θa1)[Φ(a2)Θa1α(Θa1)10τα1Φ(τa2+(1τ)(a1+a2Θ))dτ]=Φ(a1+a2Θ)Θa1+(α+1)Φ(a2)(Θa1)2α(α+1)(Θa1)210τα1Φ(τa2+(1τ)(a1+a2Θ))dτ=Φ(a1+a2Θ)Θa1+(α+1)Φ(a2)(Θa1)2+(1+α)Γ(α)Φ(a1+a2Θ)(Θa1)α+2(1+α)B(α)Γ(α)(Θa1)α+2ABIαa2{Φ(a1+a2Θ)}.

    Similarly, we get

    I2:=10(τα+11)Φ(τa1+(1τ)(a1+a2Θ))dτ=Φ(a1+a2Θ)a2Θ(α+1)Φ(a1)(a2Θ)2(1+α)Γ(α)Φ(a1+a2Θ)(a2Θ)α+2+(1+α)B(α)Γ(α)(a2Θ)α+2ABa1Iαa1+a2Θ{Φ(a1+a2Θ)}.

    By substituting the values of I1 and I2 in 2.4, we obtain our required result.

    Remark 3.2. Taking α=1 in Lemma 2.2, we get the following identity:

    Φ(a1+a2Θ)[(a2Θ)2(Θa1)2]+2[(Θa1)Φ(a2)+(a2Θ)Φ(a1)]2a2a1Φ(Θ)dΘ=[(Θa1)310(1τ2)Φ(τa2+(1τ)(a1+a2Θ))dτ(a2Θ)310(τ21)Φ(τa1+(1τ)(a1+a2Θ))dτ].

    In this section, we discuss our main results.

    Theorem 3.1. Let Φ:[a1,a2]R be a differentiable function on (a1,a2) with 0a1<a2. If |Φ|q is higher order strongly n-polynomial convex on [a1,a2] for q>1 and 1p+1q=1, then for all Θ[a1,a2] the following inequality for fractional integrals holds:

    |ABIαa2{Φ(a1+a2Θ)}+ABa1Iατ{Φ(a1+a2Θ)}Φ(a1+a2Θ)B(α)Γ(α)[(Θa1)α+(a2Θ)α+2(1α)Γ(α)]|1B(α)Γ(α)(αpαp+1)1p[(Θa1)α+1(1nns=1ss+1(|Φ(a2)|q+|Φ(a1+a2Θ)|q)2uΘa1σ(σ+1)(σ+2))1q+(a2Θ)α+1(1nns=1ss+1(|Φ(a1)|q+|Φ(a1+a2Θ)|q)2ua2Θσ(σ+1)(σ+2))1q].

    Proof. It is evident that

    10[1(1τ)s]dτ=10[1τs]dτ=ss+1.

    Applying Lemma 2.1 and properties of modulus, we have

    |ABIαa2{Φ(a1+a2Θ)}+ABa1Iατ{Φ(a1+a2Θ)}Φ(a1+a2Θ)B(α)Γ(α)[(Θa1)α+(a2Θ)α+2(1α)Γ(α)]|1B(α)Γ(α)[(Θa1)α+110|1τα||Φ(τa2+(1τ)(a1+a2Θ))|dτ+(a2Θ)α+110|τα1||Φ(τa1+(1τ)(a1+a2Θ))|dτ].

    Using Hölder's inequality and higher order strongly n-polynomial convexity of |Φ|q, we get

    10|1τα||Φ(τa2+(1τ)(a1+a2Θ))|dτ(10|1τα|pdτ)1p(10|Φ(τa2+(1τ)(a1+a2Θ))|qdτ)1q(10(1ταp)dτ)1p(10[1nns=1[1(1τ)s]|Φ(a2)|q+1nns=1[1τs]|Φ(a1+a2Θ)|qu(τσ(1τ)+τ(1τ)σ)Θa1σ]dτ)1q=(αpαp+1)1p[1nns=1ss+1(|Φ(a2)|q+|Φ(a1+a2Θ)|q)2uΘa1σ(σ+1)(σ+2)]1q. (3.1)

    Here, we use

    (AB)qAqBq

    for any A>B0 and q1.

    Similarly, we have

    10|τα1||Φ(τa1+(1τ)(a1+a2Θ))|dτ(αpαp+1)1p[1nns=1ss+1(|Φ(a1)|q+|Φ(a1+a2Θ)|q)2ua2Θσ(σ+1)(σ+2)]1q. (3.2)

    By the inequalities (3.1) and (3.2), we obtain required result.

    Corollary 3.1. Taking α=1 in Theorem 3.1 and using Remark 2.1, we have

    |Φ(a1+a2Θ)1a2a1a2a1Φ(Θ)dΘ|1a2a1(pp+1)1p[(Θa1)2(1nns=1ss+1(|Φ(a2)|q+|Φ(a1+a2Θ)|q)2uΘa1σ(σ+1)(σ+2))1q+(a2Θ)2(1nns=1ss+1(|Φ(a1)|q+|Φ(a1+a2Θ)|q)2ua2Θσ(σ+1)(σ+2))1q].

    Corollary 3.2. Taking u0+ in Theorem 3.1, we have

    |ABIαa2{Φ(a1+a2Θ)}+ABa1Iατ{Φ(a1+a2Θ)}Φ(a1+a2Θ)B(α)Γ(α)[(Θa1)α+(a2Θ)α+2(1α)Γ(α)]|1B(α)Γ(α)(αpαp+1)1p[(Θa1)α+1(1nns=1ss+1(|Φ(a2)|q+|Φ(a1+a2Θ)|q))1q+(a2Θ)α+1(1nns=1ss+1(|Φ(a1)|q+|Φ(a1+a2Θ)|q)1q].

    Corollary 3.3. Under assumptions of Theorem 3.1 with Θ=a1+a22, we have the following inequality

    |ABIαa2Φ(a1+a22)+ABa1Iα(a1+a22)Φ(a1+a22)2Φ(a1+a22)B(α)Γ(α)[(a2a12)α+(1α)Γ(α)]|1B(α)Γ(α)(αpαp+1)1p(a2a12)α+1[(1nns=1ss+1(|Φ(a2)|q+|Φ(a1+a22)|q)2u(a2a12)σ(σ+1)(σ+2))1q+(1nns=1ss+1(|Φ(a1)|q+|Φ(a1+a22)|q)2u(a2a12)σ(σ+1)(σ+2))1q].

    Theorem 3.2. Let Φ:[a1,a2]R be a differentiable function on (a1,a2) with 0a1<a2. If |Φ|q is higher order strongly n-polynomial convex on [a1,a2] for q1, then for all Θ[a1,a2] the following inequality for fractional integrals holds:

    |ABIαa2{Φ(a1+a2Θ)}+ABa1Iατ{Φ(a1+a2Θ)}Φ(a1+a2Θ)B(α)Γ(α)[(Θa1)α+(a2Θ)α+2(1α)Γ(α)]|1B(α)Γ(α)(αα+1)11q[(Θa1)α+1[|Φ(a2)|qnns=1(ss+11α+1+Γ(1+s)Γ(1+α)Γ(2+s+α))+|Φ(a1+a2Θ)|qnns=1(ss+11α+1+1α+s+1)uΘa1σ(1(σ+1)(σ+2)1(1+α+σ)(2+α+σ)Γ(2+α)Γ(1+σ)Γ(3+α+σ))]1q+(a2Θ)α+1[|Φ(a1)|qnns=1(ss+11α+1+Γ(1+s)Γ(1+α)Γ(2+s+α))+|Φ(a1+a2Θ)|qnns=1(ss+11α+1+1α+s+1)ua2Θσ(1(σ+1)(σ+2)1(1+α+σ)(2+α+σ)Γ(2+α)Γ(1+σ)Γ(3+α+σ))]1q].

    Proof. From Lemma 2.1, properties of modulus and power mean inequality, we have

    |ABIαa2{Φ(a1+a2Θ)}+ABa1Iατ{Φ(a1+a2Θ)}Φ(a1+a2Θ)B(α)Γ(α)[(Θa1)α+(a2Θ)α+2(1α)Γ(α)]|1B(α)Γ(α)[(Θa1)α+110|1τα||Φ(τa2+(1τ)(a1+a2Θ))|dτ+(a2Θ)α+110|τα1||Φ(τa1+(1τ)(a1+a2Θ))|dτ]1B(α)Γ(α)[(Θa1)α+1(10|1τα|dτ)11q(10|1τα||Φ(τa2+(1τ)(a1+a2Θ))|qdτ)1q+(a2Θ)α+1(10|τα1|dτ)11q(10|τα1||Φ(τa1+(1τ)(a1+a2Θ))|qdτ)1q].

    Using the higher order strongly n-polynomial convexity of |Φ|q, we get

    10|1τα||Φ(τa2+(1τ)(a1+a2Θ))|dτ10(1τα)[1nns=1[1(1τ)s]|Φ(a2)|q+1nns=1[1τs]|Φ(a1+a2Θ)|qu(τσ(1τ)+τ(1τ)σ)Θa1σ]dτ=|Φ(a2)|qnns=1(ss+11α+1+Γ(1+s)Γ(1+α)Γ(2+s+α))+|Φ(a1+a2Θ)|qnns=1(ss+11α+1+1α+s+1)uΘa1σ(1(σ+1)(σ+2)1(1+α+σ)(2+α+σ)Γ(2+α)Γ(1+σ)Γ(3+α+σ)). (3.3)

    Similarly, we have

    10|τα1||Φ(τa1+(1τ)(a1+a2Θ))|dτ10(τα1)[1nns=1[1(1τ)s]|Φ(a1)|q+1nns=1[1τs]|Φ(a1+a2Θ)|qu(τσ(1τ)+τ(1τ)σ)a2Θσ]dτ=|Φ(a1)|qnns=1(ss+11α+1+Γ(1+s)Γ(1+α)Γ(2+s+α))+|Φ(a1+a2Θ)|qnns=1(ss+11α+1+1α+s+1)ua2Θσ(1(σ+1)(σ+2)1(1+α+σ)(2+α+σ)Γ(2+α)Γ(1+σ)Γ(3+α+σ)). (3.4)

    By the inequalities (3.3) and (3.4), we obtain required result.

    Corollary 3.4. Taking α=1 in Theorem 3.2 and using Remark 2.1, we have

    |Φ(a1+a2Θ)1a2a1a2a1Φ(Θ)dΘ|1a2a1(12)11q[(Θa1)2[|Φ(a2)|qnns=1(ss+112+Γ(s+1)Γ(s+3))+|Φ(a1+a2Θ)|qnns=1(ss+112+1s+2)uΘa1σ(1(σ+1)(σ+2)1(2+σ)(3+σ)2Γ(1+σ)Γ(4+σ))]1q+(a2Θ)2[|Φ(a1)|qnns=1(ss+112+Γ(s+1)Γ(s+3))+|Φ(a1+a2Θ)|qnns=1(ss+112+1s+2)ua2Θσ(1(σ+1)(σ+2)1(2+σ)(3+σ)2Γ(1+σ)Γ(4+σ))]1q].

    Corollary 3.5. Taking u0+ in Theorem 3.2, we have

    |ABIαa2{Φ(a1+a2Θ)}+ABa1Iατ{Φ(a1+a2Θ)}Φ(a1+a2Θ)B(α)Γ(α)[(Θa1)α+(a2Θ)α+2(1α)Γ(α)]|1B(α)Γ(α)(αα+1)11q[(Θa1)α+1[|Φ(a2)|qnns=1(ss+11α+1+Γ(1+s)Γ(1+α)Γ(2+s+α))+|Φ(a1+a2Θ)|qnns=1(ss+11α+1+1α+s+1)]1q+(a2Θ)α+1[|Φ(a1)|qnns=1(ss+11α+1+Γ(1+s)Γ(1+α)Γ(2+s+α))+|Φ(a1+a2Θ)|qnns=1(ss+11α+1+1α+s+1)]1q].

    Corollary 3.6. Under assumptions of Theorem 3.2 with Θ=a1+a22, we have the following inequality

    |ABIαa2Φ(a1+a22)+ABa1Iα(a1+a22)Φ(a1+a22)2Φ(a1+a22)B(α)Γ(α)[(a2a12)α+(1α)Γ(α)]|1B(α)Γ(α)(αα+1)11q(a2a12)α+1[[|Φ(a2)|qnns=1(ss+11α+1+Γ(1+s)Γ(1+α)Γ(2+s+α))+|Φ(a1+a22)|qnns=1(ss+11α+1+1α+s+1)u(a2a12)σ(1(σ+1)(σ+2)1(1+α+σ)(2+α+σ)Γ(2+α)Γ(1+σ)Γ(3+α+σ))]1q+[|Φ(a1)|qnns=1(ss+11α+1+Γ(1+s)Γ(1+α)Γ(2+s+α))+|Φ(a1+a22)|qnns=1(ss+11α+1+1α+s+1)u(a2a12)σ(1(σ+1)(σ+2)1(1+α+σ)(2+α+σ)Γ(2+α)Γ(1+σ)Γ(3+α+σ))]1q].

    Theorem 3.3. Let Φ:[a1,a2]R be a twice differentiable function on (a1,a2) with 0a1<a2. If |Φ|q is higher order strongly n-polynomial convex on [a1,a2] for q>1 and 1p+1q=1, then for all Θ[a1,a2] the following inequality for fractional integrals holds:

    |Φ(a1+a2Θ)(α+1)B(α)Γ(α)[(a2Θ)α+1(Θa1)α+1][ABIαa2{Φ(a1+a2Θ)}+ABa1Iαa1+a2Θ{Φ(a1+a2Θ)}]+2(1α)B(α)Φ(a1+a2Θ)+1B(α)Γ(α)[(Θa1)αΦ(a2)+(a2Θ)αΦ(a1)]|1(α+1)B(α)Γ(α)(p(α+1)p(α+1)+1)1p[(Θa1)α+2(1nns=1ss+1(|Φ(a2)|q+|Φ(a1+a2Θ)|q)2uΘa1σ(σ+1)(σ+2))1q+(a2Θ)α+2(1nns=1ss+1(|Φ(a1)|q+|Φ(a1+a2Θ)|q)2ua2Θσ(σ+1)(σ+2))1q],

    where 1p+1q=1.

    Proof. Applying the Lemma 2.2 and properties of modulus, we have

    |Φ(a1+a2Θ)(α+1)B(α)Γ(α)[(a2Θ)α+1(Θa1)α+1][ABIαa2{Φ(a1+a2Θ)}+ABa1Iαa1+a2Θ{Φ(a1+a2Θ)}]+2(1α)B(α)Φ(a1+a2Θ)+1B(α)Γ(α)[(Θa1)αΦ(a2)+(a2Θ)αΦ(a1)]|1(α+1)B(α)Γ(α)[(Θa1)α+210|1τα+1||Φ(τa2+(1τ)(a1+a2Θ))|dτ+(a2Θ)α+210|τα+11||Φ(τa1+(1τ)(a1+a2Θ))|dτ]

    Using Hölder's inequality and higher order strongly n-polynomial convexity of |Φ|q, we get

    10|1τα+1||Φ(τa2+(1τ)(a1+a2Θ))|dτ(10|1τα+1|pdτ)1p(10|Φ(τa2+(1τ)(a1+a2Θ))|qdτ)1q(10(1τ(α+1)p)dτ)1p(10[1nns=1[1(1τ)s]|Φ(a2)|q+1nns=1[1τs]|Φ(a1+a2Θ)|qu(τσ(1τ)+τ(1τ)σ)Θa1σ]dτ)1q=((α+1)p(α+1)p+1)1p[1nns=1ss+1(|Φ(a2)|q+|Φ(a1+a2Θ)|q)2uΘa1σ(σ+1)(σ+2)]1q. (3.5)

    Here, we use

    (AB)qAqBq

    for any A>B0 and q1.

    Similarly, we have

    10|τα+11||Φ(τa1+(1τ)(a1+a2Θ))|dτ((α+1)p(α+1)p+1)1p[1nns=1ss+1(|Φ(a1)|q+|Φ(a1+a2Θ)|q)2ua2Θσ(σ+1)(σ+2)]1q. (3.6)

    By the inequalities (3.5) and (3.6), we obtain required result.

    Corollary 3.7. Taking α=1 in Theorem 3.3 and using Remark 2.2, we have

    |Φ(a1+a2Θ)[(a2Θ)2(Θa1)2]+2[(Θa1)Φ(a2)+(a2Θ)Φ(a1)]2a2a1Φ(Θ)dΘ|(2p2p+1)1p[(Θa1)3(1nns=1ss+1(|Φ(a2)|q+|Φ(a1+a2Θ)|q)2uΘa1σ(σ+1)(σ+2))1q+(a2Θ)3(1nns=1ss+1(|Φ(a1)|q+|Φ(a1+a2Θ)|q)2ua2Θσ(σ+1)(σ+2))1q]

    Corollary 3.8. Taking u0+ in Theorem 3.3, we have

    |Φ(a1+a2Θ)(α+1)B(α)Γ(α)[(a2Θ)α+1(Θa1)α+1][ABIαa2{Φ(a1+a2Θ)}+ABa1Iαa1+a2Θ{Φ(a1+a2Θ)}]+2(1α)B(α)Φ(a1+a2Θ)+1B(α)Γ(α)[(Θa1)αΦ(a2)+(a2Θ)αΦ(a1)]|1(α+1)B(α)Γ(α)(p(α+1)p(α+1)+1)1p[(Θa1)2α+3(1nns=1ss+1(|Φ(a2)|q+|Φ(a1+a2Θ)|q))1q+(a2Θ)2α+3(1nns=1ss+1(|Φ(a1)|q+|Φ(a1+a2Θ)|q))1q]

    Corollary 3.9. Under assumptions of Theorem 3.3 with Θ=a1+a22, we have the following inequality

    |2(1α)B(α)Φ(a1+a22)[ABIαa2{Φ(a1+a22)}+ABa1Iα(a1+a22){Φ(a1+a22)}]+(a2a1)α2αB(α)Γ(α)[Φ(a1)+Φ(a2)]|1(α+1)B(α)Γ(α)(p(α+1)p(α+1)+1)1p[(a2a12)α+2(1nns=1ss+1(|Φ(a2)|q+|Φ(a1+a22)|q)2ua2a1σ(σ+1)(σ+2))1q+(a2a12)α+2(1nns=1ss+1(|Φ(a1)|q+|Φ(a1+a22)|q)2ua2Θσ(σ+1)(σ+2))1q]

    Theorem 3.4. Let Φ:[a1,a2]R be a differentiable function on (a1,a2) with 0a1<a2. If |Φ|q is higher order strongly n-polynomial convex on [a1,a2] for q1, then for all Θ[a1,a2] the following inequality for fractional integrals holds:

    |Φ(a1+a2Θ)(α+1)B(α)Γ(α)[(a2Θ)α+1(Θa1)α+1][ABIαa2{Φ(a1+a2Θ)}+ABa1Iαa1+a2Θ{Φ(a1+a2Θ)}]+2(1α)B(α)Φ(a1+a2Θ)+1B(α)Γ(α)[(Θa1)αΦ(a2)+(a2Θ)αΦ(a1)]|1(α+1)B(α)Γ(α)(α+1α+2)11q[(Θa1)α+2(|Φ(a2)|qnns=1(111+s12+α+Γ(1+s)Γ(2+α)Γ(3+s+α))+|Φ(a1+a2Θ)|qnns=1(111+s12+α+12+s+α)uΘa1σ(2(1+σ)(2+σ)1(2+α+σ)(3+α+σ)Γ(3+α)Γ(1+σ)Γ(4+α+σ)))1q+(a2Θ)α+2(|Φ(a1)|qnns=1(111+s12+α+Γ(1+s)Γ(2+α)Γ(3+s+α))+|Φ(a1+a2Θ)|qnns=1(111+s12+α+12+s+α)ua2Θσ(2(1+σ)(2+σ)1(2+α+σ)(3+α+σ)Γ(3+α)Γ(1+σ)Γ(4+α+σ)))1q]

    Proof. Using Lemma 2.2 and the property of modulus, we have

    |Φ(a1+a2Θ)(α+1)B(α)Γ(α)[(a2Θ)α+1(Θa1)α+1][ABIαa2{Φ(a1+a2Θ)}+ABa1Iαa1+a2Θ{Φ(a1+a2Θ)}]+2(1α)B(α)Φ(a1+a2Θ)+1B(α)Γ(α)[(Θa1)αΦ(a2)+(a2Θ)αΦ(a1)]|1(α+1)B(α)Γ(α)[(Θa1)α+210|1τα+1||Φ(τa2+(1τ)(a1+a2Θ))|dτ+(a2Θ)α+210|τα+11||Φ(τa1+(1τ)(a1+a2Θ))|dτ]

    Using power mean inequality and the higher order strongly n-polynomial convexity of |Φ|q, we get

    10|1τα+1||Φ(τa2+(1τ)(a1+a2Θ))|dτ(10|1τα+1|dτ)1p(10(1τα+1)|Φ(τa2+(1τ)(a1+a2Θ))|qdτ)1q(α+1α+2)1p(10(1τα+1)[1nns=1[1(1τ)s]|Φ(a2)|q+1nns=1[1τs]|Φ(a1+a2Θ)|qu(τσ(1τ)+τ(1τ)σ)Θa1σ]dτ)1q=(α+1α+2)1p(|Φ(a2)|qnns=1(111+s12+α+Γ(1+s)Γ(2+α)Γ(3+s+α))+|Φ(a1+a2Θ)|qnns=1(111+s12+α+12+s+α)uΘa1σ(2(1+σ)(2+σ)1(2+α+σ)(3+α+σ)Γ(3+α)Γ(1+σ)Γ(4+α+σ)))1q. (3.7)

    Similarly, we have

    10|τα+11||Φ(τa1+(1τ)(a1+a2Θ))|dτ=(α+1α+2)1p(|Φ(a1)|qnns=1(111+s12+α+Γ(1+s)Γ(2+α)Γ(3+s+α))+|Φ(a1+a2Θ)|qnns=1(111+s12+α+12+s+α)ua2Θσ(2(1+σ)(2+σ)1(2+α+σ)(3+α+σ)Γ(3+α)Γ(1+σ)Γ(4+α+σ)))1q. (3.8)

    By the inequalities (3.7) and (3.8), we obtain required result.

    Corollary 3.10. Taking α=1 in Theorem 3.4 and using Remark 2.2, we have

    |Φ(a1+a2Θ)[(a2Θ)2(Θa1)2]+2[(Θa1)Φ(a2)+(a2Θ)Φ(a1)]2a2a1Φ(Θ)dΘ|(23)11q[(Θa1)3(|Φ(a2)|qnns=1(2311+s+2Γ(1+s)Γ(4+s))+|Φ(a1+a2Θ)|qnns=1(2311+s+13+s)uΘa1σ(2(1+σ)(2+σ)1(3+σ)(4+σ)6Γ(1+σ)Γ(5+σ)))1q+(a2Θ)3(|Φ(a1)|qnns=1(2311+s+2Γ(1+s)Γ(4+s))+|Φ(a1+a2Θ)|qnns=1(2311+s+13+s)ua2Θσ(2(1+σ)(2+σ)1(3+σ)(4+σ)6Γ(1+σ)Γ(5+σ)))1q].

    Corollary 3.11. Taking u0+ in Theorem 3.4, we have

    |Φ(a1+a2Θ)(α+1)B(α)Γ(α)[(a2Θ)α+1(Θa1)α+1][ABIαa2{Φ(a1+a2Θ)}+ABa1Iαa1+a2Θ{Φ(a1+a2Θ)}]+2(1α)B(α)Φ(a1+a2Θ)+1B(α)Γ(α)[(Θa1)αΦ(a2)+(a2Θ)αΦ(a1)]|1(α+1)B(α)Γ(α)(α+1α+2)11q[(Θa1)2α+3(|Φ(a2)|qnns=1(111+s12+α+Γ(1+s)Γ(2+α)Γ(3+s+α))+|Φ(a1+a2Θ)|qnns=1(111+s12+α+12+s+α))1q+(a2Θ)2α+3(|Φ(a1)|qnns=1(111+s12+α+Γ(1+s)Γ(2+α)Γ(3+s+α))+|Φ(a1+a2Θ)|qnns=1(111+s12+α+12+s+α))1q]

    Corollary 3.12. Under assumptions of Theorem 3.4 with Θ=a1+a22, we have the following inequality

    |2(1α)B(α)Φ(a1+a22)+1B(α)Γ(α)(a2a12)α[Φ(a1)+Φ(a2)][ABIαa2{Φ(a1+a22)}+ABa1Iα(a1+a22){Φ(a1+a22)}]|1(α+1)B(α)Γ(α)(α+1α+2)11q(a2a12)α+2[(|Φ(a2)|qnns=1(111+s12+α+Γ(1+s)Γ(2+α)Γ(3+s+α))+|Φ(a1+a22)|qnns=1(111+s12+α+12+s+α)ua2a12σ(2(1+σ)(2+σ)1(2+α+σ)(3+α+σ)Γ(3+α)Γ(1+σ)Γ(4+α+σ)))1q+(|Φ(a1)|qnns=1(111+s12+α+Γ(1+s)Γ(2+α)Γ(3+s+α))+|Φ(a1+a22)|qnns=1(111+s12+α+12+s+α)ua2a12σ(2(1+σ)(2+σ)1(2+α+σ)(3+α+σ)Γ(3+α)Γ(1+σ)Γ(4+α+σ)))1q]

    Remark 3.1. Taking Θ=a1 or Θ=a2 in our main results, we can obtain several important special cases. We omit here their proofs and the details are left to the interested readers.

    In this section, we discuss applications of our main results.

    In this section, we discuss some applications of our main results to special means of positive real numbers. First of all, we recall some previously known concepts. For a1a2, we have

    (1) The arithmetic mean: A(a1,a2)=a1+a22.

    (2) The logarithmic mean: L(a1,a2)=a2a1ln(a2)ln(a1).

    (3) The generalized logarithmic mean: Lnn(a1,a2)=[a2n+1a1n+1(a2a1)(n+1)]1n,nZ{1,0}.

    Proposition 4.1. Suppose all the assumptions of Theorem 3.1 are satisfied, then

    (1)

    |Am(a1,a2)Lmm(a1,a2)|a2a14(pp+1)1p[{ns=1ss+1(ma2(m1)q+mA(n1)q(a1,a2))}1q+{ns=1ss+1(ma1(m1)q+mA(m1)q(a1,a2))}1q],

    (2)

    |L1(a1,a2)A1(a1,a2)|a2a14(pp+1)1p[{ns=1ss+1(a22q+A2q(a1,a2))}1q+{ns=1ss+1(a12q+A2q(a1,a2))}1q].

    Proof. The proof is direct consequence of Theorem 3.1, by setting x=a1+a22,α=1,μ=0 and Φ(x)=xm, and Φ(x)=1x, respectively.

    Proposition 4.2. Suppose all the assumptions of Theorem 3.2 are satisfied, then

    (1)

    |Am(a1,a2)Lmm(a1,a2)|a2a14(12)1p[ma2(m1)qnns=1(ss+112+1(s+1)(s+2))+mA(m1)q(a1,a2)nns=1(ss+112+1s+2)]1q+[ma1(m1)qnns=1(ss+112+1(s+1)(s+2))+mA(m1)q(a1,a2)nns=1(ss+112+1s+2)]1q,

    (2)

    |L1(a1,a2)A1(a1,a2)|a2a14(12)1p[a22qnns=1(ss+112+1(s+1)(s+2))+A2q(a1,a2)nns=1(ss+112+1s+2)]1q+[a12qnns=1(ss+112+1(s+1)(s+2))+A2q(a1,a2)nns=1(ss+112+1s+2)]1q.

    Proof. The proof is direct consequence of Theorem 3.2, by taking x=a1+a22,α=1,μ=0 and Φ(x)=xm, and Φ(x)=1x, respectively.

    In this last section, we discuss applications regarding bounded functions in absolute value of the results obtained from our main results. We suppose that the following conditions are satisfied:

    |Φ|ϝ1and|Φ|ϝ2.

    Applying the above conditions, we have the following results.

    Corollary 4.1. Under the assumptions of Theorem 3.1, the following inequality holds:

    |ABIαa2{Φ(a1+a2Θ)}+ABa1Iατ{Φ(a1+a2Θ)}Φ(a1+a2Θ)B(α)Γ(α)[(Θa1)α+(a2Θ)α+2(1α)Γ(α)]|1B(α)Γ(α)(αpαp+1)1p[(Θa1)α+1(2ϝq1nns=1ss+12uΘa1σ(σ+1)(σ+2))1q+(a2Θ)α+1(2ϝq1nns=1ss+12ua2Θσ(σ+1)(σ+2))1q].

    Corollary 4.2. Under the assumptions of Theorem 3.2, the following inequality holds:

    |ABIαa2{Φ(a1+a2Θ)}+ABa1Iατ{Φ(a1+a2Θ)}Φ(a1+a2Θ)B(α)Γ(α)[(Θa1)α+(a2Θ)α+2(1α)Γ(α)]|1B(α)Γ(α)(αα+1)11q[(Θa1)α+1[ϝq1nns=1(ss+11α+1+Γ(1+s)Γ(1+α)Γ(2+s+α))+ϝq1nns=1(ss+11α+1+1α+s+1)uΘa1σ(1(σ+1)(σ+2)1(1+α+σ)(2+α+σ)Γ(2+α)Γ(1+σ)Γ(3+α+σ))]1q+(a2Θ)α+1[ϝq1nns=1(ss+11α+1+Γ(1+s)Γ(1+α)Γ(2+s+α))+ϝq1nns=1(ss+11α+1+1α+s+1)ua2Θσ(1(σ+1)(σ+2)1(1+α+σ)(2+α+σ)Γ(2+α)Γ(1+σ)Γ(3+α+σ))]1q]

    Corollary 4.3. Under the assumptions of Theorem 3.3, the following inequality holds:

    |Φ(a1+a2Θ)(α+1)B(α)Γ(α)[(a2Θ)α+1(Θa1)α+1][ABIαa2{Φ(a1+a2Θ)}+ABa1Iαa1+a2Θ{Φ(a1+a2Θ)}]+2(1α)B(α)Φ(a1+a2Θ)+1B(α)Γ(α)[(Θa1)αΦ(a2)+(a2Θ)αΦ(a1)]|1(α+1)B(α)Γ(α)(p(α+1)p(α+1)+1)1p[(Θa1)2α+3(2ϝq2nns=1ss+12uΘa1σ(σ+1)(σ+2))1q+(a2Θ)2α+3(2ϝq2nns=1ss+12ua2Θσ(σ+1)(σ+2))1q]

    Corollary 4.4. Under the assumptions of Theorem 3.4, the following inequality holds:

    |Φ(a1+a2Θ)(α+1)B(α)Γ(α)[(a2Θ)α+1(Θa1)α+1][ABIαa2{Φ(a1+a2Θ)}+ABa1Iαa1+a2Θ{Φ(a1+a2Θ)}]+2(1α)B(α)Φ(a1+a2Θ)+1B(α)Γ(α)[(Θa1)αΦ(a2)+(a2Θ)αΦ(a1)]|1(α+1)B(α)Γ(α)(α+1α+2)11q[(Θa1)2α+3(ϝq2nns=1(111+s12+α+Γ(1+s)Γ(2+α)Γ(3+s+α))+ϝq2nns=1(111+s12+α+12+s+α)uΘa1σ(2(1+σ)(2+σ)1(2+α+σ)(3+α+σ)Γ(3+α)Γ(1+σ)Γ(4+α+σ)))1q+(a2Θ)2α+3(ϝq2nns=1(111+s12+α+Γ(1+s)Γ(2+α)Γ(3+s+α))+ϝq2nns=1(111+s12+α+12+s+α)ua2Θσ(2(1+σ)(2+σ)1(2+α+σ)(3+α+σ)Γ(3+α)Γ(1+σ)Γ(4+α+σ)))1q]

    In 2015 Caputo and Fabrizio suggested a new operator with fractional order, this derivative is based on the exponential kernel. Earlier this year 2016 Atangana and Baleanu developed another version which used the generalized Mittag-Leffler function as non-local and non-singular kernel. Both operators show some properties of filter. However, the Atangana and Baleanu version has in addition to this, all properties of fractional derivative. This shown effectiveness and advantages of the Atangana-Baleanu integral operators. Inspired by this great fact that own Atangana-Baleanu integral operators, we found two new fractional integral identities involving Atangana-Baleanu fractional integrals. Applying these identities as auxiliary results, we derived new fractional counterparts of classical inequalities essentially using first and second order differentiable higher order strongly n-polynomial convex functions. We have discussed several important special cases from our main results. The efficiency of our main results is demonstrated via special means and differentiable functions of first and second order that are in absolute value bounded. We will derive as future works several new fractional integral inequalities using Chebyshev, Markov, Young and Minkowski inequalities. Since the class of higher order strongly n-polynomial convex functions have large applications in many mathematical areas, they can be applied to obtain several results in convex analysis, special functions, quantum mechanics, related optimization theory, and mathematical inequalities and may stimulate further research in different areas of pure and applied sciences. Studies relating convexity, partial convexity, and preinvex functions (as contractive operators) may have useful applications in complex interdisciplinary studies, such as maximizing the likelihood from multiple linear regressions involving Gauss-Laplace distribution. For more details, see [49,50,51,52,53,54,55,56]. We hope that our ideas and techniques of this paper will inspire interested readers working in this field.

    This research was funded by Dirección de Investigación from Pontificia Universidad Católica del Ecuador in the research project entitled, "Some integrals inequalities and generalized convexity" (Algunas desigualdades integrales para funciones con algún tipo de convexidad generalizada y aplicaciones).

    The authors declare that they have no competing interests.



    [1] J. H. B. Kemperman, On the FKG-inequalities for measures on a partially ordered space, Indagat. Math., 80 (1977), 313–331. https://doi.org/10.1016/1385-7258(77)90027-0 doi: 10.1016/1385-7258(77)90027-0
    [2] T. Z. Hu, Negatively superadditive dependence of random variables with applications, Chinese J. Appl. Probab. Statist., 16 (2000), 133–144.
    [3] K. J. Dev, F. Proschan, Negative association of random variables with applications, Ann. Statist., 11 (1983), 286–295. https://doi.org/10.1214/aos/1176346079 doi: 10.1214/aos/1176346079
    [4] T. C. Christofides, E. Vaggelatou, A connection between supermodular ordering and positive/negative association, J. Multivariate Anal., 88 (2004), 138–151. https://doi.org/10.1016/s0047-259x(03)00064-2 doi: 10.1016/s0047-259x(03)00064-2
    [5] N. Eghbal, M. Amini, A. Bozorgnia, Some maximal inequalities for quadratic forms of negative superadditive dependence random variables, Statist. Probabil. Lett., 80 (2010), 587–591. https://doi.org/10.1016/j.spl.2009.12.014 doi: 10.1016/j.spl.2009.12.014
    [6] Y. Shen, X. J. Wang, W. Z. Yang, S. H. Hu, Almost sure convergence theorem and strong stability for weighted sums of NSD random variables, Acta Math. Sin., 29 (2013), 743–756. https://doi.org/10.1007/s10114-012-1723-6 doi: 10.1007/s10114-012-1723-6
    [7] X. J. Wang, X. Deng, L. L. Zheng, S. H. Hu, Complete convergence for arrays of rowwise negatively superadditive-dependent random variables and its applications, Statistics, 48 (2014), 834–850. https://doi.org/10.1080/02331888.2013.800066 doi: 10.1080/02331888.2013.800066
    [8] N. Eghbal, M. Amini, A. Bozorgnia, On the Kolmogorov inequalities for quadratic forms of dependent uniformly bounded random variables, Scienceasia, 81 (2011), 1112–1120. https://doi.org/10.1016/j.spl.2011.03.005 doi: 10.1016/j.spl.2011.03.005
    [9] X. J. Wang, A. T. Shen, Z. Y. Chen, S. H. Hu, Complete convergence for weighted sums of NSD random variables and its application in the EV regression model, Test, 24 (2015), 166–184. https://doi.org/10.1007/s11749-014-0402-6 doi: 10.1007/s11749-014-0402-6
    [10] Y. C. Yu, H. C. Hu, L. Liu, S. Y. Huang, M-test in linear models with negatively superadditive dependent errors, J. Inequal. Appl., 235 (2017), 235. https://doi.org/10.1186/s13660-017-1509-6 doi: 10.1186/s13660-017-1509-6
    [11] A. T. Shen, X. H. Wang, Kaplan-Meier estimator and hazard estimator for censored negatively superadditive dependent data, Statistics, 50 (2016), 377–388. https://doi.org/10.1080/02331888.2015.1038269 doi: 10.1080/02331888.2015.1038269
    [12] H. Naderi, F. Boukhari, P. Matula, A note on the weak law of large numbers for weighted negatively superadditive dependent random variables, Commun. Stat. Theor. M., 51 (2022), 7465–7475. https://doi.org/10.1080/03610926.2021.1873377 doi: 10.1080/03610926.2021.1873377
    [13] S. C. Ta, C. M. Tran, D. V. Le, On the almost sure convergence for sums of negatively superadditive dependent random vectors in Hilbert spaces and its application, Commun. Stat. Theor. M., 49 (2020), 2770–2786. https://doi.org/10.1080/03610926.2019.1584304 doi: 10.1080/03610926.2019.1584304
    [14] X. H. Wang, S. H. Hu, On the strong consistency of M-estimates in linear models for negatively superadditive dependent errors, Aust. Nz. J. Stat., 57 (2015), 259–274. https://doi.org/10.1111/anzs.12117 doi: 10.1111/anzs.12117
    [15] Y. Wu, X. J. Wang, S. H. Hu, Complete convergence for arrays of rowwise negatively superadditive-dependent random variables and its applications, Appl. Math. Ser. B, 31 (2016), 439–457. https://doi.org/10.1007/s11766-016-3406-z doi: 10.1007/s11766-016-3406-z
    [16] X. H. Wang, X. Q. Li, S. H. Hu, On the complete convergence of weighted sums for an array of rowwise negatively superadditive dependent random variables, Scienceasia, 42 (2016), 66–74. https://doi.org/10.2306/scienceasia1513-1874.2016.42.066 doi: 10.2306/scienceasia1513-1874.2016.42.066
    [17] X. Deng, X. J. Wang, Y. Wu, Y. Ding, Complete moment convergence and complete convergence for weighted sums of NSD random variables, Racsam. Rev. R. Acad. A., 110 (2016), 97–120. https://doi.org/10.1007/s13398-015-0225-7 doi: 10.1007/s13398-015-0225-7
    [18] A. T. Shen, X. H. Wang, H. Y. Zhu, Convergence properties for weighted sums of NSD random variables, Commun. Stat. Theor. M., 45 (2016), 2402–2412. https://doi.org/10.1080/03610926.2014.881492 doi: 10.1080/03610926.2014.881492
    [19] A. Kheyri, M. Amini, H. Jabbari, A. Bozorgnia, Kernel density estimation under negative superadditive dependence and its application for real data, J. Stat. Comput. Sim., 89 (2019), 2373–2392. https://doi.org/10.1080/00949655.2019.1619738 doi: 10.1080/00949655.2019.1619738
    [20] B. Meng, Q. Y. Wu, D. C. Wang, On the strong convergence for weighted sums of negatively superadditive dependent random variables, J. Inequal. Appl., 269 (2017), 269. https://doi.org/10.1080/03610918.2022.2093371 doi: 10.1080/03610918.2022.2093371
    [21] K. Bertin, S. Torres, L. Viitasaari, Least-square estimators in linear regression models under negatively superadditive dependent random observations, Statistics, 55 (2021), 1018–1034. https://doi.org/10.1080/02331888.2021.1993854 doi: 10.1080/02331888.2021.1993854
    [22] M. Q. Chen, K. Chen, Z. J. Wang, Z. L. Lu, X. J. Wang, Complete moment convergence for partial sums of arrays of rowwise negatively superadditive dependent random variables, Commun. Stat. Theor. M., 49 (2020), 1158–1173. https://doi.org/10.1080/03610926.2018.1554136 doi: 10.1080/03610926.2018.1554136
    [23] Y. C. Yu, X. S. Liu, L. Liu, W. S. Liu, On adaptivity of wavelet thresholding estimators with negatively super-additive dependent noise, Math. Slovaca, 69 (2019), 1485–1500. https://doi.org/10.1515/ms-2017-0324 doi: 10.1515/ms-2017-0324
    [24] W. Z. Yang, L. Yang, D. Wei, S. H. Hu, The laws of large numbers for Pareto-type random variables with infinite means, Commun. Stat. Theor. M., 48 (2019), 3044–3054. https://doi.org/10.1080/03610926.2018.1473602 doi: 10.1080/03610926.2018.1473602
    [25] A. Adler, Laws of large numbers for two tailed Pareto random variables, Probab. Math. Stat., 28 (2008), 121–128.
    [26] A. Adler, An exact weak law of large numbers, Bull. Inst. Math. Acad., 7 (2012), 417–422.
    [27] A. Adler, Exact weak laws and one side strong laws, Bull. Inst. Math. Acad., 12 (2017), 103–124.
    [28] A. Adler, One sided strong laws for random variables with infnite mean, Open Math., 15 (2017), 828–832. https://doi.org/10.1515/math-2017-0070 doi: 10.1515/math-2017-0070
    [29] K. Matsumoto, T. Nakata, Limit theorems for a generalized Feller game, J. Appl. Probab., 50 (2013), 54–63. https://doi.org/10.1239/jap/1363784424 doi: 10.1239/jap/1363784424
    [30] T. Nakata, Limit theorems for nonnegative independent random variables with truncation, Acta Math. Hung., 145 (2015), 1–16. https://doi.org/10.1007/s10474-014-0474-5 doi: 10.1007/s10474-014-0474-5
    [31] T. Nakata, Weak laws of large numbers for weighted independent random variables with infnite mean, Statist. Probabil. Lett., 109 (2016), 124–129. https://doi.org/10.1016/j.spl.2015.11.017 doi: 10.1016/j.spl.2015.11.017
    [32] Q. M. Shao, A comparison theorem on moment inequalities between negatively associated and independent random variables, J. Theor. Probab., 13 (2000), 343–355. https://doi.org/10.1023/A:1007849609234 doi: 10.1023/A:1007849609234
  • This article has been cited by:

    1. Miguel Vivas–Cortez, Muhammad Zakria Javed, Muhammad Uzair Awan, Muhammad Aslam Noor, Silvestru Sever Dragomir, Bullen-Mercer type inequalities with applications in numerical analysis, 2024, 96, 11100168, 15, 10.1016/j.aej.2024.03.093
  • 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(1713) PDF downloads(318) Cited by(0)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog