We extend the notion of classical metric geometric mean (MGM) for positive definite matrices of the same dimension to those of arbitrary dimensions, so that usual matrix products are replaced by semi-tensor products. When the weights are arbitrary real numbers, the weighted MGMs possess not only nice properties as in the classical case, but also affine change of parameters, exponential law, and cancellability. Moreover, when the weights belong to the unit interval, the weighted MGM has remarkable properties, namely, monotonicity and continuity from above. Then we apply a continuity argument to extend the weighted MGM to positive semidefinite matrices, here the weights belong to the unit interval. It turns out that this matrix mean posses rich algebraic, order, and analytic properties, such as, monotonicity, continuity from above, congruent invariance, permutation invariance, affine change of parameters, and exponential law. Furthermore, we investigate certain equations concerning weighted MGMs of positive definite matrices. It turns out that such equations are always uniquely solvable with explicit solutions. The notion of MGMs can be applied to solve certain symmetric word equations in two letters.
Citation: Arnon Ploymukda, Pattrawut Chansangiam. Metric geometric means with arbitrary weights of positive definite matrices involving semi-tensor products[J]. AIMS Mathematics, 2023, 8(11): 26153-26167. doi: 10.3934/math.20231333
[1] | Sitalakshmi Venkatraman, Ramanathan Venkatraman . Big data security challenges and strategies. AIMS Mathematics, 2019, 4(3): 860-879. doi: 10.3934/math.2019.3.860 |
[2] | Atiqe Ur Rahman, Muhammad Saeed, Mazin Abed Mohammed, Alaa S Al-Waisy, Seifedine Kadry, Jungeun Kim . An innovative fuzzy parameterized MADM approach to site selection for dam construction based on sv-complex neutrosophic hypersoft set. AIMS Mathematics, 2023, 8(2): 4907-4929. doi: 10.3934/math.2023245 |
[3] | Pengyu Liu, Jie Jian . Effects of network topology and trait distribution on collective decision making. AIMS Mathematics, 2023, 8(5): 12287-12320. doi: 10.3934/math.2023619 |
[4] | Shahryar Sorooshian, Seyedh Mahboobeh Jamali, Nader Ale Ebrahim . Performance of the decision-making trial and evaluation laboratory. AIMS Mathematics, 2023, 8(3): 7490-7514. doi: 10.3934/math.2023376 |
[5] | Erdal Bayram, Gülşah Çelik, Mustafa Gezek . An advanced encryption system based on soft sets. AIMS Mathematics, 2024, 9(11): 32232-32256. doi: 10.3934/math.20241547 |
[6] | Esmail Hassan Abdullatif Al-Sabri, Muhammad Rahim, Fazli Amin, Rashad Ismail, Salma Khan, Agaeb Mahal Alanzi, Hamiden Abd El-Wahed Khalifa . Multi-criteria decision-making based on Pythagorean cubic fuzzy Einstein aggregation operators for investment management. AIMS Mathematics, 2023, 8(7): 16961-16988. doi: 10.3934/math.2023866 |
[7] | Muhammad Arshad, Muhammad Saeed, Khuram Ali Khan, Nehad Ali Shah, Wajaree Weera, Jae Dong Chung . A robust MADM-approach to recruitment-based pattern recognition by using similarity measures of interval-valued fuzzy hypersoft set. AIMS Mathematics, 2023, 8(5): 12321-12341. doi: 10.3934/math.2023620 |
[8] | Muhammad Ihsan, Muhammad Saeed, Atiqe Ur Rahman, Hüseyin Kamacı, Nehad Ali Shah, Wajaree Weera . An MADM-based fuzzy parameterized framework for solar panels evaluation in a fuzzy hypersoft expert set environment. AIMS Mathematics, 2023, 8(2): 3403-3427. doi: 10.3934/math.2023175 |
[9] | Esra Korkmaz . An innovative algorithm based on weighted fuzzy soft multisets and its application in selecting optimal construction materials. AIMS Mathematics, 2024, 9(10): 27512-27534. doi: 10.3934/math.20241336 |
[10] | Jia-Bao Liu, Rashad Ismail, Muhammad Kamran, Esmail Hassan Abdullatif Al-Sabri, Shahzaib Ashraf, Ismail Naci Cangul . An optimization strategy with SV-neutrosophic quaternion information and probabilistic hesitant fuzzy rough Einstein aggregation operator. AIMS Mathematics, 2023, 8(9): 20612-20653. doi: 10.3934/math.20231051 |
We extend the notion of classical metric geometric mean (MGM) for positive definite matrices of the same dimension to those of arbitrary dimensions, so that usual matrix products are replaced by semi-tensor products. When the weights are arbitrary real numbers, the weighted MGMs possess not only nice properties as in the classical case, but also affine change of parameters, exponential law, and cancellability. Moreover, when the weights belong to the unit interval, the weighted MGM has remarkable properties, namely, monotonicity and continuity from above. Then we apply a continuity argument to extend the weighted MGM to positive semidefinite matrices, here the weights belong to the unit interval. It turns out that this matrix mean posses rich algebraic, order, and analytic properties, such as, monotonicity, continuity from above, congruent invariance, permutation invariance, affine change of parameters, and exponential law. Furthermore, we investigate certain equations concerning weighted MGMs of positive definite matrices. It turns out that such equations are always uniquely solvable with explicit solutions. The notion of MGMs can be applied to solve certain symmetric word equations in two letters.
Fractional differential equations with various types of fractional derivatives arise in modeling some dynamical processes (see, for example, [15] for the globally projective synchronization of Caputo fractional-order quaternion-valued neural networks with discrete and distributed delays, [18] for the quasi-uniform synchronization issue for fractional-order neural networks with leakage and discrete delays and [11] for Mittag-Leffler stability and adaptive impulsive synchronization of fractional order neural networks in quaternion field). In contrast to the classical derivative the fractional derivative is nonlocal and it depends significantly on its lower limit. As it is mentioned in [13], this leads to some obstacles for studying impulsive fractional differential equations.
Since many phenomena are characterized by abrupt changes at certain moments it is important to consider differential equations with impulses. In the literature there are two main approaches used to introduce impulses to fractional equations:
(i) With a fixed lower limit of the fractional derivative at the initial time- the fractional derivative of the unknown function has a lower limit equal to the initial time point over the whole interval of study;
(ii) With a changeable lower limit of the fractional derivative at each time of impulse- the fractional derivative on each interval between two consecutive impulses is changed because the lower limit of the fractional derivative is equal to the time of impulse.
Both interpretations of impulses are based on corresponding interpretations of impulses in ordinary differential equation, which coincide in the case of integer derivatives. However this is not the case for fractional derivatives. In the literature both types of interpretations are discussed and studied for Caputo fractional differential equations of order α∈(0,1). We refer the reader to the papers [6,7,12,13,16] as well as the monograph [3].
We note in the case of the Caputo fractional derivative there is a similarity between both the initial conditions and the impulsive condition between fractional equations and ordinary equations (see, for example, [10] concerning the impulsive control law for the Caputo delay fractional-order neural network model). However for Riemann-Liouville fractional differential equations both the initial condition and impulsive conditions have to be appropriately given (which is different in the case of ordinary derivatives as well as the case of Caputo fractional derivatives). Riemann-Liouville fractional differential equations are considered, for example, in [1,2] for integral presentation of solutions in the case of the fractional order α∈(0,1), [5,8,17] for the case of the fractional order α∈(1,2).
In [14] the authors studied the following coupled impulsive fractional integro-differential equations with Riemann-Liouville derivatives of the form
{{Dαu(t)−ϕ1(t,Iαu(t),Iβv(t))=0 for t∈I, t≠ti, i=1,2,…,p,Δu(tj)−Ej(u(tj))=0, Δu′(tj)−E∗j(u(tj))=0, j=1,2,…,p,ν1Dα−2u(t)|t=0=u1, μ1u(t)|t=T+ν2Iα−1u(t)|t=T=u2,{Dβv(t)−ϕ2(t,Iαu(t),Iβv(t))=0 for t∈I, t≠tk, k=1,2,…,q,Δv(tk)−Ek(v(tk))=0, Δv′(tk)−E∗j(v(tk))=0, k=1,2,…,q,ν3Dβ−2v(t)|t=0=v1, μ2v(t)|t=T+ν4Iβ−1v(t)|t=T=v2, | (1.1) |
where α,β∈(1,2], I=[0,T], ϕ1,ϕ2:I×R×R→R are continuous functions, Δu(tj)=u(t+j)−u(t−j), Δu′(tj)=u′(t+j)−u′(t−j), Δv(tk)=v(t+k)−v(t−k), Δv′(tj)=v′(t+k)−v′(t−k), where u(t+j),v(t+k) and v(t−j),v(t−k) are the right limits and left limits respectively, Ej,E∗j,Ek,E∗k:R→R are continuous functions, and Dα,Iα are the α-order Riemann-Liouville fractional derivative and integral operators respectively and Dβ−2=I2−β.
Since fractional integrals and derivatives have memories, and their lower limits are very important we will use the notations RLDαa,t and Iβa,t, respectively, instead of Dα and Iβ, i.e. the Riemann-Liouville fractional derivative is defined by (see, for example [9])
RLDαa,tu(t)=1Γ(2−α)(ddt)2∫tau(s)(t−s)α−1ds, t>a, α∈(1,2), | (1.2) |
and the Riemann-Liouville fractional integral Iβa,t of order α>0 is defined by (see, for example, [9])
Iβa,tu(t)=1Γ(β)∫ta(t−s)β−1u(s)ds,t>a, | (1.3) |
where a≥0, β>0 are given numbers.
Note there are some unclear parts in the statement of coupled impulsive fractional integro-differential equations with Riemann-Liouville derivatives (1.1), such as:
a). The presence of two different integers p and q in (1.1) leads to different domains of both the unknown functions u and v. For example, in Corollary 1 [14] the solutions u(t) and v(t) are defined on [0,tp+1] and [0,tq+1] respectively, which causes some problems in the definitions of formulas (3.7) or (3.8) ([14]);
b). The impulsive functions Ej,Ek, j=1,2,…,p, k=1,2,…,q are assumed different but they are not (it is clear for example, for j=k=1). The same is about the functions E∗j,E∗k, j=1,2,…,p, k=1,2,…,q.
In this paper we sort out the above mentioned points by setting up the cleared statement of the boundary value problem with the Riemann-Liouville (RL) fractional integral for the impulsive Riemann-Liouville fractional differential equation studied in [14], and we prove a new the integral presentations of the solutions. To be more general, we study two different interpretations for the presence of impulses in fractional differential equations. The first one is the case of the fixed lower limit of the RL fractional derivative at the initial time 0 and the second one is the case of the changed lower limit of the fractional RL derivative at any point of impulse. In both cases the integral presentation of the solution is provided.
Define two different sequences of points of impulses
0=t0<t1<t2<⋯<tp<tp+1=T and 0=τ0<τ1<τ2<⋯<τq<τq+1=T, |
where p,q are given natural numbers.
We will consider the following nonlinear boundary value problem for the coupled impulsive fractional integro-differential equations with Riemann-Liouville derivatives with a lower limit at 0
{{RLDα0,tu(t)−ϕ1(t,Iα0,tu(t),Iβ0,tv(t))=0 for t∈I, t≠ti, i=1,2,…,p,Δu(tj)−Ej(u(tj))=0, Δu′(tj)−E∗j(u(tj))=0, j=1,2,…,p,−I2−α0,tu(t)|t=0=u1, μ1u(t)|t=T+ν1Iα−10,tu(t)|t=T=u2,{RLDβ0,tv(t)−ϕ2(t,Iα0,tu(t),Iβ0,tv(t))=0 for t∈I, t≠τk, k=1,2,…,q,Δv(τk)−Sk(v(τk))=0, Δv′(τk)−S∗k(v(τk))=0, k=1,2,…,q,I2−β0,tv(t)|t=0=v1, μ2v(t)|t=T+ν2Iβ−10,tv(t)|t=T=v2, | (2.1) |
where α,β∈(1,2], I=[0,T], ϕ1,ϕ2:I×R×R→R are continuous functions, Δu(tj)=u(t+j)−u(t−j), Δu′(tj)=u′(t+j)−u′(t−j), Δv(τk)=v(τ+k)−v(τ−k), Δv′(τk)=v′(τ+k)−v′(τ−k), where u(t+j),v(τ+k),u′(t+j),v′(τ+k) and u(t−j), v(t−k), u′(t−j), v′(t−k) are the right limits and left limits respectively, Ej,E∗j,Sk,S∗k:R→R are continuous functions, and RLDβ0,t,Iα0,t are the α-order Riemann-Liouville fractional derivative and integral operators, respectively, μi,νi,uk,vk, i=1,2, are given constants.
In the statement of the problem (2.1) some parts of (1.1) are cleared: there are two different points of impulses; the lower limits of the fractional integrals and fractional integrals are written; different functions at different points of impulses are used.
In our proofs we will use the following well known properties for fractional integrals (see, for example [9]).
Iαa,tIβa,tu(t)=Iα+βa,tu(t), α,β>0,Iαa,t(t−a)q=Γ(q+1)Γ(q+α+1)(t−a)q+α, α>0, q>−1. t>a. | (2.2) |
We will apply the following auxiliary result which is a generalization of the result in [4] for an arbitrary lower limit of the fractional derivative:
Lemma 1. ([4]). The general solution of the Riemann-Liouville fractional differential equation
RLDαa,tw(t)=g(t), t∈(a,T], α∈(1,2) | (2.3) |
is given by
w(t)=c1(t−a)α−1+c0(t−a)α−2+Iαa,t g(t)=c1(t−a)α−1+c0(t−a)α−2+1Γ(α)∫ta(t−s)α−1g(s)ds, t∈(a,T], | (2.4) |
where c0,c1,a≥0 are arbitrary real constants.
We will consider an appropriate boundary value problem for a scalar impulsive linear equation, we will prove a formula for its solution and later we will apply it to obatin the main result.
Consider the following boundary value problem for the linear impulsive fractional differential equation with Riemann-Liouville derivatives of the form
RLDα0,tu(t)=f(t), t∈(0,T], t≠tj, j=1,2,…p, α∈(1,2),Δu(tj)=Ej(u(tj)), Δu′(tj)=E∗j(u(tj)) j=1,2,…,p,I2−α0,tu(t)|t=0=u1, μ1u(t)|t=T+ν1Iα−10,tu(t)|t=T=u2, | (2.5) |
where f: [0,T]→R is a continuous function, Ej,E∗j:R→R are continuous functions, u1,u2∈R.
Lemma 2. The solution of (2.5) satisfies the integral equation
u(t)={c0tα−1+u1Γ(α−1)tα−2+1Γ(α)∫t0(t−s)α−1f(s)ds,t∈(0,t1],c0tα−1+(∑jk=1[t2−αkE∗k(u(tk))]−(α−2)∑jk=1[t1−αkEk(u(tk))])tα−1 +(u1Γ(α−1)+(α−1)∑jk=1t2−αkEk(u(tk))−∑jk=1t3−αkE∗k(u(tk)))tα−2 +1Γ(α)∫t0(t−s)α−1f(s)ds,t∈(tj,tj+1], |
where
c0=−(p+1∑k=1[t2−αkE∗k(u(tk))]−(α−2)p+1∑k=1[t1−αkEk(u(tk))]) −(u1Γ(α−1)+(α−1)p+1∑k=1t2−αkEk(u(tk))−p+1∑k=1t3−αkE∗k(u(tk)))T−1 −1Tα−1Γ(α)∫T0(T−s)α−1f(s)ds −ν1μ1Tα−1Γ(α−1)∫T0(T−s)α−2u(s)ds+u2μ1. | (2.6) |
Proof. We will use induction.
For t∈(0,t1] we apply Lemma 1 with a=0 and we get
u(t)=c0tα−1+c1tα−2+1Γ(α)∫t0(t−s)α−1f(s)ds | (2.7) |
and
u′(t)=c0(α−1)tα−2+c1(α−2)tα−3+α−1Γ(α)∫t0(t−s)α−2f(s)ds. | (2.8) |
From the initial condition I2−α0,tu(t)|t=0=u1 and equalities (2.2), (2.7) we get
I2−α0,tu(t)|t=0=c0Γ(α)Γ(1)t|t=0+c1Γ(α−1)Γ(0)|t=0+I20,tf(t)|t=0=c1Γ(α−1), |
i.e. c1=u1Γ(α−1).
For t∈(t1,t2] by Lemma 1 with a=0 we get
u(t)=b0tα−1+b1tα−2+1Γ(α)∫t0(t−s)α−1f(s)ds | (2.9) |
and
u′(t)=b0(α−1)tα−2+b1(α−2)tα−3+1Γ(α−1)∫t0(t−s)α−2f(s)ds. | (2.10) |
From the impulsive condition u(t1+0)−u(t1−0)=E1(u(t1)) we obtain
u(t+1)=b0tα−11+b1tα−21+1Γ(α)∫t10(t1−s)α−1f(s)ds−c0tα−11−u1Γ(α−1)tα−21−1Γ(α)∫t10(t1−s)α−1f(s)ds=(b0−c0)tα−11+(b1−u1Γ(α−1))tα−21=E1(u(t1)). | (2.11) |
and from the impulsive condition u′(t1+0)−u′(t1)=E∗1(u(t1)) we get
u′(t+1)=b0(α−1)tα−21+b1(α−2)tα−31+1Γ(α−1)∫t10(t1−s)α−2f(s)ds−c0(α−1)tα−21−u1Γ(α−1)(α−2)tα−31−1Γ(α−1)∫t10(t1−s)α−2f(s)ds=(b0−c0)(α−1)tα−21+(b1−u1Γ(α−1))(α−2)tα−31=E∗1(u(t1)). | (2.12) |
Thus we get the linear system w.r.t. b0 and b1
(b0−c0)(α−1)tα−21+(b1−u1Γ(α−1))(α−2)tα−31=E∗1(u(t1))(b0−c0)tα−11+(b1−u1Γ(α−1))tα−21=E1(u(t1)) |
or
b0=c0+t2−α1E∗1(u(t1))−(α−2)t1−α1E1(u(t1)) |
b1=(α−1)t2−α1E1(u(t1))−t3−α1E∗1(u(t1))+u1Γ(α−1). |
Therefore,
u(t)=c0tα−1+(t2−α1E∗1(u(t1))−(α−2)t1−α1E1(u(t1)))tα−1 +((α−1)t2−α1E1(u(t1))−t3−α1E∗1(u(t1))+u1Γ(α−1))tα−2 +1Γ(α)∫t0(t−s)α−1f(s)ds, t∈(t1,t2]. | (2.13) |
Assume the integral presentation of u(t) is correct on (tj−1,tj], i.e
u(t)=c0tα−1+(j−1∑k=1t2−αkE∗k(u(tk))−(α−2)j−1∑k=1t1−αkEk(u(tk)))tα−1 +((α−1)j−1∑k=1t2−αkEk(u(tk))−j−1∑k=1t3−αkE∗k(u(tk))+u1Γ(α−1))tα−2 +1Γ(α)∫t0(t−s)α−1f(s)ds, t∈(tj−1,tj]. | (2.14) |
Denote
m0=c0+j−1∑k=1t2−αkE∗k(u(tk))−(α−2)j−1∑k=1t1−αkEk(u(tk)) |
and
m1=(α−1)j−1∑k=1t2−αkEk(u(tk))−j−1∑k=1t3−αkE∗k(u(tk))+u1Γ(α−1). |
Let t∈(tj,tj+1], j=2,…,p,. By Lemma 1 with a=0 we get
u(t)=k0tα−1+k1tα−2+1Γ(α)∫t0(t−s)α−1f(s)ds, | (2.15) |
and
u′(t)=k0(α−1)tα−2+k1(α−2)tα−3+1Γ(α−1)∫t0(t−s)α−2f(s)ds. | (2.16) |
From the impulsive conditions and the equality (2.14) we obtain the linear system w.r.t. k0 and k1
(k0−m0)(α−1)tα−2j+(k1−m1)(α−2)tα−3j=E∗2(u(t2))(k0−m0)tα−1j+(k1−m1)tα−2j=Ej(u(tj)) |
or
k0=c0+j∑k=1[t2−αkE∗k(u(tk))]−(α−2)j∑k=1[t1−αkEk(u(tk))] |
k1=u1Γ(α−1)+(α−1)j∑k=1t2−αkEk(u(tk))−j∑k=1t3−αkE∗k(u(tk)). |
Therefore,
u(t)=c0tα−1+(j∑k=1[t2−αkE∗k(u(tk))]−(α−2)j∑k=1[t1−αkEk(u(tk))])tα−1 +(u1Γ(α−1)+(α−1)j∑k=1t2−αkEk(u(tk))−j∑k=1t3−αkE∗k(u(tk)))tα−2 +1Γ(α)∫t0(t−s)α−1f(s)ds, t∈(tj,tj+1], j=1,2,…,p. | (2.17) |
From the boundary condition μ1u(t)|t=T+ν1Iα−1u(t)|t=T=u2 we get
Iα−1u(t)|t=T=1Γ(α−1)∫T0(T−s)α−2u(s)ds |
and
μ1u(t)|t=T+ν1Iα−1u(t)|t=T=μ1c0Tα−1+μ1(p+1∑k=1[t2−αkE∗k(u(tk))]−(α−2)p+1∑k=1[t1−αkEk(u(tk))])Tα−1 +μ1(u1Γ(α−1)+(α−1)p+1∑k=1t2−αkEk(u(tk))−p+1∑k=1t3−αkE∗k(u(tk)))Tα−2 +μ11Γ(α)∫T0(T−s)α−1ϕ1(s,Iα0,su(s),Iβ0,sv(s))ds +ν11Γ(α−1)∫T0(T−s)α−2u(s)ds=u2. | (2.18) |
From (2.18) we have (2.6).
We will give an example to illustrate the claim of Lemma 2.
Example 1. Consider the following boundary value problem for the scalar RL fractional differential equation with an impulse at t=1
RLD1.50,tu(t)=t, t∈(0,1]∪(1,2],Δu(1)=1, Δu′(1)=0,I0.50,tu(t)|t=0=0, u(t)|t=2+I0.50,tu(t)|t=2=1. | (2.19) |
The solution of (2.19) satisfies the integral equation
u(t)={c0t0.5+t−0.5Γ(0.5)+0.266667t2.5Γ(1.5),t∈(0,1]c0t0.5+0.5t0.5+(1Γ(0.5)+0.5)t−0.5+0.266667t2.5Γ(1.5),t∈(1,2], |
where
c0=0.25−1.5084920.5Γ(1.5)−120.5Γ(0.5)∫20(2−s)−0.5u(s)ds. | (2.20) |
Consider the boundary value problem for the nonlinear impulsive fractional integro-differential equations with Riemann-Liouville derivatives of the form
RLDα0,tu(t)=ϕ1(t,Iα0,tu(t)), t∈(0,T], t≠tj, j=1,2,…p, α∈(1,2)Δu(tj)=Ej(u(tj)), Δu′(tj)=E∗j(u(tj)) j=1,2,…,p,I2−α0,tu(t)|t=0=u1, μ1u(t)|t=T+ν1Iα−10,tu(t)|t=T=u2, | (2.21) |
where ϕ1: [0,T]×R→R is a continuous function, Ej,E∗j:R→R are continuous functions, u1,u2∈R.
Corollary 1. The solution of (2.21) satisfies the integral equation
u(t)={c0tα−1+u1Γ(α−1)tα−2+1Γ(α)∫t0(t−s)α−1ϕ1(s,Iα0,su(s),)ds, t∈(0,t1]c0tα−1+(∑jk=1[t2−αkE∗k(u(tk))]−(α−2)∑jk=1[t1−αkEk(u(tk))])tα−1 +(u1Γ(α−1)+(α−1)∑jk=1t2−αkEk(u(tk))−∑jk=1t3−αkE∗k(u(tk)))tα−2 +1Γ(α)∫t0(t−s)α−1ϕ1(s,Iα0,su(s))ds, t∈(tj,tj+1], | (2.22) |
where
c0=−(p+1∑k=1[t2−αkE∗k(u(tk))]−(α−2)p+1∑k=1[t1−αkEk(u(tk))]) −(u1Γ(α−1)+(α−1)p+1∑k=1t2−αkEk(u(tk))−p+1∑k=1t3−αkE∗k(u(tk)))T−1 −1Tα−1Γ(α)∫T0(T−s)α−1ϕ1(s,Iα0,su(s),Iβ0,sv(s))ds −ν1μ1Tα−1Γ(α−1)∫T0(T−s)α−2u(s)ds+u2μ1. | (2.23) |
The proof of Corollary 1 is similar to the one of Lemma 2 with f(t)=ϕ1(t,Iα0,tu(t)) and we omit it.
Remark 1. Corollary 1 and the integral presentation (2.22) correct Theorem 3.1 and the formula (3.2) [14]. The main mistake in the proof of formula (3.2) [14] is the incorrect application of Lemma 1 with a=0 on (t1,t2] and taking the lower limit of the integral in (3.5) [14] incorrectly at t1(σ1) instead of 0. A similar comment applies to all the other intervals (tj,tj+1].
Following the proof of Lemma 2 and the integral presentation (2.22) of problem (2.21), we have the following result:
Theorem 1. The solution of (2.1) satisfies the integral equations
u(t)={c0tα−1+u1Γ(α−1)tα−2+1Γ(α)∫t0(t−s)α−1ϕ1(s,Iα0,su(s),Iβ0,sv(s))ds, t∈(0,t1],c0tα−1+(j∑k=1[t2−αkE∗k(u(tk))]−(α−2)j∑k=1[t1−αkEk(u(tk))])tα−1 +(u1Γ(α−1)+(α−1)j∑k=1t2−αkEk(u(tk))−j∑k=1t3−αkE∗k(u(tk)))tα−2 +1Γ(α)∫t0(t−s)α−1ϕ1(s,Iα0,su(s),Iβ0,sv(s))ds, t∈(tj,tj+1], | (2.24) |
and
v(t)={b0tα−1+v1Γ(α−1)tα−2+1Γ(α)∫t0(t−s)α−1ϕ2(s,Iα0,su(s),Iβ0,sv(s))ds, t∈(0,t1]b0tα−1+(j∑k=1[t2−αkS∗k(v(tk))]−(α−2)j∑k=1[t1−αkSk(v(tk))])tα−1 +(u1Γ(α−1)+(α−1)j∑k=1t2−αkSk(v(tk))−j∑k=1t3−αkS∗k(v(tk)))tα−2 +1Γ(α)∫t0(t−s)α−1ϕ2(s,Iα0,su(s),Iβ0,sv(s))ds, t∈(tj,tj+1], | (2.25) |
where
c0=−(p+1∑k=1[t2−αkE∗k(u(tk))]−(α−2)p+1∑k=1[t1−αkEk(u(tk))]) −(u1Γ(α−1)+(α−1)p+1∑k=1t2−αkEk(u(tk))−p+1∑k=1t3−αkE∗k(u(tk)))T−1 −1Tα−1Γ(α)∫T0(T−s)α−1ϕ1(s,Iα0,su(s),Iβ0,sv(s))ds −ν1μ1Tα−1Γ(α−1)∫T0(T−s)α−2u(s)ds+u2μ1b0=−(p+1∑k=1[t2−αkS∗k(v(tk))]−(α−2)p+1∑k=1[t1−αkSk(v(tk))]) −(v1Γ(α−1)+(α−1)p+1∑k=1t2−αkSk(v(tk))−p+1∑k=1t3−αkS∗k(v(tk)))T−1 −1Tα−1Γ(α)∫T0(T−s)α−1ϕ2(s,Iα0,su(s),Iβ0,sv(s))ds −ν2μ2Tα−1Γ(α−1)∫T0(T−s)α−2v(s)ds+v2μ2. | (2.26) |
The proof is similar to the one of Lemma 2 applied twice to each of the both components u and v of the coupled system (2.1) for impulsive points ti, i=1,2,…,p and τi, i=1,2,…,p and the functions f(t)=ϕ1(t,Iα0,tu(t),Iβ0,tv(t)) and f(t)=ϕ2(t,Iα0,tu(t),Iβ0,tv(t)) respectively.
Remark 2. Note the integral presentation (2.24), (2.25) of the solutions of the coupled system is the correction of Corollary 1 and integral presentation (3.7), (3.8) in [14].
Consider the following nonlinear boundary value problem for the coupled impulsive fractional integro-differential equations with Riemann-Liouville derivatives with lower limits at impulsive points ti, i=0,1,2,…,p−1 and τk, k=0,1,2,…,q−1, respectively,
{{RLDαti,tu(t)−ϕ1(t,Iα0,tu(t),Iβ0,tv(t))=0 for t∈(ti,ti+1] i=0,1,2,…,p,I2−αtj,tu′(t)|t=tj=Pju(tj)+Qj, j=1,2,…,p,I2−α0,tu(t)|t=0=u1, μ1u(t)|t=T+ν1Iα−10,tu(t)|t=T=u2,{RLDβτk,tv(t)−ϕ2(t,Iα0,tu(t),Iβ0,tv(t))=0 for t∈(τk,τk+1] k=0,1,2,…,q,I2−ατk,tu′(t)|t=τk=P∗kv(τk)+Qjk∗, k=1,2,…,q,I2−β0,tv(t)|t=0=v1, μ2v(t)|t=T+ν2Iβ−10,tv(t)|t=T=v2, | (3.1) |
where α,β∈(1,2], ϕ1,ϕ2:I×R×R→R are continuous functions, Pj,Qj, j=1,2,…,p, and P∗j,Q∗j, j=1,2,…,q, are real numbers, RLDαti,t and RLDβτk,t are the α-order Riemann-Liouville fractional derivatives with lower limits at ti and τk, respectively, μk,νk,uk,vk, i=1,2, are given constants.
Remark 3. Note problem (3.1) differs from problem (2.1):
− The lower limits of the RL fractional derivatives RLDαtj,t and RLDβτk,t in (3.1) are changed at any time of impulse tj and τk, respectively.
− The impulsive conditions are changed in (3.1). This is because the values of the unknown functions after the impulse, u(tj+0) and v(τk+0), respectively, are considered as initial values at that point. But the RL fractional derivative has a singularity at its lower limit. It requires the chang of the impulsive conditions for the unknown functions.
Consider the following boundary value problem for the scalar linear impulsive fractional equation with Riemann-Liouville derivatives of the form
RLDαti,tu(t)=f(t), for t∈(ti,ti+1], i=0,1,2,…,p, α∈(1,2)I2−αtj,tu′(t)|t=tj=Pju(tj)+Qj, j=1,2,…,p,I2−α0,tu(t)|t=0=u1, μ1u(t)|t=T+ν1Iα−10,tu(t)|t=T=u2, | (3.2) |
where the function f: [0,T]→R is a continuous function, Pj,Qj, j=1,2,…,p are real numbers, u1,u2∈R.
Now we will provide an integral presentation of the solution of (3.2).
Lemma 3. The solution of (3.2) satisfies the integral equation
u(t)={c0tα−1+u1Γ(α−1)tα−2+Iα0,tf(t), t∈(0,t1](c0+u1t1Γ(α−1))(t−tm)α−1m∏k=1Pk(tk−tk−1)α−1(α−1)Γ(α−1) +Iαtm,tf(t)+(t−tm)α−1m∑k=1PkIαtk−1,tf(t)|t=tk+Qk(α−1)Γ(α−1)m∏j=k+1Pj(tj−tj−1)α−1(α−1)Γ(α−1), for t∈(tm,tm+1],m=1,2,…,p, |
where
c0=u2μ1(T−tp)1−αM−u1t1Γ(α−1)−(T−tp)1−αMIαtp,tf(t)|t=T−Mp∑k=1PkIαtk−1,tf(t)|t=tk+Qk(α−1)Γ(α−1)m∏j=k+1Pj(tj−tj−1)α−1(α−1)Γ(α−1) −ν1μ1(T−tp)1−αM1Γ(α−1)∫T0(T−s)α−2u(s)ds,M=(α−1)pΓp(α−1)p∏k=1Pk(tk−tk−1)1−α. | (3.3) |
Proof. We will use induction.
For t∈(0,t1] similar to Lemma 2 we get
u(t)=c0tα−1+c1tα−2+1Γ(α)∫t0(t−s)α−1f(s)ds | (3.4) |
where c1=u1Γ(α−1).
For t∈(t1,t2] by Lemma 1 with a=t1 we get
u(t)=b0(t−t1)α−1+b1(t−t1)α−2+1Γ(α)∫tt1(t−s)α−1f(s)ds | (3.5) |
and
u′(t)=b0(α−1)(t−t1)α−2+b1(α−2)(t−t1)α−3+1Γ(α−1)∫tt1(t−s)α−2f(s)ds. | (3.6) |
From the impulsive condition I2−αt1,tu′(t)|t=t1=P1u(t1)+Q1, equalities (2.2) and Iαt1,tf(t))|t=t1=0 we obtain
I2−αt1,tu′(t)=b0(α−1)I2−αt1,t(t−t1)α−2+b1(α−2)I2−αt1,t(t−t1)α−3+I2−αt1,tIα−1t1,tf(t)=b0(α−1)Γ(α−1)+b1(α−2)I2−αt1,t(t−t1)α−3+I1t1,tf(t) | (3.7) |
and
I2−αt1,tu′(t)|t=t1=b0(α−1)Γ(α−1)+b1(α−2)I2−αt1,t(t−t1)α−3|t=t1=P1u(t1)+Q1<∞. | (3.8) |
Since the integral I2−αt1,t(t−t1)α−3 does not converge, it follows that b1=0 and
b0=P1(α−1)Γ(α−1)(c0(t1−t0)α−1+u1t1Γ(α−1)(t1−t0)α−1+Iαt0,tf(t)|t=t1)+Q1(α−1)Γ(α−1). |
Thus,
u(t)=(c0+u1t1Γ(α−1))P1(t1−t0)α−1(α−1)Γ(α−1)(t−t1)α−1 +P1(α−1)Γ(α−1)Iαt0,tf(t)|t=t1(t−t1)α−1 +Q1(α−1)Γ(α−1)(t−t1)α−1+Iαt1,tf(t). | (3.9) |
Similarly, for t∈(tj,tj+1], j=1,2,…,p, by Lemma 1 with a=tj we get
u(t)=k0(t−tj)α−1+k1(t−tj)α−2+1Γ(α)∫ttj(t−s)α−1ϕ1(s,Iα0,su(s),Iβ0,sv(s))ds | (3.10) |
and
u′(t)=k0(α−1)(t−tj)α−2+k1(α−2)(t−tj)α−3+1Γ(α−1)∫ttj(t−s)α−2ϕ1(s,Iα0,su(s),Iβ0,sv(s))ds. | (3.11) |
From the impulsive conditions we obtain k1=0 and
k0=Pj(α−1)Γ(α−1)u(tj)+Qj(α−1)Γ(α−1). |
From the boundary condition μ1u(t)|t=T+ν1Iα−1u(t)|t=T=u2 we get
Iα−1u(t)|t=T=1Γ(α−1)∫T0(T−s)α−2u(s)ds |
and
μ1u(t)|t=T+ν1Iα−1u(t)|t=T=μ1(c0+u1t1Γ(α−1))(T−tp)α−1p∏k=1Pk(tk−tk−1)α−1(α−1)Γ(α−1) +μ1Iαtp,tf(t)|t=T+μ1(T−tp)α−1p∑k=1PkIαtk−1,tf(t)|t=tk+Qk(α−1)Γ(α−1)m∏j=k+1Pj(tj−tj−1)α−1(α−1)Γ(α−1) +ν11Γ(α−1)∫T0(T−s)α−2u(s)ds=u2 | (3.12) |
and we obtain (3.3).
Example 2. Consider the equation
RLD1.50,tu(t)=t, t∈(0,1], RLD1.51,tu(t)=t, t∈(1,2],I0.51,tu′(t)|t=1=1,I0.50,tu(t)|t=0=0, u(t)|t=2+I0.50,tu(t)|t=2=1. | (3.13) |
Note the Eq (3.13) is similar to (2.19) but the lower limit of the fractional derivative is changed at the point of the impulse. The solution of (3.13) satisfies the integral equation
u(t)={0.266667t2.5Γ(1.5),t∈(0,1],1Γ(1.5)(−0.4(t−1)2.5+2(t−1)1.5t3)+(t−1)0.5(0.5)Γ(0.5),t∈(1,2]. |
It is clear the change of the lower limits of the fractional derivatives has a huge influence on the solution of the equation.
Based on the integral presentation of the linear problem (3.2) and Lemma 3, we obtain the following result:
Theorem 2. The solution of (3.1) satisfies the integral equations
u(t)={c0tα−1+u1Γ(α−1)tα−2+Iα0,tϕ1(t,Iα0,tu(t),Iβ0,tv(t)), t∈(0,t1](c0+u1t1Γ(α−1))(t−tm)α−1∏mk=1Pk(tk−tk−1)α−1(α−1)Γ(α−1) +Iαtm,tϕ1(t,Iα0,tu(t),Iβ0,tv(t)) +(t−tm)α−1∑mk=1Pk Iαtk−1,tϕ1(t,Iα0,tu(t),Iβ0,tv(t))|t=tk+Qk(α−1)Γ(α−1)∏mj=k+1Pj(tj−tj−1)α−1(α−1)Γ(α−1), for t∈(tm,tm+1],m=1,2,…,p, |
and
v(t)={b0tα−1+v1Γ(α−1)tα−2+Iα0,tϕ2(t,Iα0,tu(t),Iβ0,tv(t)), t∈(0,τ1](b0+v1τ1Γ(α−1))(t−τm)α−1∏mk=1P∗k(τk−τk−1)α−1(α−1)Γ(α−1) +Iατm,tϕ2(t,Iα0,tu(t),Iβ0,tv(t)) +(t−τm)α−1∑mk=1P∗k Iατk−1,tϕ2(t,Iα0,tu(t),Iβ0,tv(t))|t=τk+Q∗k(α−1)Γ(α−1)∏mj=k+1P∗j(τj−τj−1)α−1(α−1)Γ(α−1), for t∈(τm,τm+1],m=1,2,…,q, |
where
c0=u2μ1(T−tp)1−αM−u1t1Γ(α−1) −(T−tp)1−αMIαtq,tϕ1(t,Iα0,tu(t),Iβ0,tv(t))|t=T −Mp∑k=1PkIαtk−1,tϕ1(t,Iα0,tu(t),Iβ0,tv(t))|t=tk+Qk(α−1)Γ(α−1)p∏j=k+1Pj(tj−tj−1)α−1(α−1)Γ(α−1) −ν1μ1(T−tp)1−αM1Γ(α−1)∫T0(T−s)α−2u(s)ds,M=(α−1)pΓp(α−1)p∏k=1Pk(tk−tk−1)1−α, |
b0=v2μ2(T−τq)1−αC−v1τ1Γ(α−1) −(T−τq)1−αC Iατq,tϕ2(t,Iα0,tu(t),Iβ0,tv(t))t=T −Cq∑k=1P∗k Iατk−1,tϕ2(t,Iα0,tu(t),Iβ0,tv(t))|t=τk+Q∗k(α−1)Γ(α−1)q∏j=k+1P∗j(τj−τj−1)α−1(α−1)Γ(α−1) −ν2μ2(T−tq)1−αC1Γ(α−1)∫T0(T−s)α−2u(s)ds,C=(α−1)qΓq(α−1)q∏k=1P∗k(τk−τk−1)1−α. |
The proof is similar to the one of Lemma 3 applied twice to each of the both components u and v of the coupled system (3.1) for impulsive points ti, i=1,2,…,p and τi, i=1,2,…,p and the functions f(t)=ϕ1(t,Iα0,tu(t),Iβ0,tv(t)) and f(t)=ϕ2(t,Iα0,tu(t),Iβ0,tv(t)) respectively.
In this paper we set up and study a scalar nonlinear integro-differential equation with Riemann-Liouville fractional derivative and impulses. We consider a boundary value problem for the studied equation with Riemann-Liouville fractional derivative of order in (1,2). Note for Riemann-Liouville fractional differential equations both the initial condition and impulsive conditions have to be appropriately given (which is different in the case of ordinary derivatives as well as the case of Caputo fractional derivatives). We consider both interpretations in the literature on the presence of impulses in fractional differential equations: With fixed lower limit of the fractional derivative at the initial time point and with lower limits changeable at each impulsive time point. In both cases we set up in an appropriate way impulsive conditions which are dependent on the Riemann-Liouville fractional derivative. We obtain integral presentations of the solutions in both cases. These presentations could be successfully used for furure studies of existence, stability and other qualitative properties of the solutions of the integro-differential equations with Riemann-Liouville fractional derivative of order in (1,2) and impulses.
S. H. is partially supported by the Bulgarian National Science Fund under Project KP-06-N32/7.
The authors declare that they have no competing interests.
[1] | R. Bhatia, Positive Definite Matrices, New Jersey: Princeton University Press, 2007. |
[2] |
J. Lawson, Y. Lim, The geometric mean, matrices, metrics, and more, J. Amer. Math. Soc., 108 (2001), 797–812. https://doi.org/10.2307/2695553 doi: 10.2307/2695553
![]() |
[3] |
F. Hiai, Matrix analysis: matrix monotone functions, matrix means, and majorization, Interdiscip. Inf. Sci., 16 (2010), 139–248. https://doi.org/10.4036/iis.2010.139 doi: 10.4036/iis.2010.139
![]() |
[4] |
F. Kubo, T. Ando, Means of positive linear operators, Math. Ann., 246 (1980), 205–224. https://doi.org/10.1007/BF01371042 doi: 10.1007/BF01371042
![]() |
[5] |
T. Ando, Concavity of certain maps on positive definite matrices and applications to Hadamard products, Linear Alg. Appl., 26 (1979), 203–241. https://doi.org/10.1016/0024-3795(79)90179-4 doi: 10.1016/0024-3795(79)90179-4
![]() |
[6] |
A. Ploymukda, P. Chansangiam, Concavity and convexity of several maps involving Tracy-Singh products, Khatri-Rao products, and operator-monotone functions of positive operators, Sci. Asia, 45 (2019), 194–201. https://doi.org/10.2306/scienceasia1513-1874.2019.45.194 doi: 10.2306/scienceasia1513-1874.2019.45.194
![]() |
[7] |
P. Chansangiam, Cancellability and regularity of operator connections with applications to nonlinear operator equations involving means, J. Ineq. Appl., 2015 (2015), 411. https://doi.org/10.1186/s13660-015-0934-7 doi: 10.1186/s13660-015-0934-7
![]() |
[8] |
Y. Lim, Factorizations and geometric means of positive definite matrices, Linear Algebra Appl., 437 (2012), 2159–2172. https://doi.org/10.1016/j.laa.2012.05.039 doi: 10.1016/j.laa.2012.05.039
![]() |
[9] |
A. Ploymukda, P. Chansangiam, Geometric means and Tracy-Singh products for positive operators, Commun. Math. Appl., 9 (2018), 475–488. https://doi.org/10.26713/cma.v9i4.547 doi: 10.26713/cma.v9i4.547
![]() |
[10] | A. Ploymukda, P. Chansangiam, Weighted Lim's geometric mean of positive invertible operators on a Hilbert space, J. Comput. Anal. Appl., 29 (2020), 390–400. |
[11] |
Y. Lim, Geometric means on symmetric cones, Arch. der Math., 75 (2000), 39–45. https://doi.org/10.1007/s000130050471 doi: 10.1007/s000130050471
![]() |
[12] |
J. Lawson, Y. Lim, Symmetric sets with midpoints and algebraically equivalent theories, Result. Math., 46 (2004), 37–56. https://doi.org/10.1007/BF03322869 doi: 10.1007/BF03322869
![]() |
[13] | J. Lawson, Y. Lim, Geometric means and reflection quasigroups, Quasigroups Relat. Syst., 14 (2006), 43–59. |
[14] |
J. Lawson, Y. Lim, Symmetric space with convex metrics, Forum Math., 19 (2007), 571–602. https://doi.org/10.1515/FORUM.2007.023 doi: 10.1515/FORUM.2007.023
![]() |
[15] |
J. Lawson, Y. Lim, Solving symmetric matrix word equations via symmetric space machinery, Linear Algebra Appl., 414 (2006), 560–569. https://doi.org/10.1016/j.laa.2005.10.035 doi: 10.1016/j.laa.2005.10.035
![]() |
[16] | P. Chansangiam, Weighted means and weighted mean equations in lineated symmetric spaces, Quasigroups Relat. Syst., 26 (2018), 197–210. |
[17] |
D. Cheng, Semi-tensor product of matrices and its application to Morgen's problem, Sci. China Ser. F, 44 (2001), 195–212. https://doi.org/10.1007/BF02714570 doi: 10.1007/BF02714570
![]() |
[18] |
D. Cheng, H. Qi, A. Xue, A survey on semi-tensor product of matrices, J. Syst. Sci. Complexity, 20 (2007), 304–322. https://doi.org/10.1007/s11424-007-9027-0 doi: 10.1007/s11424-007-9027-0
![]() |
[19] | D. Cheng, H. Qi, Q. Li, Analysis and control of boolean networks: a semi-tensor product approach, London: Springer-Verlag, 2011. |
[20] |
G. Zhao, H. Li, P. Duan, F. E. Alsaadi, Survey on applications of semi-tensor product method in networked evolutionary games, J. Appl. Anal. Comput., 10 (2020), 32–54. https://doi.org/ 10.11948/20180201 doi: 10.11948/20180201
![]() |
[21] |
Y. Yan, D. Cheng, J. Feng, H. Li, J. Yue, Survey on applications of algebraic state space theory of logical systems to finite state machines, Sci. China Inf. Sci., 66 (2023), 111201. https://doi.org/10.1007/s11432-022-3538-4 doi: 10.1007/s11432-022-3538-4
![]() |
[22] |
Y. Yan, J. Yue, Z. Chen, Algebraic method of simplifying Boolean networks using semi‐tensor product of Matrices, Asian J. Control, 21 (2019), 2569–2577. https://doi.org/10.1002/asjc.2125 doi: 10.1002/asjc.2125
![]() |
[23] |
H. Ji, Y. Li, X. Ding, J. Lu, Stability analysis of Boolean networks with Markov jump disturbances and their application in apoptosis networks, Electronic Res. Arch., 30 (2022), 3422–3434. https://doi.org/10.3934/era.2022174 doi: 10.3934/era.2022174
![]() |
[24] |
D. Cheng, Y. Dong, Semi-tensor product of matrices and its some applications to physics, Meth. Appl. Anal., 10 (2003), 565–588. https://dx.doi.org/10.4310/MAA.2003.v10.n4.a5 doi: 10.4310/MAA.2003.v10.n4.a5
![]() |
[25] |
H. Li, G. Zhao, M. Meng, J. Feng, A survey on applications of semi-tensor product method in engineering, Sci. China Inf. Sci., 61 (2018), 010202. https://doi.org/10.1007/s11432-017-9238-1 doi: 10.1007/s11432-017-9238-1
![]() |
[26] |
Z. Ji, J. Li, X. Zhou, F. Duan, T. Li, On solutions of matrix equation AXB=C under semi-tensor product, Linear Multilinear Algebra, 69 (2019), 1–29. https://doi.org/10.1080/03081087.2019.1650881 doi: 10.1080/03081087.2019.1650881
![]() |
[27] |
P. Chansangiam, S. V. Sabau, Sylvester matrix equation under the semi-tensor product of matrices, An. Ştiinţ. Univ. Al. I. Cuza Iaşi. Mat. (N.S.), 68 (2022), 263–278. https://doi.org/10.47743/anstim.2022.00020 doi: 10.47743/anstim.2022.00020
![]() |
[28] |
J. Jaiprasert, P. Chansangiam, Solving the Sylvester-transpose matrix equation under the semi-tensor product, Symmetry, 14 (2022), 1094. https://doi.org/10.3390/sym14061094 doi: 10.3390/sym14061094
![]() |
[29] |
J. Wang, J. E. Feng, H. L. Huang, Solvability of the matrix equation AX2=B with semi-tensor product, Electronic Res. Arch., 29 (2021), 2249–2267. https://doi.org/10.3934/era.2020114 doi: 10.3934/era.2020114
![]() |
[30] |
P. Chansangiam, A. Ploymukda, Riccati equation and metric geometric means of positive semidefinite matrices involving semi-tensor products, AIMS Math., 8 (2023), 23519–23533. https://doi.org/10.3934/math.20231195 doi: 10.3934/math.20231195
![]() |
1. | Vitor Miguel Ribeiro, Pioneering paradigms: unraveling niche opportunities in green finance through bibliometric analysis of nation brands and brand culture, 2024, 6, 2643-1092, 287, 10.3934/GF.2024012 | |
2. | Manuel Mojica, Pedro R. Palos-Sanchez, Edgar Cabanas, Is there innovation management of emotions or just the commodification of happiness? A sentiment analysis of happiness apps, 2024, 1460-1060, 10.1108/EJIM-11-2023-0963 | |
3. | Pedro R. Palos-Sanchez, Raúl J. Chang-Tam, José A. Folgado-Fernández, The Role of Neobanks and FinTech in Sustainable Finance and Technology. The Customer/User Perspective for Entrepreneurs, 2025, 27730328, 100109, 10.1016/j.stae.2025.100109 |