Processing math: 100%
Research article Special Issues

Parameter estimation and fractional derivatives of dengue transmission model

  • Received: 05 November 2019 Accepted: 12 March 2020 Published: 17 March 2020
  • MSC : 34A08, 92B05

  • In this paper, we propose a parameter estimation of dengue fever transmission model using a particle swarm optimization method. This method is applied to estimate the parameters of the host-vector and SIR type dengue transmission models by using cumulative data of dengue patient in East Java province, Indonesia. Based on the parameter values, the basic reproduction number of both models are greater than one and obtained their value for SIR is R0=1.4159 and for vector host is R0=1.1474. We then formulate the models in fractional Atangana-Baleanu derivative that possess the property of nonlocal and nonsingular kernel that has been remained effective to many real-life problems. A numerical procedure for the solution of the model SIR model is shown. Some specific numerical values are considered to obtain the graphical results for both the SIR and Vector Host model. We show that the model vector host provide good results for data fitting than that of the SIR model.

    Citation: Windarto, Muhammad Altaf Khan, Fatmawati. Parameter estimation and fractional derivatives of dengue transmission model[J]. AIMS Mathematics, 2020, 5(3): 2758-2779. doi: 10.3934/math.2020178

    Related Papers:

    [1] Gonca Durmaz Güngör, Ishak Altun . Fixed point results for almost (ζθρ)-contractions on quasi metric spaces and an application. AIMS Mathematics, 2024, 9(1): 763-774. doi: 10.3934/math.2024039
    [2] Pragati Gautam, Vishnu Narayan Mishra, Rifaqat Ali, Swapnil Verma . Interpolative Chatterjea and cyclic Chatterjea contraction on quasi-partial b-metric space. AIMS Mathematics, 2021, 6(2): 1727-1742. doi: 10.3934/math.2021103
    [3] Fatima M. Azmi . New fixed point results in double controlled metric type spaces with applications. AIMS Mathematics, 2023, 8(1): 1592-1609. doi: 10.3934/math.2023080
    [4] Afrah A. N. Abdou, Maryam F. S. Alasmari . Fixed point theorems for generalized α-ψ-contractive mappings in extended b-metric spaces with applications. AIMS Mathematics, 2021, 6(6): 5465-5478. doi: 10.3934/math.2021323
    [5] Abdullah Shoaib, Tahair Rasham, Giuseppe Marino, Jung Rye Lee, Choonkil Park . Fixed point results for dominated mappings in rectangular b-metric spaces with applications. AIMS Mathematics, 2020, 5(5): 5221-5229. doi: 10.3934/math.2020335
    [6] Shaoyuan Xu, Yan Han, Suzana Aleksić, Stojan Radenović . Fixed point results for nonlinear contractions of Perov type in abstract metric spaces with applications. AIMS Mathematics, 2022, 7(8): 14895-14921. doi: 10.3934/math.2022817
    [7] Tatjana Došenović, Dušan Rakić, Stojan Radenović, Biljana Carić . Ćirić type nonunique fixed point theorems in the frame of fuzzy metric spaces. AIMS Mathematics, 2023, 8(1): 2154-2167. doi: 10.3934/math.2023111
    [8] Abdellah Taqbibt, M'hamed Elomari, Milica Savatović, Said Melliani, Stojan Radenović . Fixed point results for a new α-θ-Geraghty type contraction mapping in metric-like space via CG-simulation functions. AIMS Mathematics, 2023, 8(12): 30313-30334. doi: 10.3934/math.20231548
    [9] Hasanen A. Hammad, Hüseyin Işık, Hassen Aydi, Manuel De la Sen . Fixed point approach to the Mittag-Leffler kernel-related fractional differential equations. AIMS Mathematics, 2023, 8(4): 8633-8649. doi: 10.3934/math.2023433
    [10] Yan Han, Shaoyuan Xu, Jin Chen, Huijuan Yang . Fixed point theorems for b-generalized contractive mappings with weak continuity conditions. AIMS Mathematics, 2024, 9(6): 15024-15039. doi: 10.3934/math.2024728
  • In this paper, we propose a parameter estimation of dengue fever transmission model using a particle swarm optimization method. This method is applied to estimate the parameters of the host-vector and SIR type dengue transmission models by using cumulative data of dengue patient in East Java province, Indonesia. Based on the parameter values, the basic reproduction number of both models are greater than one and obtained their value for SIR is R0=1.4159 and for vector host is R0=1.1474. We then formulate the models in fractional Atangana-Baleanu derivative that possess the property of nonlocal and nonsingular kernel that has been remained effective to many real-life problems. A numerical procedure for the solution of the model SIR model is shown. Some specific numerical values are considered to obtain the graphical results for both the SIR and Vector Host model. We show that the model vector host provide good results for data fitting than that of the SIR model.


    In 1922, Banach first presented the Banach contraction principle [1] in metric spaces, which is a powerful and classical means to solve problems about fixed point. Subsequently, it has been generalized in many aspects. One vital generalization is to promote the concept of metric spaces. b-metric spaces is regarded as a well-known generalization of metric spaces. In 1993, Czerwik [2] first introduced the concept of b-metric spaces by modifying the third condition of the metric function. The author also provided fixed point results for contraction conditions in this type space. In the sequel, several papers have been published on the fixed point theory of various classes of single-valued and multi-valued operators in b-metric spaces (see [3,4,5,6]).

    In 1969, Boyd and Wong [7] gave a definition of ϕ-contraction in metric spaces for the first time. Afterward, Alber and Guerre [8] defined the concept of weak contraction and got some fixed point results in Hilbert space. In [9], Rhoades generalized Alber and Guerre's results to more general forms. Alutn [10] proved the common fixed point theorem for weakly contraction mappings of integral type. Later, more scholars [11,12,13,14] presented some fixed point theorems for weakly contractive mappings in different spaces.

    In particular, Perveen [15] obtained the θ-weak contraction principle in metric spaces as follows:

    Theorem 1.1. [15] Suppose (Ω,) is a complete metric space and S:ΩΩ is a θ-weak contraction. If θ is continuous, then

    (a) S has unique fixed point (say, zΩ),

    (b)limn+Snz=z, zΩ.

    Motivated and inspired by results in [15], in this paper we give some fixed point theorems for contractive mappings of the integral type in b-metric spaces. Furthermore, two examples are given to prove the feasibility of the theorems. Also, the solvability of a functional equation arising in dynamic programming is considered by means of obtained results.

    We introduce the following definitions and lemmas, which will be used to obtain our main results.

    Definition 2.1. [2] Let be a nonempty set and s1 be a given real number. A mapping ϖ:×[0,+) is said to be a b-metric if, and only if, for all κ,λ,μ, the following conditions are satisfied:

    (ⅰ) ϖ(κ,λ)=0 if, and only if, κ=λ;

    (ⅱ) ϖ(κ,λ)=ϖ(λ,κ);

    (ⅲ) ϖ(κ,λ)s(ϖ(κ,μ)+ϖ(λ,μ)).

    In general, (,ϖ) is called a b-metric space with parameter s1.

    Remark 2.2. Visibly, every metric space is a b-metric space with s=1. There are several examples of b-metric spaces that are not metric spaces (see [16]).

    Example 2.3. [17] Let (,d) be a metric space, and ϖ(κ,λ)=(d(κ,λ))p, where p>1 is a real number, then ϖ(κ,λ) is a b-metric with s=2p1.

    Definition 2.4. [18] Let (,ϖ) be a b-metric space with parameter s1, then a sequence {κι}+ι=1 in is said to be:

    (ⅰ) b-convergent if there exists κ such that ϖ(κι,κ)0 as ι+;

    (ⅱ) a Cauchy sequence if ϖ(κι,κυ)0 when ι,υ+.

    As usual, a b-metric space is called complete if, and only if, each Cauchy sequence in this space is b-convergent.

    The following lemma plays a key role in our conclusion.

    Lemma 2.5. [17] Let (,ϖ) be a b-metric space with parameter s1. Assume that {κι}+ι=1 and {λι}+ι=1 are b-convergent to κ and λ, respectively, then we have

    1s2ϖ(κ,λ)lim infι+ϖ(κι,λι)lim supι+ϖ(κι,λι)s2ϖ(κ,λ).

    In particular, if κ=λ, then we have limι+ϖ(κι,λι)=0. Moreover, for each μ, we have

    1sϖ(κ,μ)lim infι+ϖ(κι,μ)lim supι+ϖ(κι,μ)sϖ(κ,μ).

    Lemma 2.6. [19] Let φ and {κι}ιN be a nonnegative sequence with limn+κι=κ, then

    limn+κι0φ(ω)dω=κ0φ(ω)dω.

    Lemma 2.7. [19] Let φ and {κι}ιN be a nonnegative sequence, then

    limι+κι0φ(ω)dω=0

    if, and only if, limι+κι=0.

    Throughout this paper, we assume that R+=[0,+), N0=N{0}, where N stands for the set of positive integers,

    ={ξ|ξ:R+R+ satisfies that ξ is Lebesgue integrable, and δ0ξ(ω)dω>0 for each δ>0}.

    Let (,ϖ) be a b-metric space with parameter s1 and S be a self-mapping on . For any u,v, set

    (u,v)=max{ϖ(u,v),ϖ(u,Su),ϖ(v,Sv),ϖ(u,Sv)+ϖ(v,Su)2s}.

    In this part, we introduce the new concept of αsp-admissible mapping and other definitions, which will be used to prove the fixed point theorems of the integral type in b-metric space. Moreover, we also provide two examples to support our results.

    Let

    Θ1={θ|θ:(0,+)(1,+) satisfies the following conditions (1) and (3)},

    Θ2={θ|θ:(0,+)(0,1) satisfies the following conditions (2) and (3)},

    where

    (1) θ is nondecreasing and continuous;

    (2) θ is nonincreasing and continuous;

    (3) for each sequence {βι}+ι=1(0,+), limι+θ(βι)=1limι+βι=0.

    Definition 3.1. Let (,ϖ) be a b-metric space with parameter s1 and p1 be an integer. A mapping S: is said to be αspadmissible if for all z,w, one has

    α(z,w)spα(Sz,Sw)sp

    where α:×[0,+) is a given function.

    Lemma 3.2. Let φ and {κι}ιN be a nonnegative sequence. If lim supι+κι=κ, then

    κ0φ(ω)dωlim supι+κι0φ(ω)dω.

    If lim infι+κι=κ, then

    lim infι+κι0φ(ω)dωκ0φ(ω)dω.

    Proof. It follows from lim supι+κι=κ that there exists a subsequence {κις} of {κι} such that

    limς+κις=κ.

    In view of Lemma 2.6, we deduce that

    κ0φ(ω)dω=limς+κις0φ(ω)dωlim supι+κι0φ(ω)dω.

    Similarly, one can prove another inequality.

    Theorem 3.3. Let (,ϖ) be a complete b-metric space with parameter s1 and S: be a given self-mapping. Assume that α:×[0,+) and p3. If

    (ⅰ) S is αsp-admissible,

    (ⅱ) there is p0 satisfying α(p0,Sp0)sp,

    (ⅲ) α satisfies transitive property, i.e., for ξ,η,ζ if

    α(ξ,η)sp and α(η,ζ)spα(ξ,ζ)sp,

    (ⅳ) if {pι} is a sequence in satisfying pιp as ι+, then there exists a subsequence {pι(k)}+k=1 of {pι}+ι=1 with α(pι(k),p)sp,

    (ⅴ) S is a θ-weak contraction, that is, there exists (0,1), φ, θΘ1 such that: for any u,v,

    α(u,v)sp,ϖ(Su,Sv)0φ(ω)dω>0θ(α(u,v)ϖ(Su,Sv)0φ(ω)dω)[θ( (u,v)0φ(ω)dω)], (3.1)

    then S has a fixed point in . Furthermore, if

    (ⅵ) for p,qFix(S), one can get the conditions of α(p,q)sp and α(q,p)sp, where Fix(S) represents the collection of all fixed points of S,

    then the fixed point is unique.

    Proof. Under condition (ⅱ), there is a p0 satisfying α(p0,Sp0)sp. Define sequence {pn} in by pn+1=Spn for nN. If pn0=Spn0 for some n0, then pn0 is a fixed point of S. Suppose that pn+1pn for nN. It follows from condition (ⅰ) that

    α(p0,Sp0)spα(Sp0,S2p0)sp,
    α(p1,p2)spα(Sp1,Sp2)sp,
    α(p2,p3)spα(Sp2,Sp3)sp,

    Thus, for all nN, we have α(pn1,pn)sp. Using (3.1) by u=pn1 and v=pn, one gets

    θ(α(pn1,pn)ϖ(Spn1,Spn)0φ(ω)dω)[θ( (pn1,pn)0φ(ω)dω)] (3.2)

    where

    (pn1,pn)=max{ϖ(pn1,pn),ϖ(pn1,Spn1),ϖ(pn,Spn),ϖ(pn1,Spn)+ϖ(pn,Spn1)2s}=max{ϖ(pn1,pn),ϖ(pn1,pn),ϖ(pn,pn+1),ϖ(pn1,pn+1)+ϖ(pn,pn)2s}=max{ϖ(pn1,pn),ϖ(pn,pn+1)}. (3.3)

    If ϖ(pn,pn+1)ϖ(pn1,pn) for some nN, in view of (3.2) and (3.3), we have (pn1,pn)=ϖ(pn,pn+1), so

    θ(ϖ(pn,pn+1)0φ(ω)dω)<θ(spϖ(pn,pn+1)0φ(ω)dω)θ(α(pn1,pn)ϖ(Spn1,Spn)0φ(ω)dω)[θ((pn1,pn)0φ(ω)dω)]=[θ(ϖ(pn,pn+1)0φ(ω)dω)]

    which is impossible. Hence,

    ϖ(pn1,pn)>ϖ(pn,pn+1). (3.4)

    (3.4) implies that (pn1,pn)=ϖ(pn1,pn) is decreasing. Thus, we have

    θ(ϖ(pn,pn+1)0φ(ω)dω)<[θ(ϖ(pn1,pn)0φ(ω)dω)]<[θ(ϖ(pn2,pn1)0φ(ω)dω)]2<<[θ(ϖ(p0,p1)0φ(ω)dω)]n.

    Letting n+ in the above inequality, we get

    1limn+θ(ϖ(pn,pn+1)0φ(ω)dω)limn+[θ(ϖ(p0,p1)0φ(ω)dω)]n=1

    i.e., limn+θ(ϖ(pn,pn+1)0φ(ω)dω)=1, which by the definition of θ yields that

    limn+ϖ(pn,pn+1)0φ(ω)dω=0

    which implies

    limn+ϖ(pn,pn+1)=0.

    Now, we prove {pn} is a Cauchy sequence. Suppose {pn} is not Cauchy, then there exists ε>0 for which we can choose sequences {pn(k)} and {pm(k)} of {pn}, such that n(k) is the smallest index for which n(k)>m(k)>k,

    εϖ(pm(k),pn(k)),ϖ(pm(k),pn(k)1)<ε. (3.5)

    Under the triangle inequality and (3.5), we get

    εϖ(pm(k),pn(k))sϖ(pm(k),pn(k)1)+sϖ(pn(k)1,pn(k))<sε+sϖ(pn(k)1,pn(k)).

    Taking the superior limit and inferior limit as k+, we get

    εlim infk+ϖ(pm(k),pn(k))lim supk+ϖ(pm(k),pn(k))sε. (3.6)

    Similarly, one can deduce the following inequalities:

    ϖ(pm(k),pn(k))sϖ(pm(k),pm(k)1)+s2ϖ(pm(k)1,pn(k)1)+s2ϖ(pn(k)1,pn(k)), (3.7)
    ϖ(pm(k)1,pn(k)1)sϖ(pm(k)1,pm(k))+s2ϖ(pm(k),pn(k))+s2ϖ(pn(k),pn(k)1), (3.8)
    ϖ(pm(k),pn(k))sϖ(pm(k),pm(k)1)+sϖ(pm(k)1,pn(k)), (3.9)
    ϖ(pm(k)1,pn(k))sϖ(pm(k)1,pm(k))+sϖ(pm(k),pn(k)), (3.10)
    ϖ(pm(k),pn(k))sϖ(pm(k),pn(k)1)+sϖ(pn(k)1,pn(k)), (3.11)
    ϖ(pm(k),pn(k)1)sϖ(pm(k),pn(k))+sϖ(pn(k),pn(k)1). (3.12)

    By (3.6)–(3.8), we have

    εs2lim infk+ϖ(pm(k)1,pn(k)1)lim supk+ϖ(pm(k)1,pn(k)1)s3ε. (3.13)

    It follows from (3.6), (3.9), and (3.10) that

    εslim infk+ϖ(pm(k)1,pn(k))lim supk+ϖ(pm(k)1,pn(k))s2ε. (3.14)

    According to (3.6), (3.11), and (3.12), one can obtain

    εslim infk+ϖ(pm(k),pn(k)1)lim supk+ϖ(pm(k),pn(k)1)s2ε. (3.15)

    Thus, there exists NN0 such that for m(k),n(k)N, ϖ(pm(k)1,pn(k)1)0φ(ω)dω>0.

    In view of the definition of (u,v), we have

    (pm(k)1,pn(k)1)=max{ϖ(pm(k)1,pn(k)1),ϖ(pm(k)1,Spm(k)1),ϖ(pn(k)1,Spn(k)1),ϖ(pm(k)1,Spn(k)1)+ϖ(pn(k)1,Spm(k)1)2s}=max{ϖ(pm(k)1,pn(k)1),ϖ(pm(k)1,pm(k)),ϖ(pn(k)1,pn(k)),ϖ(pm(k)1,pn(k))+ϖ(pn(k)1,pm(k))2s}. (3.16)

    Letting k+ in (3.16), we get

    lim infk+(pm(k)1,pn(k)1)lim supk+(pm(k)1,pn(k)1)max{s3ε,0,0,s2ε+s2ε2s}=s3ε. (3.17)

    The transitivity property of α yields that α(pm(k)1,pn(k)1)sp. Choosing u=pm(k)1 and v=pn(k)1 in (3.1), by Lemma 3.2, one can deduce

    θ(s3ε0φ(ω)dω)lim infk+θ(spϖ(pm(k),pn(k))0φ(ω)dω)lim infk+θ(α(pm(k)1,pn(k)1)ϖ(Spm(k)1,Spn(k)1)0φ(ω)dω)lim infk+[θ((pm(k)1,pn(k)1)0φ(ω)dω)][θ(s3ε0φ(ω)dω)]

    which is a contradiction. So, {pn} is Cauchy. As is complete, there exists p such that pnp as n+.

    Next, we prove the point p to be a fixed point of S. So, we think about a set, say Q={nN0:pn=Sp}, then it has two situations. One, if Q is an infinite set, then there exists a subsequence {pn(k)}{pn}, which converges to Sp. By the uniqueness of limit, we have Sp=p. The other, if Q is a finite set, then there is NN such that ϖ(pn,Sp)0φ(ω)dω>0 for any nN. By (iv), we obtain that there exists a subsequence {pn(k)}{pn} such that α(pn(k)1,p)sp and ϖ(pn(k),Sp)0φ(ω)dω>0, kN. Taking u=pn(k)1 and v=p in (3.1), we get

    θ(α(pn(k)1,p)ϖ(Spn(k)1,Sp)0φ(t)dt)[θ( (pn(k)1,p)0φ(ω)dω)] (3.18)

    where

    (pn(k)1,p)=max{ϖ(pn(k)1,p),ϖ(pn(k)1,Spn(k)1),ϖ(p,Sp),ϖ(pn(k)1,Sp)+dϖ(p,Spn(k)1)2s}=max{ϖ(pn(k)1,p),ϖ(pn(k)1,pn(k)),ϖ(p,Sp),ϖ(pn(k)1,Sp)+ϖ(p,pn(k))2s}. (3.19)

    Putting the limit as k+ in (3.19), we get

    limk+(pn(k)1,p)=max{0,0,ϖ(p,Sp),ϖ(p,Sp)2}=ϖ(p,Sp).

    According to (3.18), (3.19), and Lemma 2.5, we get

    θ(ϖ(p,Sp)0φ(ω)dω)<θ(s31sϖ(p,Sp)0φ(ω)dω)lim supn+θ(spϖ(Spn(k)1,Sp)0φ(ω)dω)lim supn+θ(α(pn(k)1,p)ϖ(Spn(k)1Sp)0φ(ω)dω)lim supn+[θ((pn(k)1,p)0φ(ω)dω)]=[θ(ϖ(p,Sp)0φ(ω)dω)]

    which is contradiction. Hence, Sp=p.

    For the uniqueness, let q be one more fixed point of S, then (vi) yields α(p,q)sp. Using (3.1), one can arrive at

    θ(α(p,q)ϖ(Sp,Sq)0φ(ω)dω)[θ( (p,q)0φ(ω)dω)]

    where

    (p,q)=max{ϖ(p,q),ϖ(p,Sp),ϖ(q,Sq),ϖ(p,Sq)+ϖ(q,Sp)2s}=max{ϖ(p,q),0,0,ϖ(p,q)+ϖ(q,p)2s,0,0}=ϖ(p,q).

    So, we have

    θ(ϖ(p,q)0φ(ω)dω)<θ(s3ϖ(p,q)0φ(ω)dω)θ(α(p,q)ϖ(Sp,Sq)0φ(ω)dω)[θ((p,q)0φ(ω)dω)]=[θ(ϖ(p,q)0φ(ω)dω)]

    a contradiction. Thus, p=q, which proves the uniqueness of the fixed point. This completes the proof.

    Example 3.4. Let =[0,1] and ϖ(p,q)=(pq)2. It is easy to show that (,ϖ) is a b-metric space with parameter s=2. Define mappings S: by

    Sp={p4+1,  p[0,1),78, p=1

    and α:×[0,+) by

    α(p,q)=23,p,q.

    Define θ:[0,+)(1,+) and φ:[0,+)[0,+) by

    θ(ω)=e256ω+sinω and φ(ω)=2ω.

    It is easy to get that α(u,v)23, ϖ(Su,Sv)0φ(ω)dω>0 u,v[0,1] and uv. We consider the two following cases:

    Case 1. u,v[0,1). It follows that

    θ(α(u,v)ϖ(Su,Sv)0φ(ω)dω)=θ(23(u4+1+v41)202ωdω)=θ(14(uv)4)=e64(uv)4+sin(14(uv)4),
    [θ((u,v)0φ(ω)dω)]12[θ((uv)202ωdω)]12=[θ((uv)4]12=e128(uv)4+sin((uv)4)2.

    Case 2. u[0,1),v=1. One can deduce that

    θ(α(u,v)ϖ(Su,Sv)0φ(ω)dω)=θ(23(u4+178)202ωdω)=θ(14(u12)4)θ(14×16)=e4+sin164,
    [θ((u,v)0φ(ω)dω)]12[θ(ϖ(u,Sv)+ϖ(v,Su)2202ωdω)]12=[θ(14[(u78)2+u216]02ωdω)]12=[θ(141716[(u1417)2+(78)2(1417)2]02ωdω)]12[θ(141716[(78)2(1417)2]02ωdω)]12[θ(116)]12=e8+sin(116)2.

    Clearly, as =12, we have

    θ(α(u,v)ϖ(Su,Sv)0φ(ω)dω)[θ((u,v)0φ(ω)dω)].

    Hence, (3.1) holds. It follows that all conditions of Theorem 3.3 are satisfied with s=2 and p=3. Here, 45 is the fixed point of S.

    Remark 3.5. If (,ϖ) is a metric space and α(u,v)=1 in Theorem 3.3, then one can obtain Theorem 1.1 immediately.

    Theorem 3.6. Let (,ϖ) be a complete b-metric space with parameter s1 and S: be a given self-mapping. Assume that α:×[0,+) and p3. If

    (ⅰ) S is αsp-admissible,

    (ⅱ) there is p0 satisfying α(p0,Sp0)sp,

    (ⅲ) α satisfies transitive property, i.e., for ξ,η,ζ if

    α(ξ,η)sp and α(η,ζ)spα(ξ,ζ)sp,

    (ⅳ) if {pι} is a sequence in satisfying pιp as ι+, then there is a subsequence {pι(k)}+k=1 of {pι}+ι=1 with α(pι(k),p)sp,

    (ⅴ) S is a θ-ψ-weak contraction, that is, there exists φ, θΘ2 such that: for any u,vφ

    α(u,v)sp,ϖ(Su,Sv)0φ(ω)dω>0ψ(α(u,v)ϖ(Su,Sv)0φ(ω)dω)θ(ψ((u,v)0φ(ω)dω))ψ((u,v)0φ(ω)dω), (3.20)

    where ψ:[0,+)[0,+) is a continuous and increasing function with ψ(ω)=0 if, and only if, ω=0,

    then S has a fixed point in . Moreover, if

    (ⅵ) for p,qFix(S), one can get the conditions of α(p,q)sp and α(q,p)sp, where Fix(S) represents the collection of all fixed points of S,

    then the fixed point of S is unique.

    Proof. As in the proof of Theorem 3.3, we infer α(pn1,pn)sp. Using (3.16) with u=pn1 and v=pn, one can deduce that

    ψ(α(pn1,pn)ϖ(Spn1,Spn)0φ(ω)dω)θ(ψ((pn1,pn)0φ(ω)dω))ψ((pn1,pn)0φ(ω)dω) (3.21)

    where

    (pn1,pn)=max{ϖ(pn1,pn),ϖ(pn1,Spn1),ϖ(pn,Spn),ϖ(pn1,Spn)+ϖ(pn,Spn1)2s}=max{ϖ(pn1,pn),ϖ(pn1,pn),ϖ(pn,pn+1),ϖ(pn1,pn+1)+ϖ(pn,pn)2s}=max{ϖ(pn1,pn),ϖ(pn,pn+1)}. (3.22)

    If ϖ(pn,pn+1)ϖ(pn1,pn) for some nN, according to (3.22), one can obtain (pn1,pn)=ϖ(pn,pn+1). It follows that

    ψ(ϖ(pn,pn+1)0φ(ω)dω)ψ(α(pn1,pn)ϖ(Spn1,Spn)0φ(ω)dω)θ(ψ((pn1,pn)0φ(ω)dω))ψ((pn1,pn)0φ(ω)dω)=θ(ψ(ϖ(pn,pn+1)0φ(ω)dω))ψ(ϖ(pn,pn+1)0φ(ω)dω)

    which is a contradiction. Thus,

    ϖ(pn1,pn)>ϖ(pn,pn+1). (3.23)

    By (3.23), we get that (pn1,pn)=ϖ(pn1,pn) is a decreasing sequence. Hence, there exists ρ0 such that ϖ(pn1,pn)=ρ. If ρ>0, then

    ψ(ϖ(pn,pn+1)0φ(ω)dω)ψ(ϖ(pn1,pn)0φ(ω)dω)θ(ψ(ϖ(pn1,pn)0φ(ω)dω)).

    Taking n+, we obtain

    1limn+θ(ψ(ϖ(pn1,pn)0φ(ω)dω))1

    which implies limn+θ(ψ(ϖ(pn1,pn)0φ(ω)dω))=1. In view of the definition of θ and ψ, one can deduce that

    limn+ϖ(pn1,pn)0φ(ω)dω=0

    i.e.,

    limn+ϖ(pn,pn+1)=0,

    which is contradiction. It follows that limn+ϖ(pn,pn+1)=0.

    Next, we want to show {pn} is a Cauchy sequence. As in the proof of Theorem 3.3, we obtain that (3.13)–(3.17) hold. The transitivity property of α implies that α(pm(k)1,pn(k)1)sp. Putting u=pm(k)1 and v=pn(k)1 into (3.20), we get

    ψ(s3ε0φ(ω)dω)lim infk+ψ(spϖ(pm(k),pn(k))0φ(ω)dω)lim infk+ψ(α(pmk1,pnk1)ϖ(Spmk1,Spnk1)0φ(ω)dω)lim infk+[θ(ψ((pmk1,pnk1)0φ(ω)dω))ψ((pmk1,pnk1)0φ(ω)dω)]lim supk+θ(ψ((pmk1,pnk1)0φ(ω)dω)lim infk+ψ((pmk1,pnk1)0φ(ω)dω)=θ(lim infk+ψ((pmk1,pnk1)0φ(ω)dω))ψ(lim infk+(pmk1,pnk1)0φ(ω)dω)<ψ(s3ε0φ(ω)dω)

    which is a contradiction. Hence, {pn} is Cauchy. The completeness of ensures that there exists p such that {pn}p as n+.

    Next, we prove the point p to be a fixed point of S. Similar to the discussion related to Theorem 3.4, taking u=pn(k)1 and v=p in (3.20), we get

    ψ(α(pn(k)1,p)ϖ(Spn(k)1,Sp)0φ(ω)dω)θ(ψ((pn(k)1,p)0φ(ω)dω))ψ((pn(k)1,p)0φ(ω)dω) (3.24)

    where

    (pn(k)1,p)=max{ϖ(pn(k)1,p),ϖ(pn(k)1,Spn(k)1),ϖ(p,Sp),ϖ(pn(k)1,Sp)+ϖ(p,Spn(k)1)2s}=max{ϖ(pn(k)1,p),ϖ(pn(k)1,pn(k)),ϖ(p,Sp),ϖ(pn(k)1,Sp)+ϖ(p,pn(k))2s}. (3.25)

    Taking the limit as n+ in (3.25), we get

    limn+(pn(k)1,p)=max{0,0,ϖ(p,Sp),ϖ(p,Sp)2}=ϖ(p,Sp). (3.26)

    According to (3.24), (3.26), and Lemma 2.5, we get

    ψ(ϖ(p,Sp)0φ(t)dt)ψ(s31sϖ(p,Sp)0φ(t)dt)limn+ψ(α(pn(k)1,p)ϖ(Spn(k)1,Sp)0φ(ω)dω)limn+θ(ψ((pn(k)1,p)0φ(ω)dω))ψ((pn(k)1,p)0φ(ω)dω)=θ(ψ(ϖ(p,Sp)0φ(ω)dω))ψ(ϖ(p,Sp)0φ(ω)dω)<ψ(ϖ(p,Sp)0φ(ω)dω)

    which is impossible. It follows that Sp=p.

    At last, we show the uniqueness of the fixed point of S. Suppose q is another fixed point of S. It follows from the condition (ⅳ) that α(p,q)sp. In light of (3.20), one can get

    ψ(α(p,q)ϖ(Sp,Sq)0φ(ω)dω)θ(ψ((p,q)0φ(ω)dω))ψ((p,q)0φ(ω)dω),
    (p,q)=max{ϖ(p,q),ϖ(p,Sp),ϖ(q,Sq),ϖ(p,Sq)+ϖ(q,Sp)2s}=max{ϖ(p,q),0,0,ϖ(p,q)+ϖ(q,p)2s}=ϖ(p,q).

    Then

    ψ(ϖ(p,q)0φ(ω)dω)ψ(α(p,q)ϖ(Sp,Sq)0φ(ω)dω)θ(ψ((p,q)0φ(ω)dω))ψ((p,q)0φ(ω)dω)<ψ(ϖ(p,q)0φ(ω)dω)

    a contradiction, which implies that p=q. This completes the proof.

    Example 3.7. Let =[0,1] and ϖ(p,q)=(pq)2. Define mappings S: by

    Sp={p3216e,  p[0,12],13216e,p(12,1]

    and α:×[0,+) by

    α(p,q)=24,p,q[0,1].

    Define θ:[0,+)(0,1) and ψ,φ:[0,+)[0,+) by

    θ(ω)=e4ω,ψ(ω)=ω  and  φ(ω)=2ω.

    One can deduce that α(u,v)24, ϖ(Su,Sv)0φ(ω)dω>0 u,v[0,1] with uv. It follows that we also consider two cases:

    Case 1. u,v[0,12], then

    ψ(α(u,v)ϖ(Su,Sv)0φ(ω)dω)=24(u3216ev3216e)202ωdω=1642×4e(uv)4,
    θ(ψ((u,v)0φ(ω)dω))ψ((u,v)0φ(ω)dω)=e4(u,v)02ωdω(u,v)02ωdω14e(uv)4.

    Case 2. u[0,12],v(12,1]. It is easy to obtain that

    ψ(α(u,v)ϖ(Su,Sv)0φ(ω)dω)=24(u3216e13216e)202ωdω=1642×4e(u1)41642×4e,
    θ(ψ((u,v)0φ(ω)dω))ψ((u,v)0φ(ω)dω)=e4(u,v)02ωdω(u,v)02ωdω1e4×116×(113216e)41642×4e.

    That is,

    ψ(α(u,v)ϖ(Su,Sv)0φ(ω)dω)θ(ψ((u,v)0φ(ω)dω))ψ((u,v)0φ(ω)dω).

    It follows that all conditions of Theorem 3.6 are satisfied with s=2 and p=4. It is easy to get that 0 is the unique fixed point of S.

    In this section, by using the fixed point theorems obtained in Section 3, we study the existence of solutions of the following functional Eq (4.2).

    Let O and P be two Banach spaces and SO and DP be the state and decision spaces. B(S) denotes the Banach space of all bounded real-valued functions on S with norm

    m∥=sup{|m(ξ)|:ξS} for any  mB(S). (4.1)

    Bellman [20] was the first to investigate the existence and uniqueness of solutions for the following functional equations arising in dynamic programming:

    f(x)=infyDmax{r(x,y),s(x,y),f(b(x,y))},
    f(x)=infyDmax{r(x,y),f(b(x,y))}

    in a complete metric space BB(S). As suggested in Bellman and Lee [21], the basic form of the functional equations in dynamic programming is as follows:

    f(x)=optyD{H(x,y,f(T(x,y)))},xS

    where the opt represents sup or inf. Bhakta and Mitra [22] obtained the existence and uniqueness of solutions for the functional equations

    f(x)=supyD{p(x,y)+A(x,y,f(a(x,y))}

    in a Banach space B(S) and

    f(x)=supyD{p(x,y)+f(a(x,y))}

    in BB(S), respectively. After that, many authors established the existence and uniqueness of solutions or common solutions for several classes of functional equations or systems of functional equations arising in dynamic programming by means of various fixed and common fixed point theorems (see [23,24,25]).

    It is easy to get that (B(S),ϖ) is a complete b-metric space with

    ϖ(ξ,η)=∥ξη2,ξ,ηB(S).

    Consider the functional equations arising in dynamic programming:

    f(x)=infyD{u(x,y)+H(x,y,f(T(x,y)))},xS (4.2)

    where u:S×DR, T:S×DS and H:S×D×RR are mappings. Let

    Sf(x)=infyD{u(x,y)+H(x,y,f(T(x,y)))},(x,f)S×B(S). (4.3)

    Theorem 4.1. Let u:S×DR, T:S×DS, H:S×D×RR, S:B(S)B(S), α:B(S)×B(S)R. If

    (ⅰ) u and H are bounded,

    (ⅱ) S is αsp-admissible,

    (ⅲ) there is p0B(S) satisfying α(p0,Sp0)sp,

    (ⅳ) α satisfies transitive property, i.e., for ξ,η,ζB(S) such that

    α(ξ,η)sp and α(η,ζ)spα(ξ,ζ)sp,

    (ⅴ) if {pn} is a sequence in B(S) satisfying pnp as n+, then there is a subsequence {pn(k)} of {pn} with α(pn(k),p)sp,

    (ⅵ) for p,qFix(S), one can get the condition of α(p,q)sp and α(q,p)sp, where Fix(S) represents the collection of all fixed points of S,

    (ⅶ) if there exists (0,1), φ such that

    α(u,v)sp,SuSv20φ(ω)dω>0exp(2α(u,v)|H(u,v,g(T(u,v)))H(u,v,h(T(u,v)))|20φ(ω)dω)[exp( (u,v)0φ(ω)dω)], (4.4)

    then the functional Eq (4.2) has a unique solution pB(S).

    Proof. It follows from (i) that there exists M>0 satisfying

    sup{|u(x,y)|,|H(x,y,t)|:(x,y,t)S×D×R}M.

    It is easy to see that S is a self-mapping in B(S). Define α:B(S)×B(S)[0,) by

    α(u,v)={sp,  ϖ(Su,Sv)>0,0,  otherwise.

    By (i) and φ, we have for each ε>0, there exists δ>0 such that

    Cφ(t)dt<ε,C[0,2M] with m(C)δ, (4.5)

    where m(C) denotes the Lebesgue measure of C.

    Let uS,h,gB(S). By (4.3), there exists v,wD satisfying

    Sg(u)>u(u,v)+H(u,v,g(T(u,v)))2δ2,
    Sh(u)>u(u,w)+H(u,w,h(T(u,w)))2δ2,
    Sg(u)u(u,w)+H(u,w,g(T(u,w))),
    Sh(u)u(u,v)+H(u,v,h(T(u,v))).

    Thus,

    Sg(u)Sh(u)<H(u,w,g(T(u,w)))H(u,w,h(T(u,w)))+2δ2|H(u,w,g(T(u,w)))H(u,w,h(T(u,w)))|+2δ2,
    Sh(u)Sg(u)<H(u,v,h(T(u,v)))H(u,v,g(T(u,v)))+2δ2|H(u,v,h(T(u,v)))H(u,v,g(T(u,v)))|+2δ2.

    It follows that

    ||SgSh||=supuS|Sg(u)Sh(u)|max{T1,T2}+2δ2, (4.6)

    where

    T1=|H(u,w,g(T(u,w)))H(u,w,h(T(u,w)))|,
    T2=|H(u,v,h(T(u,v)))H(u,v,g(T(u,v)))|.

    It is easy to get that ||SgSh||2max{2T12,2T22}+δ. Under (4.4) and (4.6), we have

    exp(sp||Sg(u)Sh(u)||20φ(ω)dω)exp(spmax{2T21,2T22}+δ0φ(ω)dω)=max{exp(2spT21+δ0φ(ω)dω),exp(2spT22+δ0φ(ω)dω)}=max{exp(2spT210φ(ω)dω)exp(2spT21+δ2spT21φ(ω)dω),exp(2spT220φ(ω)dω)exp(2spT22+δ2spT22φ(ω)dω)}max{exp(2spT210φ(ω)dω),exp(2spT220φ(ω)dω)}max{exp(2spT21+δ2spT21φ(ω)dω),exp(2spT22+δ2spT22φ(ω)dω)}[exp( (u,v)0φ(ω)dω)]exp(ε).

    Letting ε0+ in the above inequality, we get

    exp(α(u,v)SgSh20φ(ω)dω)[exp( (u,v)0φ(ω)dω)].

    Thus, the conditions of Theorem 3.3 are satisfied by taking θ(ω)=exp(ω), so the functional Eq (4.2) has a unique fixed sloution pB(S). This completes the proof.

    In this manuscript, we first defined two new types of weak contractions named θ-weak contraction and θ-ψ-weak contraction. Second, we presented the conditions of existence and uniqueness of fixed points for them in b-metric spaces. After that, two examples were given to demonstrate the practicability of our theorems. As an application, the existence and uniqueness of solutions for a class of functional equations arising in dynamic programming were discussed.

    The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.

    This work was financially supported by the Science and Research Project Foundation of Liaoning Province Education Department (No: JYTMS20231700).

    The authors declare that they have no conflicts of interest regarding the publication of this paper.



    [1] World Health Organization, Fact sheet on the Dengue and severe dengue, WHO, 2017. Available from: http://www.who.int/mediacentre/factsheets/fs117/en/.
    [2] World Health Organization, Global strategy for dengue prevention and control 2012-2020, WHO, 2012. Available from: http://www.who.int/denguecontrol/9789241504034/en/.
    [3] Health Office (Dinas Kesehatan) of the East Java, Dinas Kesehatan Provinsi Jawa Timur, Surabaya, Indonesia, 2009.
    [4] Ministry of Health Republic of Indonesia, Kementerian Kesehatan Republik Indonesia Jakarta, 2014.
    [5] D. Aldila, T. Gotz, E. Soewono, An optimal control problem arising from a dengue disease transmission model, Math. Biosci., 242 (2013), 9-16. doi: 10.1016/j.mbs.2012.11.014
    [6] H. Tasman, A. K. Supriatna, N. Nuraini, et al. A dengue vaccination model for immigrants in a two-age-class population, Int. J. Math., 2012 (2012), 1-15.
    [7] J. P. Chavez, T. Gotz, S. Siegmund, et al. An SIR-Dengue transmission model with seasonal effects and impulsive control, Math. Biosci., 289 (2017), 29-39. doi: 10.1016/j.mbs.2017.04.005
    [8] A. Pandey, A. Mubayi, J. Medlock, Comparing vector-host and SIR models for dengue transmission, Math. Biosci., 246 (2013), 252-259. doi: 10.1016/j.mbs.2013.10.007
    [9] T. Gotz, N. Altmeier, W. Bock, et al. Modeling dengue data from Semarang, Indonesia, Ecol. Complex., 30 (2017), 57-62. doi: 10.1016/j.ecocom.2016.12.010
    [10] F. B. Agusto, M. A. Khan, Optimal control strategies for dengue transmission in pakistan, Math. Biosci., 305 (2018), 102-121. doi: 10.1016/j.mbs.2018.09.007
    [11] N. Tutkun, Parameter estimation in mathematical models using the real coded genetic algorithms, Expert Syst. Appl., 36 (2009), 3342-3345. doi: 10.1016/j.eswa.2008.01.060
    [12] R. L. Haupt, S. E. Haupt, Practical Genetic Algorithms, Second Edition, John Wiley & Sons, 2004.
    [13] W. B. Roush, S. L. Branton, A Comparison of fitting growth models with a genetic algorithm and nonlinear regression, Poultry Sci., 84 (2005), 494-502. doi: 10.1093/ps/84.3.494
    [14] W. S. W. Indratno, N. Nuraini, E. Soewono, A comparison of binary and continuous genetic algorithm in parameter estimation of a logistic growth model, American Institute of Physics Conference Series, 1587 (2014), 139-142.
    [15] Windarto, An Implementation of continuous genetic algorithm in parameter estimation of predator-prey model, AIP Conference Proceedings, 2016.
    [16] D. Akman, O. Akman, E. Schaefer, Parameter Estimation in Ordinary Differential Equations Modeling via Particle Swarm Optimization, J. Appl. Math., 2018 (2018), 1-9.
    [17] Windarto, Eridani, U. D. Purwati, A comparison of continuous genetic algorithm and particle swarm optimization in parameter estimation of Gompertz growth model, AIP Conference Proceedings, 2084 (2019), 020017.
    [18] R. Eberhart, J. Kennedy, A new optimizer using particle swarm theory, Proceedings of the Sixth International Symposium on Micro Machine and Human Science, 1995.
    [19] K. Diethelm, A fractional calculus based model for the simulation of an outbreak of dengue fever, Nonlinear Dyn., 71 (2013), 613-619. doi: 10.1007/s11071-012-0475-2
    [20] T. Sardar, S. Rana, J. Chattopadhyay, A mathematical model of dengue transmission with memory, Commun. Nonlinear Sci., 22 (2015), 511-525. doi: 10.1016/j.cnsns.2014.08.009
    [21] M. A. Khan, S. Ullah, M. Farooq, A new fractional model for tuberculosis with relapse via Atangana-Baleanu derivative, Chaos, Solitons & Fractals, 116 (2018), 227-238.
    [22] S. Ullah, M. A. Khan, M. Farooq, Modeling and analysis of the fractional HBV model with Atangana-Baleanu derivative, The European Physical Journal Plus, 133 (2018), 313.
    [23] E. O. Alzahrani, M. A. Khan, Modeling the dynamics of Hepatitis E with optimal control, Chaos, Solitons & Fractals, 116 (2018), 287-301.
    [24] Fatmawati, E. M. Shaiful, M. I. Utoyo, A fractional order model for HIV dynamics in a two-sex population, Int. J. Math., 2018 (2018), 1-11.
    [25] Fatmawati, M. A. Khan, M. Azizah, et al. A fractional model for the dynamics of competition between commercial and rural banks in Indonesia, Chaos, Solitons & Fractals, 122 (2019), 32-46.
    [26] A. Atangana, D. Baleanu, New fractional derivatives with nonlocal and non-singular kernel: theory and application to heat transfer model, Therm. Sci., 20 (2016), 763-769. doi: 10.2298/TSCI160111018A
    [27] S. Qureshi, A. Atangana, Mathematical analysis of dengue fever outbreak by novel fractional operators with field data, Physica A, 526 (2019), 121127.
    [28] O. Diekmann, J. A. P. Heesterbeek, J. A. J. Metz, On the definition and the computation of the basic reproduction ratio R0 in models for infectious diseases in heterogenous populations, J. Math. Biol., 28 (1990), 362-382.
    [29] O. Diekmann, J. A. P. Heesterbeek, Mathematical Epidemiology of Infectious Diseases, Model Building, Analysis and Interpretation, John Wiley & Son, 2000.
    [30] P. van den Driessche, J. Watmough, Reproduction numbers and sub-threshold endemic equilibria for compartmental models of disease transmition, Math. Biosci., 180 (2002), 29-48. doi: 10.1016/S0025-5564(02)00108-6
    [31] Wikipedia contributors: East Java, Wikipedia. Available from: https://en.wikipedia.org/wiki/East-Java.
    [32] M. Toufik, A. Atangana, New numerical approximation of fractional derivative with non-local and non-singular kernel: application to chaotic models, Eur. Phys. J. Plus, 132 (2017), 444.
  • Reader Comments
  • © 2020 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(6715) PDF downloads(575) Cited by(27)

Figures and Tables

Figures(12)  /  Tables(3)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog