Hindawi Publishing Corporation International Journal of Diο¬erential Equations Volume 2016, Article ID 7246027, 8 pages http://dx.doi.org/10.1155/2016/7246027
Research Article On Fuzzy Improper Integral and Its Application for Fuzzy Partial Differential Equations ElHassan ElJaoui and Said Melliani Department of Mathematics, University of Sultan Moulay Slimane, P.O. Box 523, 23000 Beni Mellal, Morocco Correspondence should be addressed to ElHassan ElJaoui;
[email protected] Received 31 October 2015; Revised 20 December 2015; Accepted 3 January 2016 Academic Editor: Najeeb A. Khan Copyright Β© 2016 E. ElJaoui and S. Melliani. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. We establish some important results about improper fuzzy Riemann integrals; we prove some properties of fuzzy Laplace transforms, which we apply for solving some fuzzy linear partial differential equations of first order, under generalized Hukuhara differentiability.
1. Introduction Wu introduced in [1] the improper fuzzy Riemann integral and presented some of its elementary properties; then he studied numerically this kind of integrals. This notion was exploited by certain researchers to study fuzzy differential equations (FDEs) of first or second order utilizing fuzzy Laplace transform, namely, by Allahviranloo and Ahmadi in [2], then by Salahshour et al. (see [3, 4]), and by ElJaoui et al. in [5]. The objective of this paper is to study the improper fuzzy Riemann integrals by establishing some important results about the continuity and the differentiability of a fuzzy improper integral depending on a given parameter. These results are then employed to prove some fuzzy Laplace transformβs properties, which we use to solve fuzzy partial differential equations (FPDEs). The organization of the remainder of this work is as follows. Section 2 is reserved for preliminaries. In Section 3, the main results are proved and new properties of fuzzy Laplace transform are investigated. Then, in Section 4, the procedure for solving first-order FPDEs by fuzzy Laplace transform is proposed. Section 5 deals with some numerical examples. In Section 6, we present conclusion and a further research topic.
2. Preliminaries By ππ (R) we meant the set of all nonempty compact convex subsets of R, which is endowed with the usual addition and scalar multiplication. Denote (see [6]) πΈ = {π : R σ³¨β [0, 1] | π verifies (1) β (4) below} ,
(1)
where (1) π is normal; that is, βπ‘ β R for which π(π‘) = 1, (2) π is convex in the fuzzy sense, (3) π is upper semicontinuous, (4) the closure of its support supp π = {π‘ β R | π(π‘) > 0} is compact. For 0 < πΌ β€ 1, [π]πΌ = {π‘ β R | π(π‘) β₯ πΌ} denotes the πΌ-level set of π β πΈ. Then, it is obvious that [π]πΌ β ππ (R) for all π β πΈ, 0 β€ πΌ β€ 1, and πΌ
πΌ
πΌ
[π1 + π2 ] = [π1 ] + [π2 ] , πΌ
(2)
πΌ
[ππ] = π [π] .
Let π· : πΈ Γ πΈ β [0, β) be a function which is defined by the identity πΌ
πΌ
π· (π1 , π2 ) = sup π ([π1 ] , [π2 ] ) , 0β€πΌβ€1
(3)
2
International Journal of Differential Equations
where π is the Hausdorff distance defined in ππ (R). Then, it is clear that (πΈ, π·) is a complete metric space (for more details about the metric π· see [7]). Definition 1 (see [2]). One defines a fuzzy number V in parametric form as a couple (V, V) of mappings V(πΌ) and V(πΌ), 0 β€ πΌ β€ 1, verifying the following properties: (1) V(πΌ) is bounded increasing left continuous in ]0, 1] and right continuous at 0. (2) V(πΌ) is bounded decreasing left continuous in ]0, 1] and right continuous at 0. (3) V(πΌ) β€ V(πΌ) for all 0 β€ πΌ β€ 1.
(4)
π
π
πΎ(πΌ) and β«π |πΉ(π₯, πΌ)|ππ₯ β€ πΎ(πΌ) for every π β₯ π. Then πΉ(π₯) is fuzzy Riemann integrable (in the sense of Wu) on [π, β[, its β improper fuzzy integral β«π πΉ(π₯)ππ₯ β πΈ, and β
β
π
π
π
(5)
For π1 , π2 β πΈ, if there exists an element π3 β πΈ such that π3 = π1 + π2 , then π3 is called the Hukuhara difference of π1 and π2 , which we denote by π1 β π2 . Definition 3 (see [2]). A mapping πΉ : (π, π) β πΈ is said to be strongly generalized differentiable at π₯ β (π, π), if there exists πΉσΈ (π₯) β πΈ, such that (i) for all β > 0 being very small, there exist πΉ(π₯ + β) β πΉ(π₯); πΉ(π₯) β πΉ(π₯ β β); and the limits πΉ (π₯ + β) β πΉ (π₯) πΉ (π₯) β πΉ (π₯ β β) = lim+ ββ0 β β
or (iv) for all β > 0 being very small, there exist πΉ(π₯) β πΉ(π₯ + β); πΉ(π₯) β πΉ(π₯ β β); and the limits
(6)
= πΉ (π₯) ,
(9)
The next theorem permits us to consider only case (i) or case (ii) of Definition 3 almost everywhere in the domain of the mappings studied. Theorem 4 (see [9]). If πΉ : (π, π) β πΈ is a strongly generalized differentiable function on (π, π) in the sense of Definition 3, (iii) or (iv), then πΉσΈ (π₯) β R for each π₯ β (π, π). Theorem 5 (see, e.g., [10]). We consider a fuzzy function πΉ : R β πΈ which is represented by πΉ(π₯) = (πΉ(π₯, πΌ), πΉ(π₯, πΌ)), for all πΌ β [0, 1]: (1) If πΉ is (i)-differentiable, then the crisp πΉ(π₯, πΌ) and πΉ(π₯, πΌ) are differentiable and σΈ (πΉσΈ (π₯, πΌ), πΉ (π₯, πΌ)). (2) If πΉ is (ii)-differentiable, then the crisp πΉ(π₯, πΌ) and πΉ(π₯, πΌ) are differentiable and σΈ (πΉ (π₯, πΌ), πΉσΈ (π₯, πΌ)).
functions πΉσΈ (π₯) = functions πΉσΈ (π₯) =
Definition 6 (see [2]). If πΉ : [0, β[β πΈ is a continuous mapping such that πβπ π₯ πΉ(π₯) is fuzzy Riemann integrable β on [0, β[ then β«0 πβπ π₯ πΉ(π₯)ππ₯ is called the fuzzy Laplace transform of πΉ which one denotes by β
L [πΉ (π₯)] = β« πβπ π₯ πΉ (π₯) ππ₯, π > 0. 0
(10)
L [πΉ (π₯)] = (L (πΉ (π₯, πΌ)) , L (πΉ (π₯, πΌ))) .
(11)
Theorem 7 (see [2]). Let πΉ : [0, β[β πΈ be a fuzzy valued function and πΉσΈ its derivative on [0, β[. Then, if πΉ is (i)differentiable
or (ii) for all β > 0 being very small, there exist πΉ(π₯) β πΉ(π₯ + β); πΉ(π₯ β β) β πΉ(π₯); and the limits πΉ (π₯) β πΉ (π₯ + β) πΉ (π₯ β β) β πΉ (π₯) lim = lim+ ββ0+ ββ0 (ββ) (ββ)
πΉ (π₯) β πΉ (π₯ + β) πΉ (π₯) β πΉ (π₯ β β) = lim+ ββ0 β (ββ)
Denote by L(π(π₯)) the classical Laplace transform of a crisp function π(π₯), and then
σΈ
= πΉσΈ (π₯)
= πΉ (π₯) ,
= πΉ (π₯) .
positive constants πΎ(πΌ) and πΎ(πΌ) such that β«π |πΉ(π₯, πΌ)|ππ₯ β€
β
(8)
σΈ
Theorem 2 (see [1]). One considers a fuzzy valued function πΉ(π₯) = (πΉ(π₯, πΌ), πΉ(π₯, πΌ)) defined on [π, β[. Suppose that, for all fixed πΌ β [0, 1], the crisp functions πΉ(π₯, πΌ), πΉ(π₯, πΌ) are integrable on [π, π], for every π β₯ π, and that there exist two
β« πΉ (π₯) ππ₯ = (β« πΉ (π₯, πΌ) ππ₯, β« πΉ (π₯, πΌ) ππ₯) .
πΉ (π₯ + β) β πΉ (π₯) πΉ (π₯ β β) β πΉ (π₯) = lim+ ββ0 β (ββ) σΈ
lim
σ΅¨σ΅¨ σ΅¨ σ΅¨σ΅¨ σ΅¨ π· (π1 , π2 ) = sup max {σ΅¨σ΅¨σ΅¨ππΌ β ππΌ σ΅¨σ΅¨σ΅¨ , σ΅¨σ΅¨σ΅¨ππΌ1 β ππΌ2 σ΅¨σ΅¨σ΅¨} . 1 2 σ΅¨ σ΅¨ 0β€πΌβ€1
ββ0
lim
ββ0+
ββ0+
The following identity holds true (see [8]):
lim+
or (iii) for all β > 0 being very small, there exist πΉ(π₯ + β) β πΉ(π₯); πΉ(π₯ β β) β πΉ(π₯); and the limits
L [πΉσΈ (π₯)] = π L [πΉ (π₯)] β πΉ (0)
(12)
or if πΉ is (ii)-differentiable (7)
L [πΉσΈ (π₯)] = (βπΉ (0)) β (βπ ) L [πΉ (π₯)] provided that the Laplace transforms of πΉ and πΉσΈ exist.
(13)
International Journal of Differential Equations
3 (see [1]), such that the real function π·(π1 (π‘), π2 (π‘)) is integrable on πΌ, and then
3. Continuity and Differentiability of Fuzzy Improper Integral In this section, πΌ denotes one of the intervals, ] β β, π] or [π, β[ or ] β β, β[, where π β R, π½ denotes another interval, and π΄ is a nonempty subset of R. Lemma 8. Let π1 (π‘), π2 (π‘) be two fuzzy valued functions, which are fuzzy Riemann integrable on πΌ, in the sense of Wu
π· (β« π1 (π‘) ππ‘, β« π2 (π‘) ππ‘) β€ β« π· (π1 (π‘) , π2 (π‘)) ππ‘. (14) πΌ
πΌ
πΌ
Proof. From identity (4), we have
σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨ π· (β« π1 (π‘) ππ‘, β« π2 (π‘) ππ‘) = sup max {σ΅¨σ΅¨σ΅¨σ΅¨β« π (π‘, πΌ) β π (π‘, πΌ) ππ‘σ΅¨σ΅¨σ΅¨σ΅¨ , σ΅¨σ΅¨σ΅¨σ΅¨β« π1 (π‘, πΌ) ππ‘ β π2 (π‘, πΌ) ππ‘σ΅¨σ΅¨σ΅¨σ΅¨} 2 σ΅¨ πΌ 1 σ΅¨ σ΅¨ πΌ σ΅¨ πΌ πΌ 0β€πΌβ€1 σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨ σ΅¨ β€ sup max {β« σ΅¨σ΅¨σ΅¨π (π‘, πΌ) β π (π‘, πΌ)σ΅¨σ΅¨σ΅¨ ππ‘, β« σ΅¨σ΅¨σ΅¨σ΅¨π1 (π‘, πΌ) β π2 (π‘, πΌ)σ΅¨σ΅¨σ΅¨σ΅¨ ππ‘} 1 2 σ΅¨ πΌσ΅¨ πΌ 0β€πΌβ€1
(15)
σ΅¨σ΅¨ σ΅¨ σ΅¨σ΅¨ σ΅¨ β€ sup β« max {σ΅¨σ΅¨σ΅¨π (π‘, πΌ) β π (π‘, πΌ)σ΅¨σ΅¨σ΅¨ , σ΅¨σ΅¨σ΅¨σ΅¨π1 (π‘, πΌ) β π2 (π‘, πΌ)σ΅¨σ΅¨σ΅¨σ΅¨} ππ‘ 1 2 σ΅¨ σ΅¨ 0β€πΌβ€1 πΌ σ΅¨σ΅¨ σ΅¨ σ΅¨σ΅¨ σ΅¨ β€ β« sup max {σ΅¨σ΅¨σ΅¨π (π‘, πΌ) β π (π‘, πΌ)σ΅¨σ΅¨σ΅¨ , σ΅¨σ΅¨σ΅¨σ΅¨π1 (π‘, πΌ) β π2 (π‘, πΌ)σ΅¨σ΅¨σ΅¨σ΅¨} ππ‘ = β« π· (π1 (π‘) , π2 (π‘)) ππ‘. 1 2 σ΅¨ σ΅¨ πΌ 0β€πΌβ€1 πΌ
Theorem 9. Let πΉ(π₯, π‘) : π΄ Γ πΌ β πΈ be a fuzzy function, satisfying the following conditions:
By tending π β β and using assumption (π»2 ), we obtain σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨πΉ (π₯, π‘, πΌ)σ΅¨σ΅¨σ΅¨ β€ max {π0 (π‘) , π1 (π‘)} = π (π‘) , σ΅¨σ΅¨ σ΅¨ σ΅¨σ΅¨πΉ (π₯, π‘, πΌ)σ΅¨σ΅¨σ΅¨ β€ max {π0 (π‘) , π1 (π‘)} = β (π‘) . σ΅¨ σ΅¨
(π»1 ) For all π₯ β π΄, π‘ σ³¨β πΉ(π₯, π‘) is continuous on πΌ. (π»2 ) For each π‘ β πΌ, π₯ σ³¨β πΉ(π₯, π‘) is continuous on π΄ β R. (π»3 ) For all πΌ β [0, 1] there exist a couple of nonnegative, continuous crisp functions ππΌ (π‘) and ππΌ (π‘), which are integrable on πΌ verifying, for all π₯ β π΄, π‘ β πΌ: σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨πΉ (π₯, π‘, πΌ)σ΅¨σ΅¨σ΅¨ β€ ππΌ (π‘) , σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨πΉ (π₯, π‘, πΌ)σ΅¨σ΅¨σ΅¨ β€ ππΌ (π‘) . σ΅¨ σ΅¨
πΉ (π₯π , π‘, 1) β€ πΉ (π₯π , π‘, πΌ) β€ πΉ (π₯π , π‘, 0) .
π· (πΉ (π₯π , π‘) , πΉ (π₯, π‘)) σ΅¨ σ΅¨ = sup max {σ΅¨σ΅¨σ΅¨πΉ (π₯π , π‘, πΌ) β πΉ (π₯, π‘, πΌ)σ΅¨σ΅¨σ΅¨ , 0β€πΌβ€1
σ΅¨σ΅¨ σ΅¨ σ΅¨σ΅¨πΉ (π₯π , π‘, πΌ) β πΉ (π₯, π‘, πΌ)σ΅¨σ΅¨σ΅¨} , σ΅¨ σ΅¨
Proof. Let π₯ β π΄ and let {π₯π }β π=1 be a sequence of elements of π΄, which converges to π₯ as π β β. For π β N, π‘ β πΌ, and πΌ β [0, 1], we have πΉ (π₯π , π‘, 0) β€ πΉ (π₯π , π‘, πΌ) β€ πΉ (π₯π , π‘, 1) ,
Therefore
(16)
Therefore, the fuzzy mapping π(π₯) = β«πΌ πΉ(π₯, π‘)ππ‘ is continuous on π΄.
(19)
(20)
π· (πΉ (π₯π , π‘) , πΉ (π₯, π‘)) β€ 2 (π (π‘) + β (π‘)) . From (π»1 ) and (π»3 ), we deduce that the mappings π(π‘), β(π‘), and π·(πΉ(π₯π , π‘), πΉ(π₯, π‘)) are all integrable on πΌ. On the other hand, we get the following inequality from Lemma 8: π· (β« πΉ (π₯π , π‘) ππ₯, β« πΉ (π₯, π‘) ππ₯)
(17)
πΌ
πΌ
(21)
β€ β« π· (πΉ (π₯π , π‘) , πΉ (π₯, π‘)) ππ₯.
Thus
πΌ
σ΅¨ σ΅¨ σ΅¨ σ΅¨ σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨πΉ (π₯π , π‘, πΌ)σ΅¨σ΅¨σ΅¨ β€ max {σ΅¨σ΅¨σ΅¨πΉ (π₯π , π‘, 1)σ΅¨σ΅¨σ΅¨ , σ΅¨σ΅¨σ΅¨πΉ (π₯π , π‘, 0)σ΅¨σ΅¨σ΅¨} β€ max {π0 (π‘) , π1 (π‘)} = π (π‘) , σ΅¨ σ΅¨ σ΅¨ σ΅¨ σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨πΉ (π₯π , π‘, πΌ)σ΅¨σ΅¨σ΅¨ β€ max {σ΅¨σ΅¨σ΅¨πΉ (π₯π , π‘, 1)σ΅¨σ΅¨σ΅¨ , σ΅¨σ΅¨σ΅¨πΉ (π₯π , π‘, 0)σ΅¨σ΅¨σ΅¨} σ΅¨ σ΅¨ σ΅¨ σ΅¨ σ΅¨ σ΅¨ β€ max {π0 (π‘) , π1 (π‘)} = β (π‘) .
That is, (18)
π· (π (π₯π ) , π (π₯)) β€ β« π· (πΉ (π₯π , π‘) , πΉ (π₯, π‘)) ππ₯. πΌ
(22)
By assumption (π»2 ), we have π·(πΉ(π₯π , π‘), πΉ(π₯, π‘)) β 0 as π β β.
4
International Journal of Differential Equations
So, by the dominated convergence theorem, β«πΌ π·(πΉ(π₯π , π‘), πΉ(π₯, π‘))ππ₯ β 0 as π β β. From inequality (22), we deduce that π(π₯π ) β π(π₯) as π β β. Consequently, π is continuous on π΄. Lemma 10. One considers two fuzzy valued functions π1 (π‘), π2 (π‘) : πΌ β πΈ, which are fuzzy Riemann integrable on πΌ (in the sense of Wu), such that π1 (π‘) β π2 (π‘) exists for all π‘ β πΌ, then π1 (π‘) β π2 (π‘) is fuzzy Riemann integrable on πΌ, the Hukuhara difference β«πΌ π1 (π‘)ππ‘ β β«πΌ π2 (π‘)ππ‘ is well defined, and β« (π1 (π‘) β π2 (π‘)) ππ₯ = β« π1 (π‘) ππ‘ β β« π2 (π‘) ππ‘. πΌ
πΌ
πΌ
(23)
Proof. Let π(π‘) = π1 (π‘) β π2 (π‘); that is, π1 (π‘) = π2 (π‘) + π(π‘). It is clear that there exist positive constants πΎ(πΌ, π1 ), πΎ(πΌ, π1 ), πΎ(πΌ, π2 ), and πΎ(πΌ, π2 ) such that, for all π β€ π in πΌ, we have πσ΅¨ σ΅¨σ΅¨ σ΅¨ β« σ΅¨σ΅¨σ΅¨π (π‘, πΌ)σ΅¨σ΅¨σ΅¨ ππ‘ β€ πΎ (πΌ, π1 ) , 1 σ΅¨ π σ΅¨ π σ΅¨ σ΅¨ β« σ΅¨σ΅¨σ΅¨σ΅¨π1 (π‘, πΌ)σ΅¨σ΅¨σ΅¨σ΅¨ ππ‘ β€ πΎ (πΌ, π1 ) , π
σ΅¨σ΅¨ σ΅¨σ΅¨ β« σ΅¨σ΅¨σ΅¨π (π‘, πΌ)σ΅¨σ΅¨σ΅¨ ππ‘ β€ πΎ (πΌ, π2 ) , σ΅¨ π σ΅¨ 2 π
β«
π
π
π
σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨π (π‘, πΌ) β π (π‘, πΌ)σ΅¨σ΅¨σ΅¨ ππ‘ σ΅¨σ΅¨ σ΅¨σ΅¨ 1 2
(24)
(25)
(26)
β€ πΎ (πΌ, π1 ) + πΎ (πΌ, π2 ) . Then from Theorem 2, π(π‘) is fuzzy Riemann integrable on πΌ. By βlinearityβ of the fuzzy integral, we get πΌ
πΌ
πΌ
(28)
Therefore, the fuzzy mapping π(π₯) = β«πΌ πΉ(π₯, π‘)ππ‘ is (i)-differentiable on π½ and πσΈ (π₯) = β«
πΌ
ππΉ (π₯, π‘) ππ‘, ππ₯
βπ₯ β π½.
(29)
then the fuzzy function π(π₯) is (ii)-differentiable on π½ and (29) remains true.
πΉ (π₯ + π, π‘) β πΉ (π₯, π‘) { , π β ]0, π0 ] { π π1 (π, π‘) = { { ππΉ π = 0, (π₯, π‘) , { ππ₯ πΉ (π₯, π‘) β πΉ (π₯ β π, π‘) { , π β ]0, π0 ] { π π2 (π, π‘) = { { ππΉ π = 0. (π₯, π‘) , { ππ₯
and similarly
β« π1 (π‘) ππ‘ = β« π2 (π‘) ππ‘ + β« π (π‘) ππ‘.
σ΅¨σ΅¨ σ΅¨σ΅¨ ππΉ σ΅¨ σ΅¨σ΅¨ (π₯, π‘, πΌ)σ΅¨σ΅¨σ΅¨ β€ ππΌ (π‘) , σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨ ππ₯ σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨ ππΉ σ΅¨σ΅¨ β€ ππΌ (π‘) . σ΅¨σ΅¨ π‘, πΌ) (π₯, σ΅¨σ΅¨ σ΅¨σ΅¨ ππ₯ σ΅¨ σ΅¨
Proof. Assume that (π΄ 1 )β(π΄ 5 ) hold true. Let π₯ β π½, π0 > 0 being very small, and define the auxiliary functions
β€ πΎ (πΌ, π1 ) + πΎ (πΌ, π2 )
π π σ΅¨ σ΅¨ σ΅¨ σ΅¨ β« σ΅¨σ΅¨σ΅¨σ΅¨π (π‘, πΌ)σ΅¨σ΅¨σ΅¨σ΅¨ ππ‘ = β« σ΅¨σ΅¨σ΅¨σ΅¨π1 (π‘, πΌ) β π2 (π‘, πΌ)σ΅¨σ΅¨σ΅¨σ΅¨ ππ‘ π π
(π΄ 5 ) For all πΌ β [0, 1] there exist a couple of continuous crisp functions ππΌ (π‘) and ππΌ (π‘), which are integrable on πΌ verifying, for all π₯ β π½, π‘ β πΌ:
(π΄σΈ 2 ) for all π‘ β πΌ, π₯ σ³¨β πΉ(π₯, π‘) is (ii)-differentiable on π½,
Hence π
(π΄ 4 ) For all π‘ β πΌ, π₯ σ³¨β (ππΉ/ππ₯)(π₯, π‘) is continuous on π½.
Moreover, if one replaces assumption (π΄ 2 ) by the alternative condition
σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨π2 (π‘, πΌ)σ΅¨σ΅¨σ΅¨ ππ‘ β€ πΎ (πΌ, π2 ) . σ΅¨ σ΅¨
σ΅¨ σ΅¨ β« σ΅¨σ΅¨σ΅¨π (π‘, πΌ)σ΅¨σ΅¨σ΅¨ ππ‘ = β« π π
(π΄ 3 ) For all π₯ β π½, π‘ σ³¨β (ππΉ/ππ₯)(π₯, π‘) is continuous on πΌ.
(27)
Thus, β«πΌ π1 (π‘)ππ‘ββ«πΌ π2 (π‘)ππ‘ exists and β«πΌ π1 (π‘)ππ‘ββ«πΌ π2 (π‘)ππ‘ = β«πΌ π(π‘)ππ‘. Theorem 11. One considers a fuzzy valued function πΉ(π₯, π‘) : π½ Γ πΌ β πΈ, verifying the following assumptions: (π΄ 1 ) For all π₯ β π½, π‘ σ³¨β πΉ(π₯, π‘) is continuous and fuzzy Riemann integrable on πΌ. (π΄ 2 ) For all π‘ β πΌ, π₯ σ³¨β πΉ(π₯, π‘) is (i)-differentiable on the interval π½.
(30)
For fixed π β]0, π0 ], we have π (π₯ + π) β π (π₯) π =
1 (β« πΉ (π₯ + π, π‘) ππ‘ β β« πΉ (π₯, π‘) ππ‘) π πΌ πΌ
=β«
πΌ
(31)
πΉ (π₯ + π, π‘) β πΉ (π₯, π‘) ππ‘ = β« π1 (π, π‘) ππ‘, π πΌ
where the existence of the Hukuhara differences is ensured by the (i)-differentiability of π₯ σ³¨β πΉ(π₯, π‘) and by Lemma 10. Analogously, we get π (π₯) β π (π₯ β π) = β« π2 (π, π‘) ππ‘. π πΌ
(32)
From assumptions (π΄ 1 )β(π΄ 4 ), we deduce that π1 and π2 satisfy conditions (π»1 )-(π»2 ) of Theorem 9.
International Journal of Differential Equations
5
On the other hand, using the finite increments theorem, we obtain σ΅¨σ΅¨ σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨π (π, π‘, πΌ)σ΅¨σ΅¨σ΅¨ = σ΅¨σ΅¨σ΅¨ πΉ (π₯ + π, π‘, πΌ) β πΉ (π₯, π‘, πΌ) σ΅¨σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨ 1 π σ΅¨ σ΅¨
Lπ [π’π (π, π)] =
σ΅¨σ΅¨ ππΉ σ΅¨σ΅¨ σ΅¨ σ΅¨ β€ sup σ΅¨σ΅¨σ΅¨ (π₯ + V, π‘, πΌ)σ΅¨σ΅¨σ΅¨ β€ ππΌ (π‘) , σ΅¨σ΅¨ 0β€Vβ€π0 σ΅¨σ΅¨ ππ₯
σ΅¨ σ΅¨ σ΅¨ σ΅¨σ΅¨ πΉ (π₯ + π, π‘, πΌ) β πΉ (π₯, π‘, πΌ) σ΅¨σ΅¨σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨π1 (π, π‘, πΌ)σ΅¨σ΅¨σ΅¨ = σ΅¨σ΅¨σ΅¨σ΅¨ σ΅¨σ΅¨ π σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨σ΅¨ ππΉ σ΅¨σ΅¨σ΅¨ β€ sup σ΅¨σ΅¨σ΅¨σ΅¨ (π₯ + V, π‘, πΌ)σ΅¨σ΅¨σ΅¨σ΅¨ β€ ππΌ (π‘) . σ΅¨σ΅¨ 0β€Vβ€π0 σ΅¨σ΅¨ ππ₯
(33)
β
=
σ΅¨σ΅¨ ππΉ σ΅¨σ΅¨ σ΅¨ σ΅¨ β€ sup σ΅¨σ΅¨σ΅¨ (π₯ β V, π‘, πΌ)σ΅¨σ΅¨σ΅¨ β€ ππΌ (π‘) , σ΅¨ ππ₯ σ΅¨ σ΅¨σ΅¨ 0β€Vβ€π0 (34)
π (π₯) β π (π₯ β π) lim+ = β« π2 (0, π‘) ππ‘ πβ0 π πΌ =β«
πΌ
(35)
πΌ
Theorem 13. Let π’(π, π) be a fuzzy valued function on [0, β[Γ[0, β[ into πΈ. Suppose that the mappings π σ³¨β πΉ(π, π) = πβπ π π’(π, π) and π σ³¨β πΊ(π, π) = πβπ π π’π (π, π) are fuzzy Riemann integrable on [0, β[, for all π β₯ π 0 for some π 0 > 0. Consider the following:
(39)
Lπ [π’π (π, π)] = (βπ’ (π, 0)) β (βπ ) Lπ [π’ (π, π)] .
(40)
Proof. This is a direct result of Theorem 12, by fixing π β₯ 0 and taking the Laplace transforms with respect to π.
Our aim now is to solve the following first-order FPDE using the fuzzy Laplace transform method under strongly generalized differentiability: π’π (π, π) + ππ’π (π, π) = π (π, π, π’ (π, π)) , π’ (π, 0) = π (π) ,
Therefore, π is (i)-differentiable at π₯ and ππΉ (π₯, π‘) ππ‘. ππ₯
π (L [π’ (π, π)]) . ππ π
4. Fuzzy Laplace Transform Algorithm for First-Order Fuzzy Partial Differential Equations
ππΉ (π₯, π‘) ππ‘. ππ₯
πσΈ (π₯) = β«
(38)
(b) If π’(π₯, π) is (ii)-differentiable with respect to π, then
π (π₯ + π) β π (π₯) = β« π1 (0, π‘) ππ‘ π πΌ
πΌ
π (β« πΉ (π, π) ππ) , ππ 0
Lπ [π’π (π, π)] = π Lπ [π’ (π, π)] β π’ (π, 0) ,
Inequalities (33) and (34), which are obviously also true for π = 0, ensure that π1 and π2 satisfy condition (π»3 ) of Theorem 9. Applying the latter theorem, we get
ππΉ (π₯, π‘) ππ‘, ππ₯
0
β
(a) If π’(π, π) is (i)-differentiable with respect to π, then
σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨ ππΉ σ΅¨ β€ sup σ΅¨σ΅¨σ΅¨σ΅¨ (π₯ β V, π‘, πΌ)σ΅¨σ΅¨σ΅¨σ΅¨ β€ ππΌ (π‘) . ππ₯ σ΅¨ σ΅¨σ΅¨ 0β€Vβ€π0 σ΅¨
=β«
β
0
σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨π (π, π‘, πΌ)σ΅¨σ΅¨σ΅¨ = σ΅¨σ΅¨σ΅¨ πΉ (π₯, π‘, πΌ) β πΉ (π₯ β π, π‘, πΌ) σ΅¨σ΅¨σ΅¨ σ΅¨σ΅¨ 2 σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨ β σ΅¨ σ΅¨
πβ0
(37)
Lπ [π’π (π, π)] = β« πβπ π π’π (π, π) ππ = β« πΉπ (π, π) ππ
Lπ [π’π (π, π)] =
σ΅¨ σ΅¨ σ΅¨ σ΅¨σ΅¨ πΉ (π₯, π‘, πΌ) β πΉ (π₯ β π, π‘, πΌ) σ΅¨σ΅¨σ΅¨σ΅¨ σ΅¨σ΅¨ σ΅¨σ΅¨π2 (π, π‘, πΌ)σ΅¨σ΅¨σ΅¨ = σ΅¨σ΅¨σ΅¨σ΅¨ σ΅¨σ΅¨ β σ΅¨σ΅¨ σ΅¨σ΅¨
π (L [π’ (π, π)]) . ππ π
Proof. For fixed π β₯ π 0 , then using Theorem 11 we have
Similarly, we have
lim+
Let Lπ [π’(π, π)] or L[π’(π, π)] (for short) denote the fuzzy Laplace transform of π’(π, π) with respect to the time variable π. Then
(41)
π’ (0, π) = β (π) , (36)
The proof under assumption (π΄σΈ 2 ) instead of (π΄ 2 ) is similar to the first case. Theorem 12. One considers a fuzzy function π’(π, π) : [0, β[Γ[0, β[β πΈ. Suppose that the mapping πΉ(π, π) = πβπ π π’(π, π) satisfies assumptions (π΄ 1 )β(π΄ 5 ) above, for all π β₯ π 0 for some π 0 > 0.
where π’(π, π) is a fuzzy function of π β₯ 0, π β₯ 0, π is a real constant, and π(π, π, π’), π(π), and β(π) are fuzzy valued functions, such that π(π, π, π’) is linear with respect to π’. For short, assume that π β₯ 0 (case π < 0 is similar). By using fuzzy Laplace transform with respect to π, we get Lπ [π’π (π, π)] + πLπ [π’π (π, π)] = Lπ [π (π, π, π’ (π, π))] . (42)
Therefore, we have to distinguish the following cases for solving (42):
6
International Journal of Differential Equations (a) Case 1: If π’ is (i)-differentiable with respect to π and π, then by Laplace transform L [π’π (π, π, πΌ)] + πL [π’π (π, π, πΌ)] (43)
= L [π (π, π, π’ (π, π))] ,
π, πΌ), π’(π, π, πΌ))} and π(π, π, π’(π, π), πΌ) = max{π(π, π, V)/V β (π’(π, π, πΌ), π’(π, π, πΌ))}. Using Theorems 12 and 13 we get the following differential system: π (L [π’ (π, π, πΌ)]) + ππ L [π’ (π, π, πΌ)] ππ (44)
(45)
L [π’ (π, π, πΌ)] = π»3 (π , πΌ) , L [π’ (π, π, πΌ)] = πΎ3 (π , πΌ) ,
Assume that this leads to (46)
where (π»1 (π , πΌ), πΎ1 (π , πΌ)) is solution of system (44) under (45). By the inverse Laplace transform we get
π’ (π, π, πΌ) = Lβ1 [πΎ1 (π , πΌ)] .
(47)
(b) Case 2: If π’ is (i)-differentiable with respect to π and (ii)-differentiable with respect to π, then by Theorems 12 and 13 we get the following differential system, satisfying the initial conditions (45):
π’ (π, π, πΌ) = Lβ1 [π»3 (π , πΌ)] , π’ (π, π, πΌ) = Lβ1 [πΎ3 (π , πΌ)] .
π (L [π’ (π, π, πΌ)]) + ππ L [π’ (π, π, πΌ)] ππ = ππ (π) + L [π (π, π, π’ (π, π))] ,
= ππ (π) + L [π (π, π, π’ (π, π))] ,
= ππ (π) + L [π (π, π, π’ (π, π))] .
= ππ (π) + L [π (π, π, π’ (π, π))] .
(53)
(d) Case 4: If π’ is (ii)-differentiable with respect to π and π, then we get the following differential system, satisfying the initial conditions (45):
π (L [π’ (π, π, πΌ)]) + ππ L [π’ (π, π, πΌ)] ππ
(48)
(52)
where (π»3 (π , πΌ), πΎ3 (π , πΌ)) is solution of system (51) under (45). Therefore
π (L [π’ (π, π, πΌ)]) + ππ L [π’ (π, π, πΌ)] ππ
π (L [π’ (π, π, πΌ)]) + ππ L [π’ (π, π, πΌ)] ππ
(51)
Assume that this implies
L [π’ (0, π, πΌ)] = L [β (π, πΌ)] ,
π’ (π, π, πΌ) = Lβ1 [π»1 (π , πΌ)] ,
= ππ (π) + L [π (π, π, π’ (π, π))] ,
= ππ (π) + L [π (π, π, π’ (π, π))] .
satisfying the following initial conditions:
L [π’ (π, π, πΌ)] = πΎ1 (π , πΌ) ,
(50)
(c) Case 3: If π’ is (ii)-differentiable with respect to π and (i)-differentiable with respect to π, then we get the following differential system, satisfying the initial conditions (45):
π (L [π’ (π, π, πΌ)]) + ππ L [π’ (π, π, πΌ)] ππ
= ππ (π) + L [π (π, π, π’ (π, π))]
L [π’ (π, π, πΌ)] = π»1 (π , πΌ) ,
π’ (π, π, πΌ) = Lβ1 [πΎ2 (π , πΌ)] .
π (L [π’ (π, π, πΌ)]) + ππ L [π’ (π, π, πΌ)] ππ
= ππ (π) + L [π (π, π, π’ (π, π))] ,
L [π’ (0, π, πΌ)] = L [β (π, πΌ)] .
(49)
where (π»2 (π , πΌ), πΎ2 (π , πΌ)) is solution of system (48) under (45). Thus π’ (π, π, πΌ) = Lβ1 [π»2 (π , πΌ)] ,
where π(π, π, π’(π, π), πΌ) = min{π(π, π, V)/V β (π’(π,
π (L [π’ (π, π, πΌ)]) + ππ L [π’ (π, π, πΌ)] ππ
L [π’ (π, π, πΌ)] = π»2 (π , πΌ) , L [π’ (π, π, πΌ)] = πΎ2 (π , πΌ) ,
= L [π (π, π, π’ (π, π))] , L [π’π (π, π, πΌ)] + πL [π’π (π, π, πΌ)]
Assume that this implies
(54)
Assume that this leads to L [π’ (π, π, πΌ)] = π»4 (π , πΌ) , L [π’ (π, π, πΌ)] = πΎ4 (π , πΌ) ,
(55)
International Journal of Differential Equations
7
where (π»4 (π, πΌ), πΎ4 (π, πΌ)) is solution of system (54) under (45). Hence
The lengths of π’, π’π , and π’π are, respectively, given by len (π’ (π, π, πΌ)) = π’ (π, π, πΌ) β π’ (π, π, πΌ) = 2 (1 β πΌ) (3π + π) β₯ 0,
π’ (π, π, πΌ) = Lβ1 [π»4 (π , πΌ)] ,
(56)
π’ (π, π, πΌ) = Lβ1 [πΎ4 (π , πΌ)] .
len (π’π (π, π, πΌ)) = π’π (π, π, πΌ) β π’π (π, π, πΌ) = 6 (1 β πΌ) β₯ 0,
5. Numerical Examples
len (π’π (π, π, πΌ)) = π’π (π, π, πΌ) β π’π (π, π, πΌ)
Example 1. Consider
= 2 (1 β πΌ) β₯ 0.
π’π (π, π) = 3π’π (π, π) + π, π’ (π, 0, πΌ) = 3π β
(πΌ, 2 β πΌ) +
So, this solution is valid for all π β₯ 0 and π β₯ 0. Case 2. If π’ is (i)-differentiable with respect to π and (ii)differentiable with respect to π, then analogously
π2 , 2
(57)
π’ (0, π, πΌ) = π β
(πΌ, 2 β πΌ) ,
Case 1. If π’ is (i)-differentiable with respect to π and π, then by Laplace transform we get 3π π (L [π’ (π, π, πΌ)]) = 3π L [π’ (π, π, πΌ)] β 9πΌπ β ππ 2
π (L [π’ (π, π, πΌ)]) = 3π L [π’ (π, π, πΌ)] + 9 (πΌ β 2) π ππ
(58)
This differential system satisfies the following initial conditions:
(2 β πΌ) . L [π’ (0, π, πΌ)] = L [(2 β πΌ) π] = π 2
(59)
L [π’ (π, π, πΌ)] =
π
{1 π» (π) = { 0 {
πΌ , π 2
((6 β 3πΌ) π + π2 /2) π
(60) +
2βπΌ . π 2
+ (2 β πΌ) π +
π2 π’ (π, π, πΌ) = (2 β πΌ) (3π + π) + . 2
π2 , 2
(61)
(65)
π2 . 2
As in Case 2, one can verify that this solution is valid only over ΞσΈ = {(π, π) | π β₯ 0, π β₯ 0, π β₯ 3π}. Case 4. If π’ is (ii)-differentiable with respect to π and π, then π2 , 2
π2 π’ (π, π, πΌ) = (2 β πΌ) (3π + π) + . 2
2
π , 2
(64)
π’ (π, π, πΌ) = 2 (πΌ β 1) (π β 3π) π» (π β 3π) + 3πΌπ
π’ (π, π, πΌ) = πΌ (3π + π) +
By the inverse Laplace transform we deduce π’ (π, π, πΌ) = πΌ (3π + π) +
π < 0.
Case 3. If π’ is (ii)-differentiable with respect to π and (i)differentiable with respect to π, then similarly
+ πΌπ + +
πβ₯0
π’ (π, π, πΌ) = 2 (1 β πΌ) (π β 3π) π» (π β 3π) + 3 (2 β πΌ) π
Solving (58) under (59), we get (3πΌπ + π2 /2)
π2 , 2
Therefore, this solution π’ is valid only over Ξ = {(π, π) | π β₯ 0, π β₯ 0, π β€ 3π}.
3π π + . 2 π
πΌ , π 2
(63)
where π» is the unit step function or the Heaviside function:
2
L [π’ (0, π, πΌ)] = L [πΌπ] =
π2 , 2
π’ (π, π, πΌ) = 2 (1 β πΌ) (π β 3π) π» (π β 3π) + 3 (2 β πΌ) π + πΌπ +
2
π + , π
β
π’ (π, π, πΌ) = 2 (πΌ β 1) (π β 3π) π» (π β 3π) + 3πΌπ + (2 β πΌ) π +
π β₯ 0, π β₯ 0.
L [π’ (π, π, πΌ)] =
(62)
(66)
One can verify that function π’ is not (ii)-differentiable with respect to either π or π. So, no solution exists in this case.
8
International Journal of Differential Equations
Conflict of Interests
Example 2. Consider π’π (π, π) = π’π (π, π) ,
The authors declare that there is no conflict of interests regarding the publication of this paper.
π’ (π, 0, πΌ) = cos (π) β
(πΌ, 2 β πΌ) , π’ (0, π, πΌ) = cos (π) β
(πΌ, 2 β πΌ) ,
(67)
References
π β₯ 0, π β₯ 0. Case 1. If π’ is (i)-differentiable with respect to π and π, then, by application of the algorithm above, one obtains π’ (π, π, πΌ) = πΌ cos (π + π) , π’ (π, π, πΌ) = (2 β πΌ) cos (π + π) .
(68)
The lengths of π’, π’π , and π’π are, respectively, given by len (π’ (π, π, πΌ)) = 2 (1 β πΌ) cos (π + π) , len (π’π (π, π, πΌ)) = β2 (1 β πΌ) sin (π + π) ,
(69)
len (π’π (π, π, πΌ)) = β2 (1 β πΌ) sin (π + π) . So, this solution is valid for all π β₯ 0, π β₯ 0: π + π β [3π/2 + 2ππ, 2π + 2ππ], π β Z. Case 2. If π’ is (i)-differentiable with respect to π and (ii)differentiable with respect to π, therefore π’ (π, π, πΌ) = (πΌ β 1) cos (π β π) + cos (π + π) , π’ (π, π, πΌ) = (1 β πΌ) cos (π β π) + cos (π + π) .
(70)
Then, this solution is valid for all π β₯ 0, π β₯ 0: π β π β [3π/2 + 2ππ, 2π + 2ππ], π β Z. Case 3. If π’ is (ii)-differentiable with respect to π and (i)differentiable with respect to π, then analogously π’ (π, π, πΌ) = (πΌ β 1) cos (π β π) + cos (π + π) , π’ (π, π, πΌ) = (1 β πΌ) cos (π β π) + cos (π + π) .
(71)
Hence, this solution is valid for all π β₯ 0, π β₯ 0: π β π β [2ππ, π/2 + 2ππ], π β Z. Case 4. If π’ is (ii)-differentiable with respect to π and π, then similarly π’ (π, π, πΌ) = πΌ cos (π + π) , π’ (π, π, πΌ) = (2 β πΌ) cos (π + π) .
(72)
So, this solution is valid for all π β₯ 0, π β₯ 0: π + π β [2ππ, π/2 + 2ππ], π β Z.
6. Conclusion Theorems of continuity and differentiability for a fuzzy function defined via a fuzzy improper Riemann integral are proved which are used to prove some results concerning fuzzy Laplace transform. Then, using Laplace transform method, the solutions for some linear fuzzy partial differential equations (FPDEs) of first order are given. For future research, one can apply this method to solve FPDEs of high order.
[1] H.-C. Wu, βThe improper fuzzy Riemann integral and its numerical integration,β Information Sciences, vol. 111, no. 1β4, pp. 109β137, 1998. [2] T. Allahviranloo and M. B. Ahmadi, βFuzzy Laplace transforms,β Soft Computing, vol. 14, no. 3, pp. 235β243, 2010. [3] S. Salahshour, M. Khezerloo, S. Hajighasemi, and M. Khorasany, βSolving fuzzy integral equations of the second kind by fuzzy laplace transform method,β International Journal of Industrial Mathematics, vol. 4, no. 1, pp. 21β29, 2012. [4] S. Salahshour and T. Allahviranloo, βApplications of fuzzy Laplace transforms,β Soft Computing, vol. 17, no. 1, pp. 145β158, 2013. [5] E. ElJaoui, S. Melliani, and L. S. Chadli, βSolving second-order fuzzy differential equations by the fuzzy Laplace transform method,β Advances in Difference Equations, vol. 2015, article 66, 14 pages, 2015. [6] O. Kaleva, βFuzzy differential equations,β Fuzzy Sets and Systems, vol. 24, no. 3, pp. 301β317, 1987. [7] M. L. Puri and D. A. Ralescu, βFuzzy random variables,β Journal of Mathematical Analysis and Applications, vol. 114, no. 2, pp. 409β422, 1986. [8] I. J. Rudas, B. Bede, and A. L. Bencsik, βFirst order linear fuzzy differential equations under generalized differentiability,β Information Sciences, vol. 177, no. 7, pp. 1648β1662, 2007. [9] B. Bede and S. G. Gal, βGeneralizations of the differentiability of fuzzy-number-valued functions with applications to fuzzy differential equations,β Fuzzy Sets and Systems, vol. 151, no. 3, pp. 581β599, 2005. [10] Y. Chalco-Cano and H. RomΒ΄an-Flores, βOn new solutions of fuzzy differential equations,β Chaos, Solitons & Fractals, vol. 38, no. 1, pp. 112β119, 2008.
Advances in
Operations Research Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Advances in
Decision Sciences Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Journal of
Applied Mathematics
Algebra
Hindawi Publishing Corporation http://www.hindawi.com
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Journal of
Probability and Statistics Volume 2014
The Scientific World Journal Hindawi Publishing Corporation http://www.hindawi.com
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
International Journal of
Differential Equations Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Volume 2014
Submit your manuscripts at http://www.hindawi.com International Journal of
Advances in
Combinatorics Hindawi Publishing Corporation http://www.hindawi.com
Mathematical Physics Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Journal of
Complex Analysis Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
International Journal of Mathematics and Mathematical Sciences
Mathematical Problems in Engineering
Journal of
Mathematics Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Volume 2014
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Discrete Mathematics
Journal of
Volume 2014
Hindawi Publishing Corporation http://www.hindawi.com
Discrete Dynamics in Nature and Society
Journal of
Function Spaces Hindawi Publishing Corporation http://www.hindawi.com
Abstract and Applied Analysis
Volume 2014
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
International Journal of
Journal of
Stochastic Analysis
Optimization
Hindawi Publishing Corporation http://www.hindawi.com
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Volume 2014