This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
|
|
- Jeffrey Lee
- 6 years ago
- Views:
Transcription
1 This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier s archiving and manuscript policies are encouraged to visit:
2 Available online at Fuzzy Sets and Systems 2 (212) Lipschitz functions and fuzzy number approximations Lucian Coroianu a,b, a Department of Mathematics and Informatics, University of Oradea, Universităţii 1, 4187 Oradea, Romania b Department of Mathematics, Babes-Bolyai University of Cluj-Napoca, Mihail Kogalniceanu 1, 484 Cluj-Napoca, Romania Received September 211; received in revised form 1 December 211; accepted 2 January 212 Available online 2 January 212 Abstract We prove that some important properties of convex subsets of vector topological spaces remain valid in the space of fuzzy numbers endowed with the Euclidean distance. We use these results to obtain a characterization of fuzzy number-valued Lipschitz functions. This fact helps us to find the best Lipschitz constant of the trapezoidal approximation operator preserving the value and ambiguity introduced in a recent paper. Finally, applications in finding within a reasonable error the trapezoidal approximation of a fuzzy number preserving the value and ambiguity in the case when the direct formula is not applicable and an estimation for the defect of additivity of the trapezoidal approximation preserving the value and ambiguity are given. 212 Elsevier B.V. All rights reserved. Keywords: Fuzzy number; Trapezoidal fuzzy number; Approximation; Ambiguity; Value; Convexity; Lipschitz constant 1. Introduction In the last few years many papers investigated on the approximation of fuzzy numbers with respect to well-known metrics [1 5,7 1,15,16,21 26,28,2 8]. Mostly, two kinds of problems are considered: approximations without any other restriction and approximations with additional conditions. We recall here some important contributions with respect to the problem of the approximation of fuzzy numbers. Firstly, we discuss about approximations without additional conditions. Chanas [15] and Grzegorzewski [21] independently proposed the interval approximation of a fuzzy number. Grzegorzewski proved that the nearest interval approximation of a fuzzy number with respect to the Euclidean metric is actually its expected interval. Abbasbandy and Asady [] proposed the trapezoidal approximation of a fuzzy number with respect to the same Euclidean distance. Yeh [4] proposed new algorithms for computing trapezoidal and triangular approximations of fuzzy numbers. Zeng and Li [8] proposed the triangular approximation of a fuzzy number with respect to a weighted L 2 -type metric. Unfortunately, the algorithm proposed by them does not always produce proper triangular fuzzy numbers as it was pointed out in the papers [9,4]. The correct algorithm is given by Ban in the paper [9]. The most general result in approximations with trapezoidal or triangular fuzzy numbers is given by Yeh in the paper [7] where algorithms for computing trapezoidal or triangular approximations of fuzzy numbers Correspondence address: Department of Mathematics and Informatics, University of Oradea, Universităţii 1, 4187 Oradea, Romania. Fax: address: lcoroianu@uoradea.ro /$ - see front matter 212 Elsevier B.V. All rights reserved. doi:1.116/j.fss
3 L. Coroianu / Fuzzy Sets and Systems 2 (212) with respect to general weighted L 2 -type metrics are given. A generalization of the notion of trapezoidal fuzzy number is the parametric (also known as semi-trapezoidal) fuzzy number introduced by Nasibov and Peker in [28]. In the same paper they proposed the parametric approximation of a fuzzy number with respect to the Euclidean metric, result improved by Ban in [8]. ThenYeh[6] generalized these results by considering general weighted L 2 -type metrics. Now, we discuss about approximations of fuzzy numbers under additional conditions. Grzegorzewski and Mrówka [22,2] proposed the trapezoidal approximation of a fuzzy number preserving the expected interval with respect to the Euclidean metric. Then Ban [7] and Yeh [5] independently improved their result. Algorithms for computing the trapezoidal approximation of a fuzzy number preserving the expected interval can be found in the papers [24,25].Then Abbasbandy and Hajjari [4] proposed approximations of fuzzy numbers preserving the core with respect to weighted L 2 -type metrics. Recently, Ban and all [1] proposed the trapezoidal approximation preserving the value and ambiguity with respect to the Euclidean metric. Interestingly, the problem of approximating fuzzy numbers by trapezoidal fuzzy numbers preserving the value and ambiguity goes back to a paper of Delgado and all [18] where they considered that two fuzzy numbers with the same value and ambiguity should be considered equal. More generally, Ban and Coroianu [1] proposed simpler methods to compute the parametric approximation of a fuzzy number with respect to the Euclidean metric preserving important characteristics such as expected interval or the ambiguity and value. Other types of nonlinear approximations can be found in the paper of Grzegorzewski and Stefanini [26] where they proposed classes of fuzzy numbers depending on four or five parameters which allow approximations with conservation of multiple characteristics of fuzzy numbers such as the support and core or the ambiguity and value. The quality of a trapezoidal, triangular or parametric approximation operator is important nevertheless. For this reason, Grzegorzewski and Mrówka [22] proposed a list of criteria that a trapezoidal approximation operator should posses. Most of these approximation operators own important properties such as: translation invariance, scale invariance, or identity criteria. Another important property that an approximation operator should posses is the continuity. One would expect that if fuzzy number A is close to fuzzy number B then their approximations are also close one to another. Yeh [4,6,7] proved that the approximation operators without additional conditions are nonexpansives. Ban and Coroianu [11] proved that the trapezoidal approximation operator preserving the expected interval satisfies the Lipschitz condition. Then Coroianu [16] found the best Lipschitz constant of the discussed operator. Recently, in the paper [1] it was proved that the trapezoidal approximation operator preserving the value and ambiguity satisfies the Lipschitz condition too. As a negative result, in the paper [12] it was proved that any trapezoidal fuzzy number-valued operator (not necessarily an approximation operator) preserving core is discontinuous with respect to any weighted L 2 -type metric and each fuzzy number with the 1-cut set as a proper interval is a point of discontinuity. The aim of this paper is to provide a method which simplifies the investigation on distance properties of the trapezoidal approximation operators and especially on the finding of the best Lipschitz constant if possible. This is motivated by the fact that the method used in the case of the trapezoidal approximation preserving the expected interval (see the proof of Theorem 7 in [16]) cannot be used in general. Indeed, while in the case of the trapezoidal approximation preserving the expected interval the problem is reduced to geometrical reasonings in the Euclidean space R 2 (this is possible because two out of the four components of the trapezoidal approximation are given by formulas which allow best possible estimations), in general all the four components of the trapezoidal approximation are needed to obtain better estimations. This seems to be the case of the trapezoidal approximation preserving the value and ambiguity where the method proposed in the case of the trapezoidal approximation operator preserving the expected interval seems to be much more difficult to apply. However, most of the approximation operators including the two operators mentioned before have something in common. They are given on cases each one corresponding to a closed convex subset of the space of fuzzy numbers. When we say convex subset of the space of fuzzy numbers we are aware of the fact that the concept of convexity is usually in relation with the notion of vector space. Still, even if the addition and the scalar multiplication of fuzzy numbers have not the necessary properties to provide a vector structure over the space of fuzzy numbers, we can define the notion of convex subset of the space of fuzzy numbers exactly as in the case of vector spaces. The idea is not new, Prakash and Sertel introduced the notion of convex subset in semivector topological spaces in [29]. Actually, one can easily check that the space of fuzzy numbers is a semivector (nowadays called semilinear) space according to Prakash and Sertel definition, with respect to the addition and scalar multiplication of fuzzy numbers using their α-cuts. Moreover, we will prove that most of the elementary properties satisfied by some convex subsets of normed vector spaces remain valid in the case of the space of fuzzy numbers endowed with the Euclidean metric and we refer here to the results from Sections and 4. These properties will allow us to obtain a characterization of Lipschitz functions (Theorem 9 in Section 4) which is very suitable with the form of the trapezoidal approximation operators
4 118 L. Coroianu / Fuzzy Sets and Systems 2 (212) given on cases as we have mentioned before. From this characterization, as a direct application we will obtain the best Lipschitz constant of the trapezoidal approximation preserving the value and ambiguity (Theorem 1 in Section 5). Theorem 9 is quite general in the sense that it can be applied for most of the approximation operators known in the literature since most of them fulfil the requirements of Theorem 9. Finally, in Section 6 we propose two applications. Firstly, we will find within a reasonable error the trapezoidal approximation of a fuzzy number in the case when the direct formula is not applicable. Then we find an estimation for the defect of additivity of the discussed operator which is an improvement of the result proposed in the paper [1]. 2. Preliminaries We consider the following well-known description of a fuzzy number A: if x a 1, l A (x) if a 1 x a 2, A(x) = 1 if a 2 x a, r A (x) if a x a 4, if a 4 x, (1) where a 1, a 2, a, a 4 R, l A :[a 1, a 2 ] [, 1] is a nondecreasing upper semicontinuous function, l A (a 1 ) =, l A (a 2 ) = 1, called the left side of the fuzzy number and r A :[a, a 4 ] [, 1] is a nonincreasing upper semicontinuous function, r A (a ) = 1, r A (a 4 ) =, called the right side of the fuzzy number. The α-cut, α (, 1], of a fuzzy number A is a crisp set defined as A α ={x R : A(x) α}. The support or -cut A of a fuzzy number is defined as A = {x R : A(x) > }. Every α-cut α [, 1], of a fuzzy number is a closed interval where A α = [A L (α), A U (α)], A L (α) = inf{x R : A(x) α}, A U (α) = sup{x R : A(x) α} for any α (, 1]. If the sides of the fuzzy number A are strictly monotone then one can see easily that A L and A U are inverse functions of l A and r A, respectively. We denote by F(R) the set of all fuzzy numbers. In this paper we say that the fuzzy numbers A and B are equal and we denote A=B if A L (α) = B L (α), A U (α) = B U (α) almost everywhere α [, 1]. Some important notions connected with the concept of fuzzy number are the ambiguity Amb(A) and the value Val(A) of a fuzzy number A, A α = [A L (α), A U (α)], α [, 1]. They are given by (see [18]) Amb(A) = Val(A) = α(a U (α) A L (α)) dα, α(a U (α) + A L (α)) dα. (2) ()
5 L. Coroianu / Fuzzy Sets and Systems 2 (212) In practical problems such as the solving of fuzzy equations, data analysis or the ranking of fuzzy numbers, it is extremely important to endow the space of fuzzy numbers with topological structures. The flexibility of the space of fuzzy numbers allows the construction of many types of metric structures over this space. For example regarding the space of fuzzy numbers as a space of functions, the classical Chebyshev type distance between functions can be defined (see e.g. [17, p. 51]). Perhaps one of the most important topological structures over the space of fuzzy numbers is the one obtained from the Hausdorff metric between compact subsets of R 2, introduced by Goetschel and Voxman in [19] (see also [17, Chapter 7] and the references given at the end of the chapter). Usually a metric is defined with a precise purpose. For instance, in the paper [27] different metrics between fuzzy numbers are introduced in order to rank fuzzy numbers. Then, in the papers [14,1] new metrics are proposed with important advantages in statistical problems. The present paper deals with the problem of the approximation of fuzzy numbers. It seems that the suitable metric for this type of problem is an extension of the Euclidean distance and it is defined by [2] d 2 (A, B) = (A L (α) B L (α)) 2 dα + (A U (α) B U (α)) 2 dα. (4) Generalizations of the above L 2 -type metric can be found in the papers [7,8] as they are usually called weighted L 2 -type metrics. In all that follows in this paper, we will deal only with the case of the above Euclidean metric d since the main application of the paper is in relation with this particular metric. Fuzzy numbers with simple membership functions are preferred in practice. The most used such fuzzy numbers are so-called trapezoidal fuzzy numbers given by if x t 1, x t 1 if t 1 x t 2, t 2 t 1 T (x) = 1 if t 2 x t, t 4 x if t x t 4, t 4 t if t 4 x. When t 2 = t we obtain a so-called triangular fuzzy number. We denote T = (t 1, t 2, t, t 4 ) a trapezoidal fuzzy number as above. It is easy to prove that and T L (α) = t 1 + (t 2 t 1 )α T U (α) = t 4 (t 4 t )α for every α [, 1]. If T = (t 1, t 2, t, t 4 )andt = (t 1, t 2, t, t 4 ) then applying formula (4) weget d 2 (T, T ) = 1 (t 1 t 1 )2 + 1 (t 2 t 2 )2 + 1 (t 1 t 1 )(t 2 t 2 ) + 1 (t t )2 + 1 (t 4 t 4 )2 + 1 (t t )(t 4 t 4 ). (5) We denote by F T (R) the set of all trapezoidal fuzzy numbers. Let A, B F(R), A α = [A L (α), A U (α)], B α = [B L (α), B U (α)], α [, 1] and λ R. We consider the sum A+B and the scalar multiplication λ A by (see e.g. [17, p. 4]) (A + B) α = A α + B α = [A L (α) + B L (α), A U (α) + B U (α)] and { [λal (α), λa U (α)] if λ, (λ A) α = λa α = [λa U (α), λa L (α)] if λ <,
6 12 L. Coroianu / Fuzzy Sets and Systems 2 (212) respectively, for every α [, 1]. In the case of the trapezoidal fuzzy numbers T = (t 1, t 2, t, t 4 )ands = (s 1, s 2, s, s 4 ) we obtain T + S = (t 1 + s 1, t 2 + s 2, t + s, t 4 + s 4 ). If (X 1, d 1 )and(x 2, d 2 ) are metric spaces then a function f : X 1 X 2 is called a Lipschitz function if there exists a real positive constant C such that d 2 ( f (x 1 ), f (x 2 )) Cd 1 (x 1, x 2 ) for all x 1, x 2 X 1. It is well-known that Lipschitz functions are continuous.. Convexity in the space of fuzzy numbers Usually, the concept of convex set is given in relation with a vector space structure. It is known that the addition and scalar multiplication of fuzzy numbers do not form a vector space. However, since these operations are closed (see also the discussion about semilinear spaces from the Introduction) and mostly because it will be of great help later in obtaining the main results of the paper, we need the notion of convex set in the space of fuzzy numbers. Of course, the definition is exactly the same as in the case of vector spaces and therefore we have the following. Definition 1. A nonempty set Ω F(R) is called a convex subset of F(R)ifforallA, B Ω and γ [, 1] we have ((1 γ)a + γb) Ω. Remark 2. If A, B, C F(R) and γ [, 1] then taking into account the definitions of addition and scalar multiplication of fuzzy numbers, we have C = (1 γ)a + γb if and only if C L (α) = (1 γ)a L (α) + γb L (α) and C U (α) = (1 γ)a U (α) + γb U (α)forallα [, 1]. If A, B F(R) then we introduce the following subsets of F(R): [A, B] ={(1 γ)a + γb : γ [, 1]}, [A, B) ={(1 γ)a + γb : γ [, 1)}, (A, B] ={(1 γ)a + γb : γ (, 1]}, (A, B) ={(1 γ)a + γb : γ (, 1)}. We observe that a nonempty set Ω F(R) is convex if and only if [A, B] Ω for all A, B Ω. The following results are familiar or easily reachable in the case of vector topological spaces. But since the space of fuzzy numbers is not a vector space, to avoid any ambiguity and also for the uniformity of the exposure, we prefer to give the proof of each result. However, at the end of the section we will propose another approach using an embedding theorem in normed spaces. We mention that in the case of topological results we always consider the topology generated over the space of fuzzy numbers by the Euclidean distance presented in the previous section. Lemma. Let A, B F(R) and let γ 1 γ 2 γ n 1. For i {1, 2,..., n} denote C i = (1 γ i )A + γ i B. Then we have [A, B] = [A, C 1 ] [C 1, C 2 ] [C n 1, C n ] [C n, B]. Proof. We give the proof only for the case n=1 because the proof of the general case is immediate by mathematical induction. So, let γ [, 1] and denote C = (1 γ)a + γb which by Remark 2 implies that C L (α) = (1 γ)a L (α) + γb L (α), (6) C U (α) = (1 γ)a U (α) + γb U (α), α [, 1]. (7) We have to prove that [A, B] = [A, C] [C, B]. Without any loss of generality we may suppose that γ (, 1) because otherwise the proof is trivial. Firstly, we prove that [A, C] [C, B] [A, B]. For this reason let us choose arbitrarily
7 L. Coroianu / Fuzzy Sets and Systems 2 (212) D [A, C] [C, B]. If D [A, C]thenletβ [, 1] be such that D = (1 β)a+βc.thensincec = (1 γ)a+γb, and noting that the scalar multiplications involve only positive reals, we get D = (1 βγ)a + βγb. Since βγ [, 1] it easily follows that D [A, B]. The proof of the case D [C, B] is similar and therefore we omit the details. Now, we prove the converse inclusion. So, let D [A, B]. Then there exists δ [, 1] such that D = (1 δ)a + δb. This implies D L (α) = (1 δ)a L (α) + δb L (α), α [, 1]. (8) Firstly, we suppose that δ γ. Combining relations (6)and(8) weget ( D L (α) = 1 δ ) A L (α) + δ γ γ C L(α), α [, 1]. In the same manner we get that ( D U (α) = 1 δ ) A U (α) + δ γ γ C U (α), α [, 1]. Since δ/γ 1 and taking into account Remark 2, we easily get that D = (1 δ/γ)a + (δ/γ)c which implies that D [A, C]. In the remaining case, that is δ γ by similar reasonings we will obtain that D [C, B]. This last assertion completes the proof. Remark 4. We have to notice that the above proved property can be done in the case of vectorial spaces with a slightly simpler computational involvement. However, it is important that in the space of fuzzy numbers this property holds too. This will be seen later in the paper. Lemma 5. If A, B, C are fuzzy numbers such that C [A, B] then d(a, B) = d(a, C) + d(c, B). Proof. Let γ [, 1] be such that C = (1 γ)a + γb. From relations (6)to(7)weget ( ) 1/2 d(a, C) + d(c, B) = ((A L (α) C L (α)) 2 + (A U (α) C U (α)) 2 ) dα ( ) 1/2 + ((C L (α) B L (α)) 2 + (C U (α) B U (α)) 2 ) dα ( ) 1/2 = (γ 2 (A L (α) B L (α)) 2 + γ 2 (A U (α) B U (α)) 2 ) dα ( ) 1/2 + ((1 γ) 2 (A L (α) B L (α)) 2 + (1 γ) 2 (A U (α) B U (α)) 2 ) dα = γd(a, B) + (1 γ)d(a, B) = d(a, B). and the proof is complete. Lemma 6. If A, B F(R) then [A,B] is a closed set. Proof. Let (C n ) n 1 be a convergent sequence of fuzzy number with lim n C n = C and such that for all n 1we have C n [A, B]. It follows that there exists a sequence (γ n ) n 1 such that for all n 1wehaveγ n [, 1] and C n = (1 γ n )A + γ n B.WehavetoprovethatC [A, B]. Since (γ n ) n 1 is bounded it follows that there exists γ R such that γ = sup n 1 γ n. Let C = (1 γ)a + γb. Since it is easy to check that γ [, 1], it follows that
8 122 L. Coroianu / Fuzzy Sets and Systems 2 (212) C [A, B]. From the definition of γ it results the existence of a subsequence (γ kn ) n 1 of the sequence (γ n ) n 1 such that lim n γ kn = γ. Wehave d 2 (C kn, C ) = ((C kn ) L (α) C L (α))2 + ((C kn ) U (α) C U (α))2 ) dα, which by elementary calculus gives d 2 (C kn, C ) = (γ γ kn ) 2 (A L (α) B L (α)) 2 dα + (γ γ kn ) 2 (A U (α) B U (α)) 2 dα = (γ γ kn ) 2 d 2 (A, B). Since lim n γ kn = γ we immediately obtain that lim n C kn = C. On the other hand, since (C kn ) n 1 is a subsequence of the sequence (C n ) n 1 and, since lim n C n = C, it necessarily follows that lim n C kn = C. From the uniqueness of the limit we obtain C = C. Just above we have proved that C [A, B] and this implies that C [A, B]. In conclusion, we have just proved that any convergent sequence with elements in [A,B] has its limit in [A,B] and this means that the set [A,B] is a closed set. Lemma 7. Let A, B F(R), A B. Then cl((a, B)) = cl((a, B]) = cl([a, B)) = [A, B], where cl is the closure operator. Proof. First we prove that cl((a, B)) = [A, B]. Let us consider the sequences (C n ) n 1, (D n ) n 1, C n = 1 ( n A ) B, n ( D n = 1 1 ) A + 1 n n B. The definition of (A,B) implies that for all n 1wehaveC n (A, B) andd n (A, B). On the other hand it is immediate that lim n C n = B and that lim n D n = A. This implies that {A, B} cl(a, B). Now, since [A, B] = (A, B) {A, B} it necessarily follows that [A, B] cl((a, B)). But since (A, B) [A, B], it also results that cl((a, B)) cl([a, B]). The previous lemma implies that [A, B] = cl([a, B]). Consequently, we obtain cl((a, B)) [A, B] which combined with the converse inclusion gives cl((a, B)) = [A, B]. The equality cl((a, B]) = [A, B] results from the following inclusions: [A, B] = cl((a, B)) cl((a, B]) cl([a, B]) = [A, B]. By similar reasonings we will obtain that cl([a, B)) = [A, B]. In what follows, we propose another approach which leads to the same conclusions of this section, based on an embedding theorem of Rådström. It can be easily proved that the quadruple (F(R), +,, d) satisfies all the requirements of Theorem 1 in []. Applying the conclusion of Rådström s theorem if follows that (F(R), +,, d) can be embedded in a normed space ( F(R),,, D). This means that A B = A + B, λ A = λ A and D(A, B) = d(a, B) for all A, B F(R) andλ. This implies that all the convex or closed subsets of F(R) can be perceived as closed or convex subsets of the normed space F(R). More recently, Yeh proved (see [,6]) that(f(r), +,, d λ )whered λ is a weighted L 2 -type metric which generalizes the Euclidean distance d (see e.g. [7]) can be embedded in a Hilbert space. In conclusion, using either Rådström s theorem or the construction of Yeh, all the results of this section can be obtained from corresponding results from normed or even Hilbert spaces. 4. A characterization of fuzzy number-valued Lipschitz functions First we need the following.
9 L. Coroianu / Fuzzy Sets and Systems 2 (212) Lemma 8. Let A, B F(R), A B. Furthermore, we consider the family of closed convex subsets of F(R), F ={Ω i : i {1, 2,..., n}} such that [A, B] n Ω i. i=1 Then, there exist k {1, 2,..., n}, {C j : j {, 1,..., k}} [A, B], with C = A and C k = B respectively, and {Ω l j : j {1, 2,..., k}} F, such that: (i) ) [A, B] = k j=1 [C j 1, C j ]; (ii) d(a, B) = k j=1 d(c j 1, C j ); (iii) [C j 1, C j ] Ω l j for all j {1, 2,..., k}. Proof. We prove the lemma by mathematical induction on the number of elements of the family F denoted in what follows with card(f). If card(f) = 1 then for k=1, C = A, C 1 = B, it is immediate that the conclusions of the lemma hold. Now, suppose that card(f) = n, n 2 and suppose that the conclusions of the lemma hold in the case of families with n 1 elements. Since [A, B] n i=1 Ω i, it follows that there exists i 1 {1, 2,..., n} such that A Ω i1. If B Ω i1 then owing to the convexity of Ω i1, it follows that [A, B] Ω i1 which means that the conclusions of the lemma hold by choosing k=1, C = A, C 1 = B, andω l1 = Ω i1. For this reason, in all what follows we will suppose that B / Ω i1. Let and let γ = inf{γ [, 1] : (γa + (1 γ)b) [A, B] Ω i1 } (1) (9) C = γ A + (1 γ )B. (11) Clearly, we have C [A, B]. Moreover, we will prove that C Ω i1 too. From the definition of γ results the existence of a sequence (γ n ) n 1 such γ n 1, lim n γ n = γ and (γ n A + (1 γ n )B) [A, B] Ω i1 for all n 1. For each n 1, denoting D n = γ n A + (1 γ n )B we obtain d 2 (D n, C) = (γ γ n ) 2 (A L (α) B L (α)) 2 dα + (γ γ n ) 2 (A U (α) B U (α)) 2 dα = (γ γ n ) 2 d 2 (A, B), which immediately implies that lim n D n = C. Since for all n 1wehaveD n [A, B] Ω i1 and since [A, B] Ω i1 is a closed subset of F(R) as an intersection of closed subsets from F(R), it follows that C [A, B] Ω i1.now,since {A, C} [A, B] Ω i1 and since [A, B] Ω i1 is a convex subset of F(R) as an intersection of convex subsets of F(R), we conclude that [A, C] [A, B] Ω i1. Consequently, by Lemma (case n=1) we get that [A, B] = [A, C] [C, B] and in addition by Lemma 5 we have d(a, B) = d(a, C) + d(c, B). (12) If C=B then the conclusions of the lemma hold by choosing k=1, C = A, C 1 = B, andω l1 = Ω i1. Therefore, in all that follows we may suppose that C B. This fact combined with relation (11) implies that γ >, information that will be very useful later. Next we prove that (C, B] ([A, B] Ω i1 ) =. By way of contradiction suppose that there exists D (C, B] ([A, B] Ω i1 ). Then let β [, 1] be such that D = βc + (1 β)b. (1) If β = thend = B which implies B [A, B] Ω i1,thatisb Ω i1, a contradiction since we have assumed that B / Ω i1.ifβ = 1 then we easily get that D = C and this is a contradiction since D (C, B]. Summarizing, we necessarily have that β (, 1). Combining relations (11)and(1) we obtain that (again, it is important that the scalar multiplication operations involve only positive reals) D = βγ A + (1 βγ )B.
10 124 L. Coroianu / Fuzzy Sets and Systems 2 (212) On the other hand, since γ > andβ (, 1), it follows that βγ < γ and this contradicts with the definition of γ (see relation (1)) since we have assumed that D [A, B] Ω i1. Thus, we obtain that (C, B] ([A, B] Ω i1 ) =. Taking into account relation (9), it follows that (C, B] i I Ω i where I ={1, 2,..., n} \{i 1 }. This implies that cl((c, B]) cl( i I Ω i) which by Lemma 7 and by the fact that i I Ω i is a closed set as finite union of some closed sets, implies that [C, B]) i I Ω i. Considering the family of n 1elementsF 1 ={Ω i : i {1, 2,..., n}\{i 1 }}, according to our assumption it follows that there exist k {1, 2,..., n 1}, {C j : j {, 1,..., k} [C, B], with C = C and C k = B respectively, and {Ω p j ; j {1, 2,..., k}} F 1, such that: (i) [C, B] = k j=1 [C j 1, C j ]; (ii) d(c, B) = k j=1 d(c j 1, C j ); (iii) [C j 1, C j ] Ω p j for all j {1, 2,..., k}. Take C = A, C j+1 = C j for j {, 1,..., k}, Ω l 1 = Ω i1, Ω l j+1 = Ω p j for j {1, 2,..., k}. From the properties of C j, j {, 1,..., k + 1} and Ω l j, j {1, 2,..., k + 1}, it is immediate that assertions (i) and (iii) of the present lemma hold. Then, since d(a, B) = d(a, C) + d(c, B) and since d(c, B) = k j=1 d(c j 1, C j ), it is immediate that assertion (ii) of the lemma holds too. This completes the proof. From the previous result we obtain the following main result of the paper. Theorem 9. Let F ={Ω i : i {1, 2,..., n}} be a family of closed convex subsets of F(R) such that F(R) = n Ω i. i=1 Furthermore, let f : F(R) F(R) denote a function with the property that there exist the positive real constants, c i, i {1, 2,..., n} such that for all i {1, 2,..., n} and A, B Ω i we have (14) d( f (A), f (B)) c i d(a, B). (15) Then d( f (A), f (B)) cd(a, B), A, B F(R), where c = max{c i : i {1, 2,..., n}}. Proof. Let us choose arbitrarily A, B F(R), A B (the case A=B is trivial). Relation (14) implies [A, B] n i=1 Ω i. Applying Lemma 8, it follows that there exist k {1, 2,..., n}, {C j : j {, 1,..., k}} [A, B], with C = A and C k = B respectively, and {Ω l j : j {1, 2,..., k}} F, such that assertions (i) (iii) of Lemma 8 hold. First of all we have d( f (A), f (B)) k d( f (C j 1 ), f (C j )). (16) j=1 Then, from assertion (iii) of Lemma 8, we obtain {C j 1, C j } Ω l j, j {1, 2,..., k} which by relation (15) implies d( f (C j 1 ), f (C j )) c l j d(c j 1, C j ). Since for each i {1, 2,..., n} we have c i c, it follows that for each j {1, 2,..., k} we have d( f (C j 1 ), f (C j )) cd(c j 1, C j ). Replacing in inequality (16) we obtain d( f (A), f (B)) c k d(c j 1, C j ) j=1 and since assertion (ii) of Lemma 8 implies d(a, B) = k j=1 d(c j 1, C j ), we get that d( f (A), f (B)) cd(a, B)and we obtain the desired conclusion.
11 L. Coroianu / Fuzzy Sets and Systems 2 (212) Remark 1. The above proved theorem says that if the space of fuzzy numbers is covered by a finite family of closed convex subsets and a function f : F(R) F(R) is Lipschitz relatively to each subset of the family then it is Lipschitz on the whole space of fuzzy numbers. This result will be very useful in finding the best Lipschitz constant of the trapezoidal approximation operator preserving the value and the ambiguity as it will be seen in the next section. We end this section by mentioning that we do not exclude the possibility of obtaining the proofs of Corollary 8 and Theorem 9 respectively, from corresponding results in normed spaces, using again the embedding theorem of Rådström. 5. Best Lipschitz constant of the trapezoidal approximation operator preserving the value and ambiguity The algorithms to compute the trapezoidal approximation (with respect to the Euclidean distance) of a fuzzy number which preserves the value and the ambiguity are given in Theorem 7 and Corollary 8 in the paper [1]. These results are obtained as solutions of minimization problems in the Euclidean vector space R 2. The so-called extended trapezoidal fuzzy numbers introduced by Yeh in the paper [4] are used. The trapezoidal approximation is given on cases and it is obtained by solving the minimization problem given in relations (25) () of the paper [1]. There are four possible algorithms to compute the proper trapezoidal approximation of a fuzzy number which preserves the value and ambiguity, each one corresponding to a nonempty subset of F(R) and these subsets form disjoint pairs. But a careful inside in the proof of Theorem 5 from page 184 (see also the geometrical interpretation of the minimization problem illustrated in Fig. 1 of the discussed paper) proves that the formulas for computing the trapezoidal approximation described in cases (i) (iv) just below the end of the proof of Theorem 5 remain valid if in all the inequalities from cases (i) (iv) we consider non-strict inequalities. Therefore, in Theorem 7 and Corollary 8 we can replace all the strict inequalities with non-strict inequalities. Taking into account this information we give a slightly modified version of Corollary 8 as follows. Theorem 11. Let A F(R), A α = [A L (α), A U (α)], α [, 1], and let T (A) = (t 1, t 2, t, t 4 ) denote the nearest trapezoidal fuzzy number to A which preserves the value and the ambiguity. (i) If then (α 1)A L (α) dα (α 1)A U (α) dα (17) t 1 = t 2 = t = t 4 = (4 6α)A L (α) dα, (6α 2)A L (α) dα, (6α 2)A U (α) dα, (4 6α)A U (α) dα. (18) (19) (2) (21) (ii) If (α 1)A L (α) dα + (α 1)A U (α) dα (22)
12 126 L. Coroianu / Fuzzy Sets and Systems 2 (212) then then and then t 1 = 6 t 2 = t = t 4 = 2 (iii) If αa L (α) dα 4 (α 1)A L (α) dα + t 1 = t 2 = t = 2 t 4 = 6 (iv) If αa U (α) dα. αa U (α) dα 4 (α 1)A L (α) dα (α 1)A L (α) dα + (α 1)A L (α) dα + t 1 = 2 t 2 = t = t 4 = A L (α) dα αa L (α) dα, αa U (α) dα, (2) (α 1)A U (α) dα (25) (α 1)A L (α) dα + (2 6α)A L (α) dα + 2 (24) (26) αa L (α) dα. (27) (α 1)A U (α) dα, (28) (α 1)A U (α) dα (29) (α 1)A U (α) dα () (6α 2)A U (α) dα, (1) (α 1)A U (α) dα, (2) A U (α) dα. () The following auxiliary result will help us in the main application of this section. Lemma 12. Let (A n ) n N,(A n ) α = [(A n ) L (α), (A n ) U (α)], α [, 1] be a sequence in the space F(R) such that lim n A n = A, where Then A α = [A L (α), A U (α)], α [, 1]. lim n α k (A n ) L (α) dα = α k A L (α) dα
13 and lim n for all k N. α k (A n ) U (α) dα = L. Coroianu / Fuzzy Sets and Systems 2 (212) α k A U (α) dα Proof. Let ε >. Because lim n A n = A, there exists n ε N such that d 2 (A n, A) = for every n n ε.weget (((A n ) L (α) A L (α)) 2 + ((A n ) U (α) A U (α)) 2 ) dα < ε 2 ((A n ) L (α) A L (α)) 2 dα < ε 2 for every n n ε and ((A n ) U (α) A U (α)) 2 dα < ε 2 for every n n ε. Let k N.Then α k (A n ) L (α) dα α k 1 A L (α) dα α k ((A n ) L (α) A L (α)) dα α 2k ((A n ) L (α) A L (α)) 2 dα for every n n ε,thatis ((A n ) L (α) A L (α)) 2 dα < ε lim n α k (A n ) L (α) dα = α k A L (α) dα. The proof is similar in the case of the second limit and therefore we omit the details. We present now the main result of this section. Theorem 1. If T : F(R) F T (R) is the trapezoidal approximation operator which preserves the value and the ambiguity given in Theorem 11, then d(t (A), T (B)) d(a, B) for all A, B F(R) and the value ( )/ is the best possible Lipschitz constant of the operator T. Proof. Let us introduce the sets Ω 1 ={A F(R) :A satisfies condition (17)}, Ω 2 ={A F(R) :A satisfies condition (22)}, Ω ={A F(R) :A satisfies condition (25)}, Ω 4 ={A F(R) :A satisfies conditions (28).()}. It is elementary to verify the convexity of the sets Ω i, i {1, 2,, 4}. Then from Lemma 12 it follows that each one of the sets Ω i, i {1, 2,, 4}, is a closed set. We exemplify with the set Ω 1 because the proof in the remaining
14 128 L. Coroianu / Fuzzy Sets and Systems 2 (212) cases is similar. Let (A n ) n 1,(A n ) α = [(A n ) L (α), (A n ) U (α)], α [, 1] be a sequence in the space F(R) such that lim n A n = A and such that for all n 1wehaveA n Ω 1. As a first implication we have (α 1)(A n ) L (α) dα (α 1)(A n ) U (α) dα. Passing to limit with n and taking into account Lemma 12, it follows that (α 1)A L (α) dα (α 1)A U (α) dα, that is A Ω 1 and clearly this implies that Ω 1 is a closed subset of F(R). As a first conclusion it follows that the space F(R) is covered by the family of closed convex subsets of F(R), F ={Ω i : i {1, 2,, 4}}. Weintendto apply Theorem 9 with respect to the family F and therefore, in what follows we will determine the Lipschitz constants of the operator T relatively to the sets of the family F. Thus, for A, B F(R) with T (A) = (t 1, t 2, t, t 4 )and T (B) = (t 1, t 2, t, t 4 ), we distinguish the following four cases: (i) A, B Ω 1. It is immediate that the case (i) in Theorem 11 is applicable to compute T(A) andt(b). We observe that formulas (18) (21) coincide with the formulas for computing the nearest extended trapezoidal approximation of Grzegorzewski and Mrówka (see e.g. [4], the formulas from the first four lines from the top on page 1 or Theorem in [7]). Therefore, in this case we have T (A) = T e (A)andT(B) = T e (B)whereT e (A)andT e (B) denote the extended trapezoidal approximations of A and B respectively. Since by Proposition 4.4 in [2] we have d(t e (A), T e (B)) d(a, B), we obtain in this case the relation d(t (A), T (B)) d(a, B). (4) (ii) A, B Ω 2. It is immediate that the case (ii) in Theorem 11 is applicable to compute T(A) andt(b). Noting formula (5) and since by relations (2) (24)wehavet 2 = t = t 4 and t 2 = t = t 4, we obtain d 2 (T (A), T (B)) = 1 (t 1 t 1 )2 + 4 (t 2 t 2 )2 + 1 (t 1 t 1 )(t 2 t 2 ), which again by formulas (2) (24), after some simple calculations give ( 2 ( d 2 (T (A), T (B)) = 12 α(a L (α) B L (α)) dα) + 8 α(a U (α) B U (α)) dα ( )( ) 12 α(a L (α) B L (α)) dα α(a U (α) B U (α)) dα. Now, applying the Schwarz inequality we get d 2 (T (A), T (B)) 12 (A L (α) B L (α)) 2 dα α 2 dα + 8 (A U (α) B U (α)) 2 dα +12 (A L (α) B L (α)) 2 dα α 2 dα (A U (α) B U (α)) 2 dα = 4 (A L (α) B L (α)) 2 dα + 8 (A U (α) B U (α)) 2 dα ) 2 α 2 dα α 2 dα +4 (A L (α) B L (α)) 2 dα (A U (α) B U (α)) 2 dα. (5) Let m be such that (A U (α) B U (α)) 2 dα = m (A L (α) B L (α)) 2 dα. (6)
15 L. Coroianu / Fuzzy Sets and Systems 2 (212) Using the above equality in relation (5)weget ( d 2 (T (A), T (B)) 4 + 8m ) + 4 m (A L (α) B L (α)) 2 dα Noting that (m + 1) it follows that = 4 + 8m + 4 m m + 1 (A L (α) B L (α)) 2 dα = m = (m + 1) (A L (α) B L (α)) 2 dα. (A L (α) B L (α)) 2 dα + (A U (α) B U (α)) 2 dα + (A L (α) B L (α)) 2 dα (A L (α) B L (α)) 2 dα = d 2 (A, B), d 2 (T (A), T (B)) 4 + 8m + 4 m d 2 (A, B). (7) m + 1 Let us introduce the function f :[, ) R, f (m) = 4 + 8m + 4 m. m + 1 We have f (m) = ( 6m 4 m + 6)/( m(m + 1) 2 ). It is immediate that f ((( 1 + 1)/) 2 ) =. Then it is easy to check that f is strictly increasing on [, (( 1 + 1)/) 2 ] and strictly decreasing on ((( 1 + 1)/) 2, ). This implies that m = (( 1 + 1)/) 2 is a global maximum point of the function f. Since by simple calculations we get f ((( 1 + 1)/) 2 ) = ( )/, introducing the maximum value in relation (7)weget d(t (A), T (B)) d(a, B). (8) (iii) A, B Ω. To avoid laborious calculus we propose a different approach. We easily observe that A and B both belong to the set Ω 2. Then, having in mind the scalar multiplication of fuzzy numbers presented in the first section, one can easily prove that T ( A) = T (A)andT ( B) = T (B). This is not surprising since it can easily be proved that the operator T is scale invariant. On the other hand, noting that in general for two arbitrarily chosen fuzzy numbers C and D we have d(c, D) = d( C, D), it follows that d(t (A), T (B)) = d( T (A), T (B)) = d(t ( A), T ( B)) and since A and B both belong to the set Ω 2, by relation (8) we obtain d(t ( A), T ( B)) d( A, B) = d(a, B) and finally, we have d(t (A), T (B)) d(a, B). (9) (iv) A, B Ω 4. It is immediate that case (iv) in Theorem 11 is applicable to compute T(A)andT(B). Noting formula (5) and since by relations (1) ()wehavet 2 = t and t 2 = t, we obtain d 2 (T (A), T (B)) = 1 (t 1 t 1 )2 + 2 (t 2 t 2 )2 + 1 (t 4 t 4 )2 + 1 (t 2 t 2 )[(t 1 t 1 ) + (t 4 t 4 )]. (4)
16 1 L. Coroianu / Fuzzy Sets and Systems 2 (212) We introduce the following notations: a 1 = a 2 = a = a 4 = (A L (α) B L (α)) dα, (A U (α) B U (α)) dα, Relations (1) ()imply (α 1)(A L (α) B L (α)) dα, (α 1)(A U (α) B U (α)) dα. t 1 t 1 = 2a 1 2a 4, t 2 t 2 = a + a 4, t 4 t 4 = 2a 2 2a. Using these equalities in relation (4) we obtain d 2 (T (A), T (B)) = 4 (a 1 a 4 ) (a + a 4 ) (a 2 a ) (a + a 4 )(a 1 a 4 + a 2 a ), which by simple calculations give d 2 (T (A), T (B)) = 4 (a2 1 + a2 2 + a2 + a2 4 ) + 2 (a 1a + a 2 a 4 ) 2(a 1 a 4 + a 2 a ). Now, using the obvious inequalities a 1 a + a 2 a (a2 1 + a2 2 + a2 + a2 4 )and a 1a 4 + a 2 a 1 2 (a2 1 + a2 2 + a2 + a2 4 ), we easily obtain d 2 (T (A), T (B)) 8 (a2 1 + a2 2 + a2 + a2 4 ). Then, replacing a i, i {1, 2,, 4}, with their initial values and then using the Schwarz inequality, we get [ ( d 2 (T (A), T (B)) ( ) 2 ] (A L (α) B L (α)) dα) + (A U (α) B U (α)) dα + 8 ( ) 2 (α 1)(A L (α) B L (α)) dα + 8 ( ) 2 (α 1)(A U (α) B U (α)) dα 8 [ ] (A L (α) B L (α)) 2 dα + (A U (α) B U (α)) 2 dα + 8 (α 1) 2 dα (A L (α) B L (α)) 2 dα + 8 (α 1) 2 dα (A U (α) B U (α)) 2 dα = 16 [ (A L (α) B L (α)) 2 dα + Summarizing, in this last case we obtain ] (A U (α) B U (α)) 2 dα = 16 d2 (A, B). (41) d(t (A), T (B)) 4 d(a, B). (42)
17 L. Coroianu / Fuzzy Sets and Systems 2 (212) Now we return to the purpose of this theorem, that is to find the best Lipschitz constant of the operator T. Analyzing cases (i) (iv) we conclude that thelipschitz constants corresponding to the elements of the family F ={Ω i : i {1, 2,, 4}} are c 1 = 1, c 2 = c = ( )/ andc 4 = 4 (see relations (4), (8), (9) and(42)). Applying Theorem 9 and noting that max{c i : i {1, 2,, 4}} = ( )/, it follows that for all A, B F(R)wehave d(t (A), T (B)) d(a, B). To prove that the value ( )/ is the best Lipschitz constant of the operator T we need to find A, B F(R), A B, such that d(t (A ), T (B )) = d(a, B ). (4) Looking over relation (5), in order to obtain equality instead of inequality, we need to find A, B Ω 2 with B of the form (B ) L (α) = (A ) L (α) + k 1 α, (B ) U (α) = (A ) U (α) k 2 α, α [, 1], where k 1 >, k 2 >. Actually, it is not necessary to impose the conditions of strict positivity for k 1 and k 2 but in this way, provided that (A ) L (1) + k 1 (A ) U (1) k 2, we get that B is a proper fuzzy number. Moreover, it is elementary to check that if A Ω 2 then B Ω 2 too when k 1 > andk 2 >. Then, A and B have to satisfy relation (6)for m = (( 1 1)/) 2, which immediately implies that ( 1 ) 1 k 1 k 2 =. Inspired by the above considerations we introduce the fuzzy numbers A, B, (A ) L (α) = 9 α + 1, (A ) U (α) = 96, 1 1 (B ) L (α) = (A ) L (α) + α, (B ) U (α) = (A ) U (α) α, α [, 1]. It is immediate that A, B Ω 2 and therefore applying formulas (2) (24) we obtain T (A ) = (27, 96, 96, 96) and ( T (B ) = ( 1 1) 9, 96 2( 1 1) 9 ), 96 2( 1 1), 96 2( 1 1). 9 9 Now, it is easy to verify that relation (4) holds. This last assertion ends the proof. From the previous theorem it results that the value ( )/ is the best Lipschitz constant of the operator T applied for the entire set of fuzzy numbers. Still, since the operator T is given on four cases, it would be of some interest to determine the best Lipschitz constant for each case in part. That is to find the best Lipschitz constant for each one of the sets Ω i, i {1, 2,, 4}. Clearly, from the example given at the end of the proof of the previous theorem it follows that the value ( )/ is the best Lipschitz constant applied for the set Ω 2. Then, for A and B from the discussed example we observe that A and B both belong to Ω. Noting again that the operator T is scale invariant we easily obtain that (4) holds if we replace A and B with A and B respectively. This means
18 12 L. Coroianu / Fuzzy Sets and Systems 2 (212) that ( )/ is the best Lipschitz constant applied for the set Ω. Moreover, if A and B are trapezoidal fuzzy numbers then it is immediate that (17) holds and therefore we have A, B Ω 1. In addition, since the operator T fulfils the identity criteria it follows that T (A) = A and T (B) = B. This fact combined with inequality (4) proves that the value 1 is the best Lipschitz constant applied for the set Ω 1. Unfortunately, in the case of the set Ω 4 we cannot find A, B Ω 4, A B such that relation (42) would become equality. By way of contradiction let us suppose that this possibility exits. Because in relation (41) we used the inequalities ( 2 (A L (α) B L (α)) dα) (A L (α) B L (α)) 2 dα, ( 2 (A U (α) B U (α)) dα) ( 2 (α 1)(A L (α) B L (α)) dα) ( 2 (α 1)(A U (α) B U (α)) dα) (A U (α) B U (α)) 2 dα, (α 1) 2 dα (α 1) 2 dα (A L (α) B L (α)) 2 dα, (A U (α) B U (α)) 2 dα, it follows that all these inequalities become equalities and therefore, in what follows we consider them as equalities. From the first and the third equality it results the existence of the real constants k 1 and k 2 such that and A L (α) B L (α) = k 1 A L (α) B L (α) = k 2 (α 1) almost everywhere α [, 1]. Clearly, the above two equalities are possible only when k 1 = k 2 = and so we get that A L (α) = B L (α) almost everywhere α [, 1]. Analogously, from the second and the fourth equality we get that A U (α) = B U (α) almost everywhere α [, 1] which combined with the previous relation implies that A=B (see the discussion about the equality of two fuzzy numbers given in Section 2) and this clearly contradicts our assumption. In conclusion, the best Lipschitz constant applied for the set Ω 4 remains an open question. 6. Some applications In this last section we will apply the estimation obtained in Theorem 1 to calculate the trapezoidal approximation of a fuzzy number preserving the value and ambiguity within a reasonable error in the case when the direct algorithm is not applicable. We will consider the same example as in the case of the trapezoidal approximation operator preserving the expected interval (see [16, Example ]). Then we will improve the estimation on the defect of additivity of the trapezoidal approximation operator preserving the value and the ambiguity, estimation which was obtained in the paper [1]. Example 14. Let us consider the fuzzy number A, A L (α) = e α2, A U (α) = 4 α, α [, 1]. We will determine T(A) with an error less than 1 2 with respect to the Euclidean metric d. For this purpose let us consider the sequence of fuzzy numbers (A n ) n 1,(A n ) L (α) = 1 + α 2 + α 4 /2! + α 6 /! + +α 2n /n!, A U (α) = 4 α, α [, 1]. Following the same root as in the case of the trapezoidal approximation operator preserving the expected interval we get that d 2 e 2 (A n, A) ((n + 1)!) 2, n 1. (4n + 5) Applying Theorem 1 we obtain d 2 (T (A n ), T (A)) e 2 ((n + 1)!) 2 (4n + 5).
19 L. Coroianu / Fuzzy Sets and Systems 2 (212) Obviously, for n 5wehave d(t (A n ), T (A)) 1 2. For n=5 case (i) in Theorem 11 is applicable to compute the trapezoidal approximation of A 5 preserving the value and the ambiguity, therefore T (A 5 ) = (.69595, ,, 4) and we thus obtained an approximation of T(A) with an error less than 1 2. Note that in the case of the trapezoidal approximation preserving the expected interval, since the best Lipschitz 5 constant is the value, we obtained the desired approximation for n=4. It is known by Example 24 in the paper [1] that the trapezoidal approximation operator preserving the value and the ambiguity is not additive. Following the ideas in [6] the notion of defect of additivity of a trapezoidal approximation operator was introduced in the paper [1]. Definition 15 (Ban et al. [1, Definition 25]). Let A F(R) andt : F(R) F T (R) be a trapezoidal approximation operator with respect to a distance denoted with D. The defect of additivity of the operator t with respect to fuzzy number A is given by δ t,d (A) = sup D(t(A) + t(b), t(a + B)). B F(R) To obtain an estimation for δ T,d (A) in the case of the trapezoidal approximation preserving the value and ambiguity we need the following. Lemma 16. Let t : F(R) F T (R) be a trapezoidal approximation operator with respect to a distance denoted with D. Suppose that the following requirements hold: (i) t(o) = O where O is the trapezoidal fuzzy number (,,,); (ii) there exists a positive real constant c such that D(t(A), t(b)) cd(a, B), A, B F(R); (iii) we have D(A + C, B + C) = D(A, B) for all A, B, C F(R). Then δ t,d (A) 2cD(O, A) for all A F(R). Proof. Let us choose arbitrarily a fuzzy number A. Then, for B F(R) wehave D(t(A) + t(b), t(a + B)) D(t(A) + t(b), t(b)) + D(t(B), t(a + B)) D(t(A), O) + cd(b, A + B) = D(t(A), t(o)) + cd(o, A) cd(o, A) + cd(o, A) = 2cD(O, A) and the proof is complete.
20 14 L. Coroianu / Fuzzy Sets and Systems 2 (212) Corollary 17. Let A be a fuzzy number. If T denotes the trapezoidal approximation operator preserving the value and the ambiguity, then ( 1 1 ) 1/2 δ T,d (A) 2 (A 2 L (α) + A2 U (α)) dα. Proof. Firstly, we notice that the operator T and the Euclidean distance d fulfil the requirements of Lemma 16.Nowthe proof is immediatefromthe above lemma, replacingc with the best Lipschitz constant of the trapezoidal approximation operator preserving the value and ambiguity, that is c = ( )/, and noting that ( ) 1/2 d(o, A) = (A 2 L (α) + A2 U (α)) dα. 7. Conclusions In this paper we have proved that many important properties concerning convex subsets in normed spaces remain valid in the space of fuzzy numbers. These properties were used to obtain a characterization of Lipschitz fuzzy valued functions (Theorem 9), which is useful in proving that a trapezoidal approximation operator satisfies the Lipschitz condition and in the case of the trapezoidal approximation preserving the value and ambiguity even the best Lipschitz constant was found. The method is suitable to be applied for most of the triangular trapezoidal or parametric approximation operators that can be found in the literature since these operators fulfil the requirements of Theorem 9. Future work will be dedicated to find better estimates for the defect of additivity of fuzzy approximation operators. We have already started a research in the case of the trapezoidal approximation operator preserving the expected interval and the results are encouraging. Acknowledgments This work was possible with the financial support of the Sectoral Operational Programme for Human Resources Development 27 21, co-financed by the European Social Fund, under the project number POSDRU/17/1.5/S/ with the title Modern Doctoral Studies: Internationalization and Interdisciplinarity and, with the support of a grant of the Romanian National Authority for Scientific Research, CNCS-UEFISCDI, project number PN-II-ID-PCE References [1] S. Abbasbandy, M. Amirfakhrian, The nearest trapezoidal form of a generalized left right fuzzy number, Int. J. Approx. Reason. 4 (26) [2] S. Abbasbandy, M. Amirfakhrian, The nearest approximation of a fuzzy quantity in parametric form, Appl. Math. Comput. (New York) 172 (26) [] S. Abbasbandy, B. Asady, The nearest trapezoidal fuzzy number to a fuzzy quantity, Appl. Math. Comput. (New York) 156 (24) [4] S. Abbasbandy, T. Hajjari, Weighted trapezoidal approximation-preserving cores of a fuzzy number, Comput. Math. Appl. 59 (21) [5] T. Allahviranloo, M. Adabitabar Firozja, Note on Trapezoidal approximation of fuzzy numbers, Fuzzy Sets Syst. 158 (27) [6] A.I. Ban, S.G. Gal, Defects of Properties in Mathematics. Quantitative Characterizations, World Scientific, New Jersey, 22. [7] A.I. Ban, Approximation of fuzzy numbers by trapezoidal fuzzy numbers preserving the expected interval, Fuzzy Sets Syst. 159 (28) [8] A.I. Ban, On the nearest parametric approximation of a fuzzy number-revisited, Fuzzy Sets Syst. 16 (29) [9] A.I. Ban, Trapezoidal and triangular approximations of fuzzy numbers inadvertences and corrections, Fuzzy Sets Syst. 16 (29) [1] A.I. Ban, A. Brândaş, L. Coroianu, C. Negruţiu, O. Nica, Approximations of fuzzy numbers by trapezoidal fuzzy numbers preserving the value and ambiguity, Comput. Math. Appl. 61 (211) [11] A.I. Ban, L. Coroianu, Continuity and linearity of the trapezoidal approximation preserving the expected interval operator, in: IFSA-EUSFLAT World Congress, 2 24 July 29, pp
Nearest piecewise linear approximation of fuzzy numbers
Nearest piecewise linear approximation of fuzzy numbers Lucian Coroianu 1, Marek Gagolewski,3, Przemyslaw Grzegorzewski,3, Abstract The problem of the nearest approximation of fuzzy numbers by piecewise
More informationApproximations by interval, triangular and trapezoidal fuzzy numbers
Approximations by interval, triangular and trapezoidal fuzzy numbers Chi-Tsuen Yeh Department of Mathematics Education, National University of Tainan 33, Sec., Shu-Lin St., Tainan city 75, Taiwan Email:
More informationInt. J. Industrial Mathematics (ISSN ) Vol. 5, No. 1, 2013 Article ID IJIM-00188, 5 pages Research Article. M. Adabitabarfirozja, Z.
Available online at http://ijim.srbiau.ac.ir/ Int. J. Industrial Mathematics (ISSN 28-562) Vol. 5, No., 23 Article ID IJIM-88, 5 pages Research Article Triangular approximations of fuzzy number with value
More informationSimplifying the Search for Effective Ranking of Fuzzy Numbers
1.119/TFUZZ.214.231224, IEEE Transactions on Fuzzy Systems IEEE TRANSACTIONS ON FUZZY SYSTEMS, VOL., NO., 1 Simplifying the Search for Effective Ranking of Fuzzy Numbers Adrian I. Ban, Lucian Coroianu
More informationFuzzy Mathematics, Approximation Theory, Optimization, Linear and Quadratic Programming, Functional Analysis.
1. Personal information Name and surname: Lucian COROIANU Date and place of birth: January 26, 1976, Oradea, Romania Marital status: Married, 2 children, 7 and 9 years old Present academic position: Assistant
More informationThis article was published in an Elsevier journal. The attached copy is furnished to the author for non-commercial research and education use, including for instruction at the author s institution, sharing
More informationThis article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution
More informationTRAPEZOIDAL APPROXIMATION OF FUZZY NUMBERS
BABES -BOLYAI UNIVERSITY, CLUJ-NAPOCA FACULTY OF MATHEMATICS AND COMPUTER SCIENCE TRAPEZOIDAL APPROXIMATION OF FUZZY NUMBERS Ph.D. Thesis Summary Scienti c Supervisor: Professor Ph.D. BLAGA PETRU Ph.D.
More informationThis article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution
More informationThis article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution
More informationTopology, Math 581, Fall 2017 last updated: November 24, Topology 1, Math 581, Fall 2017: Notes and homework Krzysztof Chris Ciesielski
Topology, Math 581, Fall 2017 last updated: November 24, 2017 1 Topology 1, Math 581, Fall 2017: Notes and homework Krzysztof Chris Ciesielski Class of August 17: Course and syllabus overview. Topology
More informationWEAK CONVERGENCE OF RESOLVENTS OF MAXIMAL MONOTONE OPERATORS AND MOSCO CONVERGENCE
Fixed Point Theory, Volume 6, No. 1, 2005, 59-69 http://www.math.ubbcluj.ro/ nodeacj/sfptcj.htm WEAK CONVERGENCE OF RESOLVENTS OF MAXIMAL MONOTONE OPERATORS AND MOSCO CONVERGENCE YASUNORI KIMURA Department
More information(This is a sample cover image for this issue. The actual cover is not yet available at this time.)
(This is a sample cover image for this issue. The actual cover is not yet available at this time.) This article appeared in a journal published by Elsevier. The attached copy is furnished to the author
More informationThis article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution
More informationChebyshev Type Inequalities for Sugeno Integrals with Respect to Intuitionistic Fuzzy Measures
BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 9, No 2 Sofia 2009 Chebyshev Type Inequalities for Sugeno Integrals with Respect to Intuitionistic Fuzzy Measures Adrian I.
More informationAdrian I. Ban and Delia A. Tuşe
18 th Int. Conf. on IFSs, Sofia, 10 11 May 2014 Notes on Intuitionistic Fuzzy Sets ISSN 1310 4926 Vol. 20, 2014, No. 2, 43 51 Trapezoidal/triangular intuitionistic fuzzy numbers versus interval-valued
More informationON GENERALIZED-CONVEX CONSTRAINED MULTI-OBJECTIVE OPTIMIZATION
ON GENERALIZED-CONVEX CONSTRAINED MULTI-OBJECTIVE OPTIMIZATION CHRISTIAN GÜNTHER AND CHRISTIANE TAMMER Abstract. In this paper, we consider multi-objective optimization problems involving not necessarily
More informationThis article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution
More informationThis article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution
More informationThis article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution
More informationCiric-type δ-contractions in metric spaces endowedwithagraph
Chifu and Petruşel Journal of Inequalities and Applications 2014, 2014:77 R E S E A R C H Open Access Ciric-type δ-contractions in metric spaces endowedwithagraph Cristian Chifu 1* and Adrian Petruşel
More informationChapter 1. Measure Spaces. 1.1 Algebras and σ algebras of sets Notation and preliminaries
Chapter 1 Measure Spaces 1.1 Algebras and σ algebras of sets 1.1.1 Notation and preliminaries We shall denote by X a nonempty set, by P(X) the set of all parts (i.e., subsets) of X, and by the empty set.
More informationThe Skorokhod reflection problem for functions with discontinuities (contractive case)
The Skorokhod reflection problem for functions with discontinuities (contractive case) TAKIS KONSTANTOPOULOS Univ. of Texas at Austin Revised March 1999 Abstract Basic properties of the Skorokhod reflection
More informationThis article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution
More informationThis article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution
More informationGeometry and topology of continuous best and near best approximations
Journal of Approximation Theory 105: 252 262, Geometry and topology of continuous best and near best approximations Paul C. Kainen Dept. of Mathematics Georgetown University Washington, D.C. 20057 Věra
More information(This is a sample cover image for this issue. The actual cover is not yet available at this time.)
This is a sample cover image for this issue. The actual cover is not yet available at this time.) This article appeared in a journal published by Elsevier. The attached copy is furnished to the author
More informationA Note of the Strong Convergence of the Mann Iteration for Demicontractive Mappings
Applied Mathematical Sciences, Vol. 10, 2016, no. 6, 255-261 HIKARI Ltd, www.m-hikari.com http://dx.doi.org/10.12988/ams.2016.511700 A Note of the Strong Convergence of the Mann Iteration for Demicontractive
More informationThis article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution
More informationZERO DUALITY GAP FOR CONVEX PROGRAMS: A GENERAL RESULT
ZERO DUALITY GAP FOR CONVEX PROGRAMS: A GENERAL RESULT EMIL ERNST AND MICHEL VOLLE Abstract. This article addresses a general criterion providing a zero duality gap for convex programs in the setting of
More informationLecture Notes in Advanced Calculus 1 (80315) Raz Kupferman Institute of Mathematics The Hebrew University
Lecture Notes in Advanced Calculus 1 (80315) Raz Kupferman Institute of Mathematics The Hebrew University February 7, 2007 2 Contents 1 Metric Spaces 1 1.1 Basic definitions...........................
More informationEcon Lecture 3. Outline. 1. Metric Spaces and Normed Spaces 2. Convergence of Sequences in Metric Spaces 3. Sequences in R and R n
Econ 204 2011 Lecture 3 Outline 1. Metric Spaces and Normed Spaces 2. Convergence of Sequences in Metric Spaces 3. Sequences in R and R n 1 Metric Spaces and Metrics Generalize distance and length notions
More informationSolution of Fuzzy Growth and Decay Model
Solution of Fuzzy Growth and Decay Model U. M. Pirzada School of Engineering and Technology, Navrachana University of Vadodara, salmap@nuv.ac.in Abstract: Mathematical modelling for population growth leads
More informationResearch Article On λ-statistically Convergent Double Sequences of Fuzzy Numbers
Hindawi Publishing Corporation Journal of Inequalities and Applications Volume 2008, Article ID 47827, 6 pages doi:0.55/2008/47827 Research Article On λ-statistically Convergent Double Sequences of Fuzzy
More informationContents: 1. Minimization. 2. The theorem of Lions-Stampacchia for variational inequalities. 3. Γ -Convergence. 4. Duality mapping.
Minimization Contents: 1. Minimization. 2. The theorem of Lions-Stampacchia for variational inequalities. 3. Γ -Convergence. 4. Duality mapping. 1 Minimization A Topological Result. Let S be a topological
More informationPart V. 17 Introduction: What are measures and why measurable sets. Lebesgue Integration Theory
Part V 7 Introduction: What are measures and why measurable sets Lebesgue Integration Theory Definition 7. (Preliminary). A measure on a set is a function :2 [ ] such that. () = 2. If { } = is a finite
More informationHABILITATION THESIS. New results in the theory of Countable Iterated Function Systems
HABILITATION THESIS New results in the theory of Countable Iterated Function Systems - abstract - Nicolae-Adrian Secelean Specialization: Mathematics Lucian Blaga University of Sibiu 2014 Abstract The
More informationViscosity approximation methods for the implicit midpoint rule of asymptotically nonexpansive mappings in Hilbert spaces
Available online at www.tjnsa.com J. Nonlinear Sci. Appl. 9 016, 4478 4488 Research Article Viscosity approximation methods for the implicit midpoint rule of asymptotically nonexpansive mappings in Hilbert
More informationarxiv: v1 [math.ca] 7 Jul 2013
Existence of Solutions for Nonconvex Differential Inclusions of Monotone Type Elza Farkhi Tzanko Donchev Robert Baier arxiv:1307.1871v1 [math.ca] 7 Jul 2013 September 21, 2018 Abstract Differential inclusions
More information1 Lesson 1: Brunn Minkowski Inequality
1 Lesson 1: Brunn Minkowski Inequality A set A R n is called convex if (1 λ)x + λy A for any x, y A and any λ [0, 1]. The Minkowski sum of two sets A, B R n is defined by A + B := {a + b : a A, b B}. One
More informationThis article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution
More informationAn introduction to Mathematical Theory of Control
An introduction to Mathematical Theory of Control Vasile Staicu University of Aveiro UNICA, May 2018 Vasile Staicu (University of Aveiro) An introduction to Mathematical Theory of Control UNICA, May 2018
More informationNotes on Distributions
Notes on Distributions Functional Analysis 1 Locally Convex Spaces Definition 1. A vector space (over R or C) is said to be a topological vector space (TVS) if it is a Hausdorff topological space and the
More informationThe Generalized Viscosity Implicit Rules of Asymptotically Nonexpansive Mappings in Hilbert Spaces
Applied Mathematical Sciences, Vol. 11, 2017, no. 12, 549-560 HIKARI Ltd, www.m-hikari.com https://doi.org/10.12988/ams.2017.718 The Generalized Viscosity Implicit Rules of Asymptotically Nonexpansive
More informationMath 341: Convex Geometry. Xi Chen
Math 341: Convex Geometry Xi Chen 479 Central Academic Building, University of Alberta, Edmonton, Alberta T6G 2G1, CANADA E-mail address: xichen@math.ualberta.ca CHAPTER 1 Basics 1. Euclidean Geometry
More informationBased on the Appendix to B. Hasselblatt and A. Katok, A First Course in Dynamics, Cambridge University press,
NOTE ON ABSTRACT RIEMANN INTEGRAL Based on the Appendix to B. Hasselblatt and A. Katok, A First Course in Dynamics, Cambridge University press, 2003. a. Definitions. 1. Metric spaces DEFINITION 1.1. If
More information2. The Concept of Convergence: Ultrafilters and Nets
2. The Concept of Convergence: Ultrafilters and Nets NOTE: AS OF 2008, SOME OF THIS STUFF IS A BIT OUT- DATED AND HAS A FEW TYPOS. I WILL REVISE THIS MATE- RIAL SOMETIME. In this lecture we discuss two
More informationSOME FIXED POINT RESULTS FOR ADMISSIBLE GERAGHTY CONTRACTION TYPE MAPPINGS IN FUZZY METRIC SPACES
Iranian Journal of Fuzzy Systems Vol. 4, No. 3, 207 pp. 6-77 6 SOME FIXED POINT RESULTS FOR ADMISSIBLE GERAGHTY CONTRACTION TYPE MAPPINGS IN FUZZY METRIC SPACES M. DINARVAND Abstract. In this paper, we
More informationEQUIVALENCE OF TOPOLOGIES AND BOREL FIELDS FOR COUNTABLY-HILBERT SPACES
EQUIVALENCE OF TOPOLOGIES AND BOREL FIELDS FOR COUNTABLY-HILBERT SPACES JEREMY J. BECNEL Abstract. We examine the main topologies wea, strong, and inductive placed on the dual of a countably-normed space
More informationBest proximity points of Kannan type cyclic weak ϕ-contractions in ordered metric spaces
An. Şt. Univ. Ovidius Constanţa Vol. 20(3), 2012, 51 64 Best proximity points of Kannan type cyclic weak ϕ-contractions in ordered metric spaces Erdal Karapınar Abstract In this manuscript, the existence
More informationPart III. 10 Topological Space Basics. Topological Spaces
Part III 10 Topological Space Basics Topological Spaces Using the metric space results above as motivation we will axiomatize the notion of being an open set to more general settings. Definition 10.1.
More informationBoundedly complete weak-cauchy basic sequences in Banach spaces with the PCP
Journal of Functional Analysis 253 (2007) 772 781 www.elsevier.com/locate/jfa Note Boundedly complete weak-cauchy basic sequences in Banach spaces with the PCP Haskell Rosenthal Department of Mathematics,
More informationThis article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution
More informationB. Appendix B. Topological vector spaces
B.1 B. Appendix B. Topological vector spaces B.1. Fréchet spaces. In this appendix we go through the definition of Fréchet spaces and their inductive limits, such as they are used for definitions of function
More information1.1. MEASURES AND INTEGRALS
CHAPTER 1: MEASURE THEORY In this chapter we define the notion of measure µ on a space, construct integrals on this space, and establish their basic properties under limits. The measure µ(e) will be defined
More informationIntroduction to Dynamical Systems
Introduction to Dynamical Systems France-Kosovo Undergraduate Research School of Mathematics March 2017 This introduction to dynamical systems was a course given at the march 2017 edition of the France
More informationIntegral Jensen inequality
Integral Jensen inequality Let us consider a convex set R d, and a convex function f : (, + ]. For any x,..., x n and λ,..., λ n with n λ i =, we have () f( n λ ix i ) n λ if(x i ). For a R d, let δ a
More informationSome Properties of the Augmented Lagrangian in Cone Constrained Optimization
MATHEMATICS OF OPERATIONS RESEARCH Vol. 29, No. 3, August 2004, pp. 479 491 issn 0364-765X eissn 1526-5471 04 2903 0479 informs doi 10.1287/moor.1040.0103 2004 INFORMS Some Properties of the Augmented
More informationSet, functions and Euclidean space. Seungjin Han
Set, functions and Euclidean space Seungjin Han September, 2018 1 Some Basics LOGIC A is necessary for B : If B holds, then A holds. B A A B is the contraposition of B A. A is sufficient for B: If A holds,
More information1.4 The Jacobian of a map
1.4 The Jacobian of a map Derivative of a differentiable map Let F : M n N m be a differentiable map between two C 1 manifolds. Given a point p M we define the derivative of F at p by df p df (p) : T p
More information3 (Due ). Let A X consist of points (x, y) such that either x or y is a rational number. Is A measurable? What is its Lebesgue measure?
MA 645-4A (Real Analysis), Dr. Chernov Homework assignment 1 (Due ). Show that the open disk x 2 + y 2 < 1 is a countable union of planar elementary sets. Show that the closed disk x 2 + y 2 1 is a countable
More information(This is a sample cover image for this issue. The actual cover is not yet available at this time.)
(This is a sample cover image for this issue. The actual cover is not yet available at this time.) This article appeared in a journal published by Elsevier. The attached copy is furnished to the author
More informationProblem Set 2: Solutions Math 201A: Fall 2016
Problem Set 2: s Math 201A: Fall 2016 Problem 1. (a) Prove that a closed subset of a complete metric space is complete. (b) Prove that a closed subset of a compact metric space is compact. (c) Prove that
More informationStrong Convergence Theorem by a Hybrid Extragradient-like Approximation Method for Variational Inequalities and Fixed Point Problems
Strong Convergence Theorem by a Hybrid Extragradient-like Approximation Method for Variational Inequalities and Fixed Point Problems Lu-Chuan Ceng 1, Nicolas Hadjisavvas 2 and Ngai-Ching Wong 3 Abstract.
More informationOn Semicontinuity of Convex-valued Multifunctions and Cesari s Property (Q)
On Semicontinuity of Convex-valued Multifunctions and Cesari s Property (Q) Andreas Löhne May 2, 2005 (last update: November 22, 2005) Abstract We investigate two types of semicontinuity for set-valued
More informationChapter 2 Metric Spaces
Chapter 2 Metric Spaces The purpose of this chapter is to present a summary of some basic properties of metric and topological spaces that play an important role in the main body of the book. 2.1 Metrics
More informationAumann-Shapley Values on a Class of Cooperative Fuzzy Games
Journal of Uncertain Systems Vol.6, No.4, pp.27-277, 212 Online at: www.jus.org.uk Aumann-Shapley Values on a Class of Cooperative Fuzzy Games Fengye Wang, Youlin Shang, Zhiyong Huang School of Mathematics
More informationInterpolation on lines by ridge functions
Available online at www.sciencedirect.com ScienceDirect Journal of Approximation Theory 175 (2013) 91 113 www.elsevier.com/locate/jat Full length article Interpolation on lines by ridge functions V.E.
More informationMax-min (σ-)additive representation of monotone measures
Noname manuscript No. (will be inserted by the editor) Max-min (σ-)additive representation of monotone measures Martin Brüning and Dieter Denneberg FB 3 Universität Bremen, D-28334 Bremen, Germany e-mail:
More informationSTAT 7032 Probability Spring Wlodek Bryc
STAT 7032 Probability Spring 2018 Wlodek Bryc Created: Friday, Jan 2, 2014 Revised for Spring 2018 Printed: January 9, 2018 File: Grad-Prob-2018.TEX Department of Mathematical Sciences, University of Cincinnati,
More informationNotions such as convergent sequence and Cauchy sequence make sense for any metric space. Convergent Sequences are Cauchy
Banach Spaces These notes provide an introduction to Banach spaces, which are complete normed vector spaces. For the purposes of these notes, all vector spaces are assumed to be over the real numbers.
More informationRESTRICTED UNIFORM BOUNDEDNESS IN BANACH SPACES
RESTRICTED UNIFORM BOUNDEDNESS IN BANACH SPACES OLAV NYGAARD AND MÄRT PÕLDVERE Abstract. Precise conditions for a subset A of a Banach space X are known in order that pointwise bounded on A sequences of
More informationTopology. Xiaolong Han. Department of Mathematics, California State University, Northridge, CA 91330, USA address:
Topology Xiaolong Han Department of Mathematics, California State University, Northridge, CA 91330, USA E-mail address: Xiaolong.Han@csun.edu Remark. You are entitled to a reward of 1 point toward a homework
More informationMAT 570 REAL ANALYSIS LECTURE NOTES. Contents. 1. Sets Functions Countability Axiom of choice Equivalence relations 9
MAT 570 REAL ANALYSIS LECTURE NOTES PROFESSOR: JOHN QUIGG SEMESTER: FALL 204 Contents. Sets 2 2. Functions 5 3. Countability 7 4. Axiom of choice 8 5. Equivalence relations 9 6. Real numbers 9 7. Extended
More informationAppendix B Convex analysis
This version: 28/02/2014 Appendix B Convex analysis In this appendix we review a few basic notions of convexity and related notions that will be important for us at various times. B.1 The Hausdorff distance
More informationExtreme points of compact convex sets
Extreme points of compact convex sets In this chapter, we are going to show that compact convex sets are determined by a proper subset, the set of its extreme points. Let us start with the main definition.
More informationThis article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution
More informationLocally convex spaces, the hyperplane separation theorem, and the Krein-Milman theorem
56 Chapter 7 Locally convex spaces, the hyperplane separation theorem, and the Krein-Milman theorem Recall that C(X) is not a normed linear space when X is not compact. On the other hand we could use semi
More informations P = f(ξ n )(x i x i 1 ). i=1
Compactness and total boundedness via nets The aim of this chapter is to define the notion of a net (generalized sequence) and to characterize compactness and total boundedness by this important topological
More information1 Directional Derivatives and Differentiability
Wednesday, January 18, 2012 1 Directional Derivatives and Differentiability Let E R N, let f : E R and let x 0 E. Given a direction v R N, let L be the line through x 0 in the direction v, that is, L :=
More informationIntroduction to Convex Analysis Microeconomics II - Tutoring Class
Introduction to Convex Analysis Microeconomics II - Tutoring Class Professor: V. Filipe Martins-da-Rocha TA: Cinthia Konichi April 2010 1 Basic Concepts and Results This is a first glance on basic convex
More informationCHARACTERIZATION OF (QUASI)CONVEX SET-VALUED MAPS
CHARACTERIZATION OF (QUASI)CONVEX SET-VALUED MAPS Abstract. The aim of this paper is to characterize in terms of classical (quasi)convexity of extended real-valued functions the set-valued maps which are
More informationConvex Optimization Notes
Convex Optimization Notes Jonathan Siegel January 2017 1 Convex Analysis This section is devoted to the study of convex functions f : B R {+ } and convex sets U B, for B a Banach space. The case of B =
More informationSome Background Material
Chapter 1 Some Background Material In the first chapter, we present a quick review of elementary - but important - material as a way of dipping our toes in the water. This chapter also introduces important
More information01. Review of metric spaces and point-set topology. 1. Euclidean spaces
(October 3, 017) 01. Review of metric spaces and point-set topology Paul Garrett garrett@math.umn.edu http://www.math.umn.edu/ garrett/ [This document is http://www.math.umn.edu/ garrett/m/real/notes 017-18/01
More informationThe problems I left behind
The problems I left behind Kazimierz Goebel Maria Curie-Sk lodowska University, Lublin, Poland email: goebel@hektor.umcs.lublin.pl During over forty years of studying and working on problems of metric
More informationvan Rooij, Schikhof: A Second Course on Real Functions
vanrooijschikhof.tex April 25, 2018 van Rooij, Schikhof: A Second Course on Real Functions Notes from [vrs]. Introduction A monotone function is Riemann integrable. A continuous function is Riemann integrable.
More informationOptimization and Optimal Control in Banach Spaces
Optimization and Optimal Control in Banach Spaces Bernhard Schmitzer October 19, 2017 1 Convex non-smooth optimization with proximal operators Remark 1.1 (Motivation). Convex optimization: easier to solve,
More informationON THE INVERSE FUNCTION THEOREM
PACIFIC JOURNAL OF MATHEMATICS Vol. 64, No 1, 1976 ON THE INVERSE FUNCTION THEOREM F. H. CLARKE The classical inverse function theorem gives conditions under which a C r function admits (locally) a C Γ
More informationThis article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution
More informationCourse 212: Academic Year Section 1: Metric Spaces
Course 212: Academic Year 1991-2 Section 1: Metric Spaces D. R. Wilkins Contents 1 Metric Spaces 3 1.1 Distance Functions and Metric Spaces............. 3 1.2 Convergence and Continuity in Metric Spaces.........
More information(convex combination!). Use convexity of f and multiply by the common denominator to get. Interchanging the role of x and y, we obtain that f is ( 2M ε
1. Continuity of convex functions in normed spaces In this chapter, we consider continuity properties of real-valued convex functions defined on open convex sets in normed spaces. Recall that every infinitedimensional
More informationLocal strong convexity and local Lipschitz continuity of the gradient of convex functions
Local strong convexity and local Lipschitz continuity of the gradient of convex functions R. Goebel and R.T. Rockafellar May 23, 2007 Abstract. Given a pair of convex conjugate functions f and f, we investigate
More information2 (Bonus). Let A X consist of points (x, y) such that either x or y is a rational number. Is A measurable? What is its Lebesgue measure?
MA 645-4A (Real Analysis), Dr. Chernov Homework assignment 1 (Due 9/5). Prove that every countable set A is measurable and µ(a) = 0. 2 (Bonus). Let A consist of points (x, y) such that either x or y is
More informationCONVERGENCE OF THE STEEPEST DESCENT METHOD FOR ACCRETIVE OPERATORS
PROCEEDINGS OF THE AMERICAN MATHEMATICAL SOCIETY Volume 127, Number 12, Pages 3677 3683 S 0002-9939(99)04975-8 Article electronically published on May 11, 1999 CONVERGENCE OF THE STEEPEST DESCENT METHOD
More informationCARISTI TYPE OPERATORS AND APPLICATIONS
STUDIA UNIV. BABEŞ BOLYAI, MATHEMATICA, Volume XLVIII, Number 3, September 2003 Dedicated to Professor Gheorghe Micula at his 60 th anniversary 1. Introduction Caristi s fixed point theorem states that
More informationBest proximity problems for Ćirić type multivalued operators satisfying a cyclic condition
Stud. Univ. Babeş-Bolyai Math. 62(207), No. 3, 395 405 DOI: 0.2493/subbmath.207.3. Best proximity problems for Ćirić type multivalued operators satisfying a cyclic condition Adrian Magdaş Abstract. The
More informationThis article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research education use, including for instruction at the authors institution
More informationTopological vectorspaces
(July 25, 2011) Topological vectorspaces Paul Garrett garrett@math.umn.edu http://www.math.umn.edu/ garrett/ Natural non-fréchet spaces Topological vector spaces Quotients and linear maps More topological
More informationApproximation of the attractor of a countable iterated function system 1
General Mathematics Vol. 17, No. 3 (2009), 221 231 Approximation of the attractor of a countable iterated function system 1 Nicolae-Adrian Secelean Abstract In this paper we will describe a construction
More information