Abstract
Convex and non-convex fuzzy mappings are well known to be important in the research of fuzzy optimization. Symmetry and the idea of convexity are closely related. Therefore, the concept of symmetry and convexity is important in the discussion of inequalities because of how its definition behaves. This study aims to consider new class of generalized fuzzy variational-like inequality for fuzzy mapping which is known as perturbed fuzzy mixed variational-like inequality. We also introduce strongly fuzzy mixed variational inequality, as a particular case of perturbed fuzzy mixed variational-like inequality which is also a new one. Furthermore, by using the generalized auxiliary principle technique and some new analytic techniques, some existence results and efficient numerical techniques of perturbed fuzzy mixed variational-like inequality are established. As exceptional cases, some known and new results are obtained. Results obtained in this paper can be viewed as refinement and improvement of previously known results.
Keywords:
fuzzy mappings; perturbed fuzzy mixed variational-like inequality; auxiliary principle technique; existence; iterative method MSC:
26A33; 26A51; 26D10
1. Introduction
In [1], Zadeh introduced fuzzy set theory, which has been extensively used to simulate the uncertainty found in practical applications. The expansion of fuzzy set theory and its applications have drawn the interest of numerous academics, see [2]. With the passage of time, these have been expanded and generalized to analyze a vast class of problems like in optimization and control system, mechanics, economics and transportation, physics and so forth. The idea of variational inequality was initiated by Hartman and Stampacchia [3] in 1964. A useful generalization of variational inequality is generalized mixed variational-like inequality (in short, GMVLI). The “GMVLI” have capability and noteworthy applications in different fields like structural analysis [4], optimization theory [5,6,7,8,9,10,11], and economics [12,13]. Motivated and inspired by going on research work, many authors discussed fuzzy variational-like inequalities and its generalizations, and its applications in different fields. For more informations related to fuzzy theory and inequalities, see [14,15,16,17]. In fuzzy optimization, Noor [18,19] studied the characterization of minimum of convex fuzzy mapping through fuzzy variational inequality and fuzzy mixed variational inequality, and obtained some iterative algorithms. It is worthy to mention one of the most considered generalization of convex fuzzy mapping is preinvex fuzzy mapping. The idea of fuzzy preinvex mapping on the invex set was introduced and studied by Noor [20], Moreover, any local minimum of a preinvex fuzzy mapping is a global minimum on invex set and necessary and sufficient condition for fuzzy mapping is to be preinvex if its epigraph is an invex set. Furthermore, it has been verified that a fuzzy optimality conditions of differentiable fuzzy preinvex mappings can be distinguished by variational-like inequalities. Then many another’s extended this concept and discussed the applications of fuzzy variational-like inequalities. Chang [21], Chang and Zhu [22], and Chang et al. [23] and Kumam and Petrot [24], studied the idea of “GMVLI” and complementarity problems for fuzzy mappings in different context. By using the Berge maximum theorem in finite and infinite dimensional space, some particular cases of “GMVLI” are studied by Tian [25], Parida and Sen [26], and Yao [27]. It is convenient to mention that their methods are not productive. Therefore, the enlargement of a well-organized and applicable technique for solving variational-like inequality is one of the impotent and engaging problem. Although, there exist substantial numbers of numerical methods as well as projection method and its alternative forms, Newton’s methods, linear approximation and descent. But the projection type method can be used to recommend iterative method for variational-like inequalities. To overcome this drawback of above mentioned method, Glowinski, Lions and Tremolieres [28], was suggested auxiliary principle technique. Then many authors expended the auxiliary principle technique to study the existence and uniqueness of a solution of “GMVLI” for set valued mappings with compact and non-compact values and single valued mappings, see [29,30,31,32,33,34,35,36,37,38]. By using auxiliary principle technique, “GMVLI” can easily be handled as well as its particular cases.
Motivated and inspired by going on research work in this interesting and fascinating field, the objective of this article to introduce new class of “GMVLI”, which is known as perturbed fuzzy mixed variational-like inequality (PFMVLI) and to study some existence theorems. We also prove the existence of auxiliary problem for “PFMVLI”. By utilizing the theorems, we construct an iterative algorithms auxiliary problem for “PFMVLI”. The results given in this paper are up to date and they generalize, refine and consolidate a number of recent results in [3,21,28,31].
2. Preliminaries
Let be a real Hilbert space and be a convex set. we denote the collection CB of all nonempty bounded and closed subsets of and is the Hausdorff metric on CB defined by
A fuzzy set on is a mapping , for each fuzzy set and , then -level sets of is denoted and defined as follows . If , then is called support of . By we define the closure of .
In what follows, denote the family of all fuzzy sets on . A mapping from to is called a fuzzy mapping. If is a fuzzy mapping, then the set , for is a fuzzy set in (in the sequel we denote by ) and , is the degree of membership of in .
Definition 1.
(i) If for each the function is upper semicontinuous, then fuzzy mapping
is called closed. If is a net and satisfying → , then have following property, we have
(ii) A closed fuzzy mapping is said to satisfy the condition (∗), if there exists a function such that for each the set
is a bounded subset of .
Remark 1.
It is important to mention that, from Definition 1, (i and ii), it can easily be seen that CB
. Indeed, let be a net and → Then, from (1) and (2), we have and for each Since is closed with condition (∗), we have
This implies that and so .
Problem 1.
Let be a real Hilbert space. Then, for given nonlinear mappings , we consider the problem of finding and constant , such that
where is a nondifferentiable function and is a fuzzy mapping satisfying condition (∗) with function , such that
The inequality (3) is known as perturbed fuzzy mixed variational-like inequality (in short, PFMVLI). It can easily be seen that; this inequality is more general than fuzzy mixed variational-like inequality (in short, FMVLI) and include classical mixed variational-like inequality (in short, MVLI) and associated fuzzy optimizations problems are particular cases. For applications, see [3,17,21,22,26] and the references therein.
Now we study some certain cases of Problem (1).
- (1)
- Let be an ordinary multivalued mapping and be the mapping in problem (1).
Now we define a fuzzy mapping as follows
where is the characteristic functions of the From (4), it can straightforwardly be noticed that the is a closed fuzzy mapping fulfilling condition (∗) with constant function for all Also
- then problem (1), is parallel to finding , such that
This kind of problem is called the set valued “PMVLI”. This inequality is also new one.
If (identity mapping), and is single valued, then inequality (5) is parallel to finding a such that
This is known as “PMVLI” and studied by Noor et al. [31].
- If then problem (1) is called strongly fuzzy mixed variational inequality and is parallel to finding , such that
This class of “FMVI” is also new one. This inequality is more general as well as classical “FMVI” and related fuzzy optimization problems as particular cases. In case of ordinary set-valued mapping as flourished by Noor [9].
- If then problem (1), is parallel to finding , such that
- For inequality (7) reduces to
- When and is a fuzzy mapping, then inequality (9) is parallel to finding ,
This is known as “FMVI”, see [18].
- If is an indicator mapping of a closed invex set in that is
- when then inequality (10), is analogous to finding , such that
- If then inequality (10), is analogous to finding , such that
- If , then inequality (12), is analogous to finding , such that
This is also known as “FVI”, see [10,22]. The “VI” is mainly due to Stampacchia and Guido [13], and Lions and Stampacchia [5], as developed and studied by Noor [10]. For the applications of inequality (13), see [5,9,26,28] and the references therein.
From above discussion, it can easily be seen that inequalities (7)–(13) are particular cases of “PFMVLI” (3). In fact, “PFMVLI” is more generalize and unifying one, which is main motivation of our work. For a proper and suitable choice of , and , we can choose a number of known and unknown “FVLI” and complementary problems.
Next, we will use mathematical terminologies S-monotone and L-continuous for strongly monotone and Lipschitz continuous, respectively.
Definition 2.
If fuzzy mapping is closed and fulfil the condition (∗) with function then nonlinear mapping is said to be:
- (a)
- -S-Monotone: if there exist a constant such that
- (b)
- γ-L-continuous: if there exist a constant such that
Definition 3.
A fuzzy mapping is said to be:
- (c)
- -L-continuous: if there exist a function and constant such thatwhere is the Housdroff metric on
In particular, from (a) and (b), we have
From (c), we have
which implies that
Definition 4.
The bi-function is said to be:
- (d)
- S-monotone: if there exist constant
such that
- (e)
- L-continuous: if there exist constant
such that
From (d) and (e), we can observe that
Definition 5.
A mapping is said to be -L-continuous if there exist constant such that
For any we denote the conv(
), the convex hull of . A set valued mapping is called a KKM mapping if, for every finite subset of
Lemma 1.
[8] Let be an arbitrary nonempty in a topological vector space and let is a KKM mapping. If is closed for all and is compact for at least one , then
Theorem 1.
[21] Let be a locally convex Hausdorff topologically vector space and be a properly convex functional. Then is lower semicontinuous on if and only if, is weakly lower semicontinuous on
Assumption 1.
Let and be two mappings satisfying the following condition
- (a)
- (and so for all , for all
- (b)
- For any given the mapping is concave, where
- (c)
- For any given the mapping is lower semicontinuous, where
Lemma 2.
[11] Let be a complete metric space and CB(
) and be any real number. Then, for every then there exist
such that
In next sections, we will use the above results.
3. Auxiliary Principle and Algorithm
In this portion, we explore the auxiliary principle technique which is mainly due to Glowinski, Lions and Tremolieres [28], as flourished and improved by Noor [32], and Huang and Deng [17], to examine “PFMVLI”. We give an existence result of the solution of auxiliary problem for the “PFMVLI” (3). Furthermore, based on this existence result, we suggest an iterative algorithm for the “PFMVLI” (3).
For a given satisfying the problem (1), we consider the problem of finding such that
where is a constant. The inequality (14) is also called auxiliary “PFMVLI”.
Theorem 2.
Let be real Hilbert space and be a closed bounded subset of . Let be a closed continuous fuzzy mapping satisfying condition (∗) with function . Let be a properly convex functional (i.e., A functional is called proper, if for all and ) and -L-continuous with constant . If the Assumption 1 is satisfied and bi-function is L-continuous with constant . Then auxiliary problem (14), has a unique solution.
Proof.
Let given , we consider the mapping define by
for all .
From (15), it can easily be seen that, for each , since To obtain the solution, firstly we show that is a KKM mapping. Suppose the contrary, that is, is not KKM mapping, then there exist a finite subset of and constant , with such that”
Then, we have
From Assumption 1 and by the convexity of , so the above inequality yield
This is a contradiction. Hence, is a KKM mapping.
Since the weak closure of ] is weakly closed subset of a bounded set so it is weakly compact. Hence by Lemma 1
Let
Then, there exists a sequence in (fix ), such that Then
Now, by using the property of inner product
Since is weakly lower semicontinuous, we have
By using Lipchitz continuity and convexity of and Assumption 1, we get
This implies that
for all
Hence is solution of auxiliary problem (14).
Uniqueness: Let be also a solution of auxiliary problem. Then we have
for all
Replacing by in (19) and by in (20), we have
and
using the Assumption 1, that and then adding the (21) and (22), we have
which implies that is the uniqueness of the solution of auxiliary problem (14).This complete the proof of Theorem 2. □
Algorithm 1.
.
At start with initial value
from Theorem 2, the auxiliary problem (14) has a unique solution , such that
Since then by Nadler’s Lemma 2, there exist , such that
For
again from Theorem 2, the auxiliary problem (14) has a unique solution , such that
Since then by Nadler’s Lemma 2, there exist , such that
At step we can obtain sequences
such that
- (i)
- (ii)
4. Existence and Convergence
In this section, we discuss the existence and convergence of perturbed fuzzy-mixed variational-like inequality with the help of auxilary prencilple.
Theorem 3.
Let statement of Theorem 2 hold and let be a -L-continuous with constant . Let nonlinear continuous mapping be -S-monotone and -L-continuous with constants and respectively and is nondifferentiable. If the bi-function is L-continuous with constant , respectively, then for constant , such that
where
Then there exist
satisfying the “PFMVLI” (3) and the sequences and , generated by (23) converge strongly to and , respectively.
Proof .
From Algorithm 1 and auxiliary problem (14), for any we have
and
By taking in (25) and in (26), we get
and
Adding (27) and (28) and using we have
It follows that
By the -S-monotonicity and -L-continuity of and -L-continuity of we have
and
Similarly, by the S-monotonicity and L-continuity of bifunction , we have
Combining (29), (30) and (31) and using the -L-continuity of and -L-continuity of and , respectively, we have
which implies that
It follows that,
where
Clearly
where
From (24), it follows that . Hence, it follows from (32) and (33) that {} is Cauchy sequence in . Hence it converges to some point. Since is closed convex set in , then there exist a in such that which satisfy the “PFMVLI” (3).
On the other hand, from algorithm 1, we have
This implies that {} is Cauchy sequence in , since {} is converging sequence. Then, we consider that when. Since so we have
When , we have
Hence,
Now finally we show that
We again study (23), as follows
Now from Lipchitz continuity and convexity of and by Assumption 1, we get
This implies that
This completes the proof. □
Special Cases:
Theorem 4.
Let the operator be -S-monotone with constant and -L-continuous with constant respectively. Let nonlinear continuous mapping be -S-monotone and -L-continuous with constants and respectively. Let be a properly convex functional and -L-continuous with constant and is nondifferentiable. If the bi-function is S-monotone with constant and L-continuous with constant , respectively. If Assumption 1 hold, then for constant , such that
where
Then there exist
satisfying the set-valued “PMVLI” (5) and the sequences and , generated by (23) converge strongly to and , respectively.
Proof .
Therefore, the conclusion of Theorem 4 can be obtained from Theorem 2 and Theorem 3 immediately. □
By using the set valued mapping we define the fuzzy mapping , as follows
where is a characteristic function of the sets It is easy to see that is a closed fuzzy mapping satisfying condition (∗) with constant functions for all Also
- then Problem (1), is analogous to finding , such that
Algorithm 2.
If (identity mapping), and is single valued, then Theorem 4, reduces to:
At start with initial value
from Theorem 2, the auxiliary problem (14) has a unique solution , such that
Since then by Nadler’s Lemma 2, there exist , such that
For
again from Theorem 2, the auxiliary problem (14) has a unique solution , such that
Since then by Nadler’s Lemma 2, there exist , such that
At step we can obtain sequences
such that
- (i)
- (ii)
Theorem 5.
Let be >-S-monotone with constant and -L-continuous with constant respectively. Let be a nondifferentiable mapping and the bi-function is S-monotone with constant and L-continuous with constant , respectively. If satisfies following condition
then for constant , such that
where
Then there exist satisfying “PVLI” (6) and the sequences generated by (23) converge strongly to , see [31].
Remark 2.
At the end, we would like to mention that many earlier defined familiar methods as well as decent, projection techniques and its mixed forms, relaxation, and Newton’s methods that can be obtained form auxiliary “PFMVLI”. Similarly, for suitable choice of the fuzzy mapping and bi-function many old of (fuzzy) “VI” and (fuzzy) variational inclusion and corresponding (fuzzy) optimization problems from the “PFMVLI” (3) can be obtain. In future, we will try to find “PFMVLI” for higher order strongly-preinvex fuzzy mappings.
5. Conclusions
In this paper, we have proposed the idea of “PFMVLI”. As a particular case of “PFMVLI”, strongly fuzzy mixed variational inequality are also introduced. With the help of generalized auxiliary principle technique and some new analytic techniques, some existence theorems of auxiliary “PFMVLI” are studied for “PFMVLI” and some iterative methods are obtained for the solution of “PFMVLI”. Then we have obtained some known and new results. There is much room for further study to explore this concept because for suitable choice of the fuzzy mapping , , and many new classes of (fuzzy) “VI” and (fuzzy) variational inclusion and corresponding (fuzzy) optimization problems can be obtain from the “PFMVLI” (3).
Author Contributions
Conceptualization, M.B.K.; methodology, M.B.K.; validation, M.A.N. and M.S.S.; formal analysis, G.S.-G. and S.T.; investigation, M.A.N. and G.S.-G.; resources, M.B.K.; data curation, M.S.S.; writing—original draft preparation, M.B.K., G.S.-G. and M.S.S.; writing—review and editing, M.B.K.; visualization, M.S.S.; supervision, M.B.K. and M.A.N.; project administration, M.B.K. and S.T.; funding acquisition, G.S.-G. All authors have read and agreed to the published version of the manuscript.
Funding
The research of Santos-García was funded by the project ProCode-UCM (PID2019-108528RB-C22) from the Spanish Ministerio de Ciencia e Innovación.
Data Availability Statement
Not applicable.
Acknowledgments
The authors would like to thank the Rector, COMSATS University Islamabad, Islamabad, Pakistan, for providing excellent research and academic environments.
Conflicts of Interest
The authors declare no conflict of interest.
References
- Zadeh, L.A. Fuzzy sets, Information and control; Springer Science & Business Media: Berlin, Germany, 1965; Volume 8, pp. 338–353. [Google Scholar]
- Zimmermann, H.J. Fuzzy Set Theory—And Its Applications; Springer Science & Business Media: Berlin, Germany, 2011. [Google Scholar]
- Hartman, P.; Stampacchia, G. On some non-linear elliptic differential-functional equations. Acta Math. 1966, 115, 271–310. [Google Scholar] [CrossRef] [Scilit]
- Panagiotopoulos, P.D.; Stavroulakis, G.E. New types of variational principles based on the notion of quasi differentiability. Acta Mech. 1992, 94, 171–194. [Google Scholar] [CrossRef] [Scilit]
- Lions, J.L.; Stampacchia, G. Variational inequalities. Commun. Pure Appl. Math. 1967, 20, 493–519. [Google Scholar] [CrossRef] [Scilit]
- Noor, M.A. Variational-like inequalities. Optimization 1994, 30, 323–330. [Google Scholar] [CrossRef] [Scilit]
- Noor, M.A. Fundamentals of mixed quasi variational inequalities. Int. J. Pure Appl. Math. 2004, 15, 137–258. [Google Scholar]
- Fan, K. A generalization of Tychonoff’s fixed point theorem. Math. Ann. 1961, 142, 305–310. [Google Scholar] [CrossRef] [Scilit]
- Noor, M.A. On a class of variational inequalities. J. Math. Anal. Appl. 1987, 128, 138–155. [Google Scholar] [CrossRef] [Scilit]
- Noor, M.A. Variational inequalities for fuzzy mappings (I). Fuzzy Sets Syst. 1993, 55, 309–312. [Google Scholar] [CrossRef] [Scilit]
- Rus, I.A. Generalized Contractions and Applications; Cluj University Press: Cluj-Napoca, Romania, 2001. [Google Scholar]
- Hanson, M.A. On sufficiency of the Kuhn-Tucker conditions. J. Math. Anal. Appl. 1981, 80, 545–550. [Google Scholar] [CrossRef] [Scilit]
- Stampacchia, G. Formes bilinéaires coercitives sur les ensembles convexes. Comptes Rendus Hebd. Seances L Acad. Sci. 1964, 258, 4413. [Google Scholar]
- Khan, M.B.; Santos-García, G.; Noor, M.A.; Soliman, M.S. Some new concepts related to fuzzy fractional calculus for up and down convex fuzzy-number valued functions and inequalities. Chaos Solitons Fractals 2022, 164, 112692. [Google Scholar] [CrossRef] [Scilit]
- Khan, M.B.; Santos-García, G.; Treanța, S.; Soliman, M.S. New Class Up and Down Pre-Invex Fuzzy Number Valued Mappings and Related Inequalities via Fuzzy Riemann Integrals. Symmetry 2022, 14, 2322. [Google Scholar] [CrossRef] [Scilit]
- Khan, M.B.; Zaini, H.G.; Santos-García, G.; Noor, M.A.; Soliman, M.S. New Class Up and Down λ-Convex Fuzzy-Number Valued Mappings and Related Fuzzy Fractional Inequalities. Fractal Fract. 2022, 6, 67. [Google Scholar] [CrossRef] [Scilit]
- Huang, N.J.; Deng, C.X. Auxiliary principle and iterative algorithms for generalized set-valued strongly nonlinear mixed variational-like inequalities. J. Math. Anal. Appl. 2001, 256, 345–359. [Google Scholar] [CrossRef] [Scilit]
- Noor, M.A. Variational inequalities for fuzzy mappings (II). Fuzzy Sets Syst. 1998, 97, 101–107. [Google Scholar] [CrossRef] [Scilit]
- Noor, M.A. Variational inequalities for fuzzy mappings (III). Fuzzy Sets Syst. 2000, 110, 101–108. [Google Scholar] [CrossRef] [Scilit]
- Noor, M.A. Fuzzy preinvex functions. Fuzzy Sets Syst. 1994, 64, 95–104. [Google Scholar] [CrossRef] [Scilit]
- Chang, S.S. Variational Inequality and Complementarity Problems Theory and Applications; Shanghai Scientific and Technological Literature Publishing House: Shanghai, China, 1991. [Google Scholar]
- Chang, S.S.; Zhu, Y.G. On variational inequalities for fuzzy mappings. Fuzzy Sets Syst. 1989, 32, 359–367. [Google Scholar] [CrossRef] [Scilit]
- Chang, S.S.; O’Regan, D.; Tan, K.K.; Zeng, L.C. Auxiliary principle and fuzzy variational-like inequalities. J. Inequalities Appl. 2005, 2005, 479–494. [Google Scholar] [CrossRef] [Scilit]
- Kumam, P.; Petrot, N. Mixed variational-like inequality for fuzzy mappings in reflexive Banach spaces. J. Inequalities Appl. 2009, 2009, 209485. [Google Scholar] [CrossRef] [Scilit]
- Tian, G.Q. Generalized quasi-variational-like inequality problem. Math. Oper. Res. 1993, 8, 752–764. [Google Scholar] [CrossRef] [Scilit]
- Parida, J.; Sen, A. A variational-like inequality for multifunctions with applications. J. Math. Anal. Appl. 1987, 124, 73–81. [Google Scholar] [CrossRef] [Scilit]
- Yao, J.C. The generalized quasi-variational inequality problem with applications. J. Math. Anal. Appl. 1991, 15, 139–160. [Google Scholar] [CrossRef] [Scilit]
- Glowinski, R.; Lions, J.L.; Tremolieres, R. Numerical Analysis of Variational Inequalities; Elsevier Science Ltd.: North Holland, Amsterdam, 1981. [Google Scholar]
- Khan, M.B.; Noor, M.A.; Al-Shomrani, M.M.; Abdullah, L. Some Novel Inequalities for LR-h-Convex Interval-Valued Functions by Means of Pseudo Order Relation. Math. Meth. Appl. Sci. 2022, 45, 1310–1340. [Google Scholar] [CrossRef] [Scilit]
- Khan, M.B.; Noor, M.A.; Noor, K.I.; Chu, Y.M. New Hermite–Hadamard–type inequalities for (h1, h2)—Convex fuzzy-interval-valued functions. Adv. Differ. Equ. 2021, 2021, 149. [Google Scholar] [CrossRef] [Scilit]
- Bloach, M.I.; Noor, M.A. Perturbed mixed variational-like inequalities. AIMS Math. 2020, 5, 2153. [Google Scholar] [CrossRef] [Scilit]
- Noor, M.A. Auxiliary principle for generalized mixed variational-like inequalities. J. Math. Anal. Appl. 1997, 215, 75–85. [Google Scholar] [CrossRef] [Scilit]
- Noor, M.A.; Noor, K.I. Some characterizations of strongly preinvex functions. J. Math. Anal. Appl. 2006, 316, 697–706. [Google Scholar] [CrossRef] [Scilit]
- Noor, M.A.; Noor, K.I.; Baloch, M.I. Auxiliary Principle Technique for strongly mixed variational-like inequalities. UPB. Sci. Bull. Series A 2018, 80, 93–100. [Google Scholar]
- Khan, M.B.; Noor, M.A.; Noor, K.I.; Chu, Y.M. Higher-order strongly preinvex fuzzy mappings and fuzzy mixed variational-like inequalities. Int. J. Comput. Intell. Syst. 2021, 14, 1856–1870. [Google Scholar] [CrossRef] [Scilit]
- Aslam Noor, M.; Inayat Noor, K.; MYAl-Bayatti, H. Higher order variational inequalities. Inf. Sci. Lett. 2022, 11, 5. [Google Scholar]
- Van Hieu, D.; Cholamjiak, P. Modified extragradient method with Bregman distance for variational inequalities. Appl. Anal. 2022, 101, 655–670. [Google Scholar] [CrossRef] [Scilit]
- Cai, T.; Yu, D.; Liu, H.; Gao, F. Computational analysis of variational inequalities using mean extra-gradient approach. Mathematics 2022, 10, 2318. [Google Scholar] [CrossRef] [Scilit]
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |
© 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).