Solving linear and nonlinear Abel fuzzy integral equations by homotopy analysis method

(1)

JournalofTaibahUniversityforScience9(2015)104–115

Availableonlineatwww.sciencedirect.com

ScienceDirect

Solving

linear

and

nonlinear

Abel

fuzzy

integral

equations

by

homotopy

analysis

method

Farshid

Mirzaee

a,∗

,

Mohammad

Komak

Yari

a

,

Mahmoud

Paripour

b

a_Department_of_Mathematics,_Faculty_of_Science,_Malayer_University,_Malayer_65719-95863,_Iran

b_Department_of_Mathematics,_Hamedan_University_of_Technology,_Hamedan_65155-579,_Iran

Availableonline1July2014

Abstract

ThemainpurposeofthisarticleistopresentanapproximationmethodforsolvingAbelfuzzyintegralequationinthemost generalform.Theproposedapproachisbasedonhomotopyanalysismethod.ThismethodtransformslinearandnonlinearAbel fuzzyintegralequationsintotwocrisplinearandnonlinearintegralequations.Theconvergenceanalysisfortheproposedmethod isalsointroduced.Wegivesomenumericalapplicationstoshowefficiencyandaccuracyofthemethod.Allofthenumerical computationshavebeenperformedonacomputerwiththeaidofaprogramwritteninMatlab.

Keywords:Fuzzynumber;Fuzzyintegralequation;Abelfuzzyintegralequations;Homotopyanalysismethod

1. Introduction

Fuzzyintegralequationsareimportantinstudyingandsolvingalargeproportionoftheproblemsinmanytopicsin appliedmathematics,inparticularinrelationtophysics,geographic,medicalandbiology.Usuallyinmanyapplications, someoftheparametersinourproblemsarerepresentedbyfuzzynumberratherthancrisp,andhenceitisimportantto developmathematicalmodelsandnumericalproceduresthatwouldappropriatelytreatgeneralfuzzyintegralequations andsolvethem.

TheconceptofintegrationoffuzzyfunctionswasfirstintroducedbyDuboisandPrade[1].Alternativeapproaches werelatersuggestedbyGoetschelandVoxman[2],Kaleva[3],Nanda[4]andothers.WhileGoetschelandVoxman[2]

preferredaRiemannintegraltypeapproach,Kalva[3]definedtheintegraloffuzzyfunction,usingtheLebesguetype conceptforintegration.OneofthefirstapplicationsoffuzzyintegrationwasgivenbyWuandMa[5],whoinvestigated thefuzzyFredholmintegralequationofthesecondkind(FFIE-2).Thisworkwhichestablishedtheexistenceofaunique solutionfor(FFIE-2)wasfollowedbyotherworkssuchasMirzaeeetal.[6]andNguyen[7]whereanoriginalfuzzy

∗_{Corresponding}_author._Tel.:₊₉₈_8132355466;_fax:₊₉₈_8132355466.

E-mailaddresses:[email protected],[email protected](F.Mirzaee).

PeerreviewunderresponsibilityofTaibahUniversity

http://dx.doi.org/10.1016/j.jtusci.2014.06.006

CORE Metadata, citation and similar papers at core.ac.uk

(2)

differentialequation isreplacedbyafuzzy integralequation. RecentlyLiao,inhis Ph.D.thesis[8],hasproposed thehomotopy analysismethod(HAM)tosolvesomeclassesofnonlinearequations.Stepbystep,themethodwas developedanditseffectivenesswasprovedinhandlingnonlinearequations[8–11].

Abelintegralequationsoccurinmanybranchesofscientificfields,suchasmicroscopy,seismology,radioastronomy, electronemission,atomicscattering,radarranging,plasmadiagnostics,X-rayradiography,andopticalfiberevaluation

[12].

Recently,Mirzaeeetal.[13–15]havestudiedthenumericalsolutionsoftheFredholmfuzzyintegral equations. Since thehomotopy analysismethodisapowerfuldevicefor solvingawidevarietyof problemsarisinginmany scientificapplications,wewilldevelopthenumericalmethodsfortheapproximatesolutionsoflinearandnonlinear Abelfuzzyintegralequations.

Thestructureofthispaperisorganizedasfollows:inSection2,somebasicdefinitionsandresultswhichwillbeused lateraregiven.InSection3,Abelfuzzyintegralequationsareintroduced.InSection4,weapplyhomotopyanalysis methodtosolveAbelfuzzyintegralequations,thentheproposedmethodisimplementedforsolvingthreeillustrative examplesinSection5andfinally,conclusionisdrawninSection6.

2. Preliminaries

Wenowrecallsomedefinitionsneededthroughthepaper.

Deﬁnition1. (Kaleva[3]).Afuzzynumberisafuzzysetv:R1→I=[0,1]whichsatisfies • visuppersemicontinuous,

• v(x)=0outsidesomeinterval[c,d],

• Therearerealnumbersa,b:c≤a≤b≤dforwhich • v(x)ismonotonicincreasingon[c,a],

• v(x)ismonotonicdecreasingon[b,d],

• v(x)=1,a≤x≤b.

ThesetofallsuchfuzzynumberisdenotedbyRF.

Deﬁnition2. (Kaleva[3]).LetVbeafuzzysetonR.Viscalledafuzzyintervalif: • Visnormal:thereexistsx0∈RsuchthatV(x0)=1.

• Visconvex:forallx,t∈Rand0≤λ≤1,itholdsthatV(λx+(1−λ)t)≥min{V(x),V(t)}, • Visuppersemi-continuous:foranyx0∈R,itholdsthatV(x0)≥ lim

x→0±

V(x), • [V]α₌_Cl_{x_∈_R_|V(x)_>₀_}_is_a_compact_subset_of_R.

Theα-cutofafuzzyintervalVwith0<α≤1isthecrispset[V]α={x∈R|V(x)>0}.ForafuzzyintervalV,itsα-cutare closedintervalsinR.Theywillbedenotedbythemby[V]α =[V(α),V(α)].Analternativedefinitionorparametric formofafuzzynumberwhichyieldsthesameRFisgivenbyKaleva[8]asfollows:

Deﬁnition3. (Maetal.[16]).Anarbitraryfuzzynumber˜uintheparametricformisrepresentedbyanorderedpair offunctions(u(r),u(r))whichsatisfythefollowingrequirements:

• u(r)isaboundedleft-continuousnon-decreasingfunctionover[0,1], • u(r)isaboundedright-continuousnon-increasingfunctionover[0,1], • u(r)≤u(r),forall0≤r≤1.

Forarbitraryfuzzynumbers ˜v=(v(r),v(r)), ˜w=(w(r),w(r))andrealnumberλ,onemaydefinetheadditionandthe scalarmultiplicationofthefuzzynumbersbyusingtheextensionprincipleasfollows:

(3)

• ˜v = ˜w ifandonlyifv(r)=w(r)andv(r)=w(r), • ˜v⊕ ˜w=(v(r)+w(r),v(r)+w(r)), • (λ ˜v)= (λv(r),λv(r)) λ≥0 (λv(r),λv(r)) λ<0.

Deﬁnition4. (Ga[17]).Forarbitrarynumbers ˜v=(v(r),v(r))and ˜w=(w(r),w(r))

D(˜v, ˜w)=max{ sup

0≤r≤1|v(r)−

w(r)|, sup

0≤r≤1|v(r)−

w(r)|},

inthedistancebetween ˜vand ˜w.Itisprovedthat(RF,D)isacompletemetricspacewithfollowingproperties[5]

• D( ˜u+ ˜w, ˜v+ ˜w)=D( ˜u, ˜v); ∀˜u, ˜v, ˜w∈RF,

• D(k˜u,k ˜v)=|k|D(˜u, ˜v); ∀˜u, ˜v∈RF ∀k∈R,

• D( ˜u⊕ ˜v, ˜w⊕˜e)≤D( ˜u, ˜w)+D(˜v,˜e); ∀˜u, ˜v, ˜w,˜e∈RF.

Deﬁnition5. (Anastassiou[18]).Let ˜f,˜g:[a,b]→RF,befuzzyrealnumbervaluedfunctions.Theuniformdistance

between ˜f,˜gisdefinedby

D( ˜f ,˜g)=sup{D( ˜f (x),˜g(x))|x∈[a,b]},

InGoetschelandVoxman[2]theauthors provedthatif thefuzzyfunction ˜f(x) iscontinuousinthemetricD,its definiteintegralexistsandalso,

b a f(x,r)dx= b a f(x,r)dx, b a f(x,r)dx= b a f(x,r)dx.

Where(f(x,r),f(x,r))istheparametricformof ˜f(x).Itshouldbenotedthatthefuzzyintegralcanbealsodefined usingtheLebesgue-typeapproach[3].However,if ˜f(x)becontinuous,bothapproachesyieldthesamevalue.Moreover, therepresentationofthefuzzyintegralismoreconvenientfornumericalcalculations.Moredetailsabouttheproperties ofthefuzzyintegralaregivenin[6,8].

Deﬁnition6. (Wu[19]).Afuzzyrealnumbervaluedfunction ˜f :[a,b]→RF,issaidtobecontinuousinx0∈[a,b],

ifforeach>0thereisδ>0suchthatD( ˜f(x), ˜f(x0))<,wheneverx∈[a,b]and|x−x0|<δ.Wesaythatfisfuzzy

continuouson[a,b]iffiscontinuousateachx0∈[a,b]anddenotethespaceofallsuchfunctionsbyCF([a,b]).

Lemma1. (Anastassiou[18]).If ˜f,˜g:[a,b]⊆R→RF are fuzzycontinuousfunction,then thefunction F:[a,

b]→R+by ˜F(x)=D( ˜f(x),˜g(x))iscontinuouson[a,b],and D b a ˜ f(x)dx, b a ˜g(x)dx ≤ b a D( ˜f(x),˜g(x))dx.

Theorem1. (Hc[20]).Let ˜f(x)beafuzzyvaluefunctionon[a,∞)anditisrepresentedby(f(x,r),f(x,r)).For anyﬁxedr∈[0,1],assumethatf(x,r)andf(x,r)areRiemann-integrableon[a,b]foreverybaandassumethere aretwopositivefunctionsM(r)andM(r)suchthat_ab|f(x,r)|dxM(r)and_ab|f(x,r)|dxM(r)foreveryba. Then ˜f(x)isimproperfuzzyRiemann-integrableon[a,∞)andtheimproperfuzzyRiemann-integralisafuzzynumber. Further,wehave: _∞ a ˜ f(x)dx= _∞ a f(x,r)dx, _∞ a f(x,r)dx .

(4)

3. Abelfuzzyintegralequations TheAbelintegralequationis[21,22]

f(x)= x

a

u(t)

(x−t)αdt; a≤x≤b, (1)

whereαisaknownconstantsuchthat0<α<1,f(x)isapredetermineddatafunctionandu(x)isunknownfunction thatwillbedetermined.Theexpression(x−t)−αiscalledthekernelofAbelintegralequation,orsimplyAbelkernel, thatissingularast→x.

Iff(x)isacrispfunction,thenthesolutionsof Eq.(1) arecrisptoo.However,iff(x)isafuzzyfunction,these equationsmayonlypossessfuzzysolutions.Inthispaper,theAbelfuzzyintegralequationsarediscussed.Introducing theparametricformsoff(x)andu(x),wehavetheparametricformoffuzzyAbleintegralequationasfollows:

(f(x,r),f(x,r))= x a u(t,r) (x−t)αdt, x a u(t,r) (x−t)αdt , (2)

where0≤r≤1andαisaknownconstantsuchthat0<α<1, ˜f(x)=(f(x,r),f(x,r))isapredetermineddatafunction and˜u(x)=(u(x,r),u(x,r))isthesolutionthatwillbedetermined.

Byputtingα=1/2inEq.(2),weobtainthestandardformofthenonlinearAbelfuzzyintegralequationas

(f(x,r),f(x,r))= x a F(u(t,r)) √ x−t dt, x a F(u(t,r)) √ x−t dt , (3)

wherethefunction(f(x,r),f(x,r))isagivenreal-valuedfunction,and(F (u(x,r)),F(u(x,r)))isanonlinearfunction of(u(x,r),u(x,r)).Recallthattheunknownfunction(u(x,r),u(x,r))occursonlyinsidetheintegralsignfortheAbel fuzzyintegralEq.(3).

4. ThehomotopyanalysismethodforsolvingAbelfuzzyintegralequations

LetusconsidertheAbelfuzzyintegralEq.(2).WefirstremarkthatEq.(2)isnotwritteninthecanonicalform ofHAM,necessaryforcalculatingthedecompositionsolutionseries.Furthermore,thelinearoperatordefinedbyEq.

(2) generallydoes nothaveaninverse so itis difficulttoobtain aprecisenumericalsolution byHAM.Forthese considerations,webeginouranalysisbyputtingα=1/2andwritingEq.(2)as:

(f(x,r),f(x,r))= x a u(x,r)+u(t,r)−u(x,r) √ x−t dt, x a u(x,r)+u(t,r)−u(x,r) √ x−t dt , (4) thus (f(x,r),f(x,r))= x a u(x,r) √ x−tdt+ x a u(t,r)−u(x,r) √ x−t dt, x a u(x,r) √ x−tdt+ x a u(t,r)−u(x,r) √ x−t dt , (5) so (f(x,r),f(x,r))= 2√x−au(x,r)+ x a u(t,r)−u(x,r) √ x−t dt,2 √ x−au(x,r)+ x a u(t,r)−u(x,r) √ x−t dt , (6) therefore,itisclearthatEq.(2)canbereplacedbyasuitableequivalentexpression(6),whichiswritteninthecanonical formandthenitcanbesolvedbymeansoftheHAMdecompositionmethod.PriortoapplyingHAMforEq.(6)we rewriteEq.(6)inthefollowingform

(u(x,r),u(x,r))= f(x,r) 2√x−a− 1 2√x−a x a u(t,r)−u(x,r) √ x−t dt, f(x,r) 2√x−a− 1 2√x−a x a u(t,r)−u(x,r) √ x−t dt , (7)

(5)

Eq.(7)isasystemoflinearAbelintegralequationsincrispcaseforeach0≤r≤1.Tosolvesystem(7)byHAM,we constructthezero-orderdeformationequation

(1−p)L[U(x,p;r)−Z0(x;r)]=pc U(x,p;r)− f(x,r) 2√x−a + 1 2√x−a x a U(t,p;r)−U(x,p;r) √ x−t dt , [(1−p)L[U(x,p;r)−Z0(x;r)]=pc U(x,p;r)− f(x,r) 2√x−a + 1 2√x−a x a U(t,p;r)−U(x,p;r) √ x−t dt , (8)

wherep∈[0,1]istheembeddingparameter,cisnon-zeroauxiliaryparameter,Lisanauxiliarylinearoperator,Z₀(x,r) andZ0(x,r)areinitialguessesofu(x,r)andu(x,r)respectivelyandU(x,p;r)andU(x,p;r)areunknownfunction

dependonthevariablep.Usingtheabovezero-orderdeformationequation,withassumptionL[u]=u,wehave

(1−p)[U(x,p;r)−Z0(x;r)]=pc U(x,p;r)−√f(x,r) x−a + 1 2√x−a x a U(t,p;r)−U(x,p;r) √ x−a dt , (1−p)[U(x,p;r)−Z0(x;r)]=pc U(x,p;r)−√f(x,r) x−a + 1 2√x−a x a U(t,p;r)−U(x,p;r) √ x−a dt . (9)

Obviously,whenp=0andp=1,itholds ⎧ ⎨ ⎩ U(x,0;r)=Z₀(x;r) U(x,0;r)=Z0(x;r) , (10) and U(x,1;r)= _f (x;r) 2√x−a − 1 2√x−a x a U(t,1;r)−U(x,1;r) √ x−t dt , U(x,1;r)= f(x;r) 2√x−a − 1 2√x−a x a U(t,1;r)−U(x,1;r) √ x−t dt . (11)

Thus,aspincreasesfrom0to1,thesolution(U(x,p;r),U(x,p;r))variesfrominitialguess(Z0(x;r),Z0(x;r))tothe

solution(u(x;r),u(x;r)).ExpandingU(x,p;r)andU(x,p;r)inTaylorserieswithrespectp,wehave

U(x,p;r)=Z₀(x;r)+ ∞ m=1 u_m(x;r)pm, U(x,p;r)=Z0(x;r)+ ∞ m=1 um(x;r)pm, (12) where u_m(x;r)= 1 m! dmU(x,p;r) dpm |p=0, um(x;r)= 1 m! dmU(x,p;r) dpm |p=0. (13)

ItshouldbenotedthatU(x,0;r)=Z0(x;r)andU(x,0;r)=Z0(x;r).Differentiatingthezero-orderdeformationEq. (9)mtimeswithrespecttotheembeddingparameterpandthensettingp=0andfinallydividingthembym!,wehave

um(x;r)−χmum−1(x;r)= um−1(x;r)+(χm−1) f(x;r) 2√x−a+ 1 2√x−a x a u_m₋₁(t;r)−u_m₋₁(x;r) √ x−t dt , um(x;r)−χmum−1(x;r)= um−1(x;r)+(χm−1) f(x;r) 2√x−a+ 1 2√x−a x a um−1(t;r)−um−1(x;r) √ x−t dt . (14)

(6)

Wherem≥1and

χm=

0 m≤1

1 m≥2 , (15)

andu₀(x;r)=Z0(x;r)andu0(x;r)=Z0(x;r).IfwetakeZ0(x;r)=Z0(x;r)=0,thenwehave

u₁(x;r)=−c f(x;r) 2√x−a, u1(x;r)=−c f(x;r) 2√x−a, .. . u_m(x;r)=(1+c)u_m₋₁+c 1 2√x−a x a u_m₋₁(t;r)−u_m₋₁(x;r) √ x−t dt, um(x;r)=(1+c)um−1+c 1 2√x−a x a um−1(t;r)−um−1(x;r) √ x−t dt, (16)

wherem≥2.Usingthefact 1 2√x−a x a u_m₋₁(t;r)−u_m₋₁(x;r) √ x−t dt= 1 2√x−a x a u_m₋₁(t;r) √ x−t dt− 1 2√x−a x a u_m₋₁(x;r) √ x−t dt, 1 2√x−a x a um−1(t;r)−um−1(x;r) √ x−t dt= 1 2√x−a x a um−1(t;r) √ x−t dt− 1 2√x−a x a um−1(x;r) √ x−t dt, (17) and 1 2√x−a x a u_m₋₁(x;r) √ x−t dt= u_m₋₁(x;r) 2√x−a x a 1 √ x−tdt, 1 2√x−a x a u_m₋₁(x;r) √ x−t dt= um−1(x;r) 2√x−a x a 1 √ x−tdt, (18)

andusingthefact x a 1 √ x−tdt=2 √ x−a, (19)

forEqs.(16)–(19),weobtain

u₁(x;r)=−c f(x;r) 2√x−a, u1(x;r)=−c f(x;r) 2√x−a, .. . u_m(x;r)=u_m_−1(x;r)+c 1 2√x−a x a u_m₋₁(t;r) √ x−t dt, um(x;r)=um−1(x;r)+c 1 2√x−a x a um_√−1(t;r) x−t dt (20) wherem≥2.

Proposition1. ConsiderthefollowingAbelfuzzyintegralequations

2n+1Γ(n+1) 1×3×5···×(2n+1)g(r)x n+(1/2)_, 2n+1Γ(n+1) 1×3×5···×(2n+1)g(r)x n+(1/2) = x 0 ˜u(t) √ x−tdt. (21)

(7)

and √ πΓ((n+2)/2) Γ((n+3)/2)g(r)x (n+1)/2_,√_πΓ((n+2)/2) Γ((n+3)/2)g(r)x (n+1)/2 = x 0 ˜u(t) √ x−tdt, (22) Theexactsolutionsinthosecasesaregivenby

˜u(x)=(g(r)xn,g(r)xn), (23)

and

˜u(x)=(g(r)xn/2,g(r)xn/2), (24)

respectively.Forc=−1,inthosecasestheserieswillconvergetotheexactsolutions.

Proof. WeconsiderEq.(21),forn=1,wehave 4 3g(r)x 3/2_,4 3g(r)x 3/2 = x 0 ˜u(t) √ x−tdt , (25)

theexactsolutioninthiscaseisgivenby

˜u(x)=(g(r)x,g(r)x). (26) Bysubstitutingc=−1in(20),wehave u₁(x,r)=f(x,r) 2√x = 4/3g(r)x3/2 2√x = 2 3g(r)x, u₂(x,r)=2 3g(r)x− 1 2√x 2 3g(r) ⎛ ⎝4 3x 3 2 ⎞ ⎠ = 2 3g(r) 1−2 3 x, u₃(x,r)=2 3g(r) 1−2 3 2 x, .. . u_n(x,r)= 2 3g(r) 1−2 3 n x. (27) Then u(x,r)= ∞ n=0 2 3g(r) 1−2 3 n x=g(r)x. Similarly,wehave u1(x,r)= f(x,r) 2√x = 4/3g(r)x 3 2 2√x = 2 3g(r)x, u2(x,r)= 2 3g(r)x− 1 2√x 2 3g(r) 4 3x 3/2 = 2 3g(r) 1−2 3 x, u3(x,r)= 2 3g(r) 1−2 3 2 x, .. . un(x,r)= 2 3g(r) 1−2 3 n x. (28)

(8)

Then u(x,r)= ∞ n=0 2 3g(r) 1−2 3 n x=g(r)x.

Whichistheexactsolutionis˜u(x)=(g(r)x,g(r)x).Now,weassumethatEq.(21)istrueforn=m−1.Weprovethe relationsforn=m.Weconsiderthefollowingequation

2m+1_Γ_(m₊₁₎ 1×3×5···×(2m+1)g(r)x m+(1/2)_, 2m+1Γ(m+1) 1×3×5···×(2m+1)g(r)x m+(1/2) = x 0 u(t,r) √ x−tdt, (29)

whichistheexactsolution

˜u(x)=(g(r)xm,g(r)xm), (30)

wheremisanintegernumber.Usingc=−1inEq.(20)wehave

u₁(x,r)= 2 m_Γ_(m₊₁₎ 1×3×5×···×(2m+1)g(r)x m_, u₂(x,r)= 2 m_Γ_(m₊₁₎ 1×3×5×···×(2m+1)g(r) 1− 2 m_Γ_(m₊₁₎ 1×3×5×···×(2m+1) 2 xm, u₃(x,r)= 2 m_Γ_(m₊₁₎ 1×3×5×···×(2m+1)g(r) 1− 2 m_Γ_(m₊₁₎ 1×3×5×···×(2m+1) 3 xm, .. . u_n(x,r)= 2 m_Γ_(m₊₁₎ 1×3×5×···×(2m+1)g(r) 1− 2 m_Γ_(m₊₁₎ 1×3×5×···×(2m+1) n xm. (31) Then u(x,r)= ∞ n=0 2mΓ(m+1) 1×3×5×···×(2m+1)g(r) 1− 2 m_Γ_(m₊₁₎ 1×3×5×···×(2m+1) n xm=g(r)xm. Similarly,wehave u1(x,r)= 2m_Γ_(m₊₁₎ 1×3×5×···×(2m+1)g(r)x m_, u2(x,r)= 2m_Γ_(m₊₁₎ 1×3×5×···×(2m+1)g(r) 1− 2 m_Γ_(m₊₁₎ 1×3×5×···×(2m+1) 2 xm, u3(x,r)= 2mΓ(m+1) 1×3×5×···×(2m+1)g(r) 1− 2 m_Γ_(m₊₁₎ 1×3×5×···×(2m+1) 3 xm, .. . un(x,r)= 2mΓ(m+1) 1×3×5×···×(2m+1)g(r) 1− 2 m_Γ_(m₊₁₎ 1×3×5×···×(2m+1) n xm. (32) Then u(x,r)= ∞ n=0 2m_Γ_(m₊₁₎ 1×3×5×···×(2m+1)g(r) 1− 2 m_Γ_(m₊₁₎ 1×3×5×···×(2m+1) n xm=g(r)xm.

(9)

5. Numericalexamples

Here,we considerthree examples to illustrate the homotopy analysis method for solving Abel fuzzy integral equations.

Example1. ConsiderthefollowingAbelfuzzyintegralequation

4 3rx (3/2)_,4 3(2−r)x (3/2) = x 0 ˜u(t) √ x−tdt.

Theexactsolutioninthiscaseisgivenby ˜u(x)=(rx,(2−r)x) and 0≤r≤1. Bysubstitutingc=−1inEq.(20) u₁(x,r)=2 3rx, u₂(x,r)=2 3r 1−2 3 x, u₃(x,r)=2 3r 1−2 3 2 x, .. . u_n(x,r)= 2 3r 1−2 3 n x, andalso u1(x,r)= 2 3(2−r)x, u2(x,r)= 2 3(2−r) 1−2 3 x, u3(x,r)= 2 3(2−r) 1−2 3 2 x, .. . un(x,r)= 2 3(2−r) 1−2 3 n x. Thus, (u(x,r),u(x,r))= _∞ n=0 2 3r 1−2 3 n x, ∞ n=0 2 3(2−r) 1−2 3 n x ,

wheretheabovesummationyieldstotheexactsolution ˜u(x)=(rx,(2−r)x).

Example2. ConsiderthefollowingAbelfuzzyintegralequation

5 16(r 2₊_r_)πx3_, 5 16(4−r 3₋_r_)πx3 = x 0 ˜u(t) √ x−tdt.

(10)

Theexactsolutioninthiscaseisgivenby ˜u(x)=((r2+r)x(5/2),(4−r3−r)x(5/2)) and 0≤r≤1. Bysubstitutingc=−1inEq.(20) u₁(x,r)= 5 32(r 2₊_r )πx(5/2), u₂(x,r)= 5 32(r 2₊_r ) 1− 5 32π πx(5/2), u₃(x,r)= 5 32(r 2₊_r₎ 1− 5 32π 2 πx(5/2), .. . u_n(x,r)= 5 32(r 2₊_r ) 1− 5 32π n πx(5/2), andalso u1(x,r)= 5 32(4−r 3₋_r )πx(5/2), u2(x,r)= 5 32(4−r 3₋_r ) 1− 5 32π πx(5/2), u3(x,r)= 5 32(4−r 3₋_r₎ 1− 5 32π 2 πx(5/2), .. . un(x,r)= 5 32(4−r 3₋_r ) 1− 5 32π n πx(5/2). Thus, ˜u(x)= _∞ n=0 5 32(r 2₊_r ) 1− 5 32π n πx(5/2), ∞ n=0 5 32(4−r 3₋_r ) 1− 5 32π n πx(5/2) ,

wheretheabovesummationyieldstotheexactsolution ˜u(x)=((r2+r)x(5/2),(4−r3−r)x(5/2)).

Example3. ConsiderthefollowingnonlinearAbelfuzzyintegralequation

2043 3003(r 2₊ 2r)3x(13/2),2043 3003(6−3r 3 )3x(13/2) = x 0 ˜u3(t) √ x−tdt. (33)

Theexactsolutioninthiscaseisgivenby

˜u(x)=((r2+2r)x2,(6−3r3)x2) and 0≤r≤1. Thetransformation

˜v(x)=˜u3(x), ˜u(x)=3

˜v(x), (34)

carriesEq.(34)into 2043 3003(r 2₊ 2r)x(13/2),2043 3003(6−3r 3 )x(13/2) = x 0 ˜v(t) √ x−tdt. (35)

(11)

SubstitutingEq.(35)inEq.(20)gives: v₁(x,r)=1024 3003(r 2₊ 2r)3x6, v₂(x,r)=1024 3003(r 2₊_2r)3 1−1024 3003 x6, v₃(x,r)=1024 3003(r 2₊ 2r)3 1−1024 3003 2 x6, .. . v_n(x,r)= 1024 3003(r 2₊ 2r)3 1−1024 3003 n x6, andalso v1(x,r)= 1024 3003(6−3r 3 )3x6, v2(x,r)= 1024 3003(6−3r 3₎3 1−1024 3003 x6, v3(x,r)= 1024 3003(6−3r 3 )3 1−1024 3003 2 x6, .. . vn(x,r)= 1024 3003(6−3r 3 )3 1−1024 3003 n x6. Thus, ˜v(x)=(v(x,r),v(x,r))= _∞ n=0 1024 3003(r 2₊_2r)3 1−1024 3003 n x6, ∞ n=0 1024 3003(6−3r 3₎3 1−1024 3003 n x6 ,

wheretheabovesummationyieldstotheexactsolution ˜v(x)=((r2+2r)3x6,(6−3r3)3x6).

Then,byusingEq.(34),wehave: ˜u(x)=((r2+2r)x2,(6−3r3)x2).

6. Conclusion

Inthispaper,linearandnonlinearAbelfuzzyintegralequationswereconvertedintotwocrisplinearandnonlinear Abelintegralequationsbasedontheembeddingmethod.Then,weappliedhomotopyanalysismethodtoobtainthe uniquesolutionofAbelfuzzyintegralequations.Itwasshownthatthisnewtechniqueiseasytoimplementandproduces accurateresults.Aconsiderableadvantageofthemethodisthattheapproximatesolutionsarefoundveryeasilyby usingcomputerprogramssuchasMatlab.Themethodcanalsobeextendedtothesystemoflinearintegro-differential equationswithvariablecoefficients,butsomemodificationsareneeded.

Acknowledgement

(12)

References

[1]D.Dubois,H.Prade,Towardsfuzzydifferentialcalculus,FuzzySetsSyst.8(1982)1–7. [2]R.Goetschel,W.Voxman,Elementaryfuzzycalculus,FuzzySetsSyst.18(1986)31–43. [3]O.Kaleva,Fuzzydifferentialequations,FuzzySetsSyst.24(1987)301–317.

[4]S.Nanda,Onintegrationoffuzzymappings,FuzzySetsSyst.32(1989)95–101.

[5]C.Wu,M.Ma,Ontheintegralsseriesandintegralequationsoffuzzyset-valuedfunctions,J.HarbinInst.Technol.21(1990)11–19. [6]F.Mirzaee,M.Paripour,M.KomakYari,NumericalsolutionofFredholmfuzzyintegralequationsofthesecondkindviadirectmethodusing

triangularfunction,J.Hypertens.1(2)(2012)46–60.

[7]H.T.Nguyen,Anoteontheextensionprincipleforfuzzysets,J.Math.Anal.Appl.64(1978)369–380.

[8]S.J.Liao,ProposedHomotopyAnalysisTechniquefortheSolutionofNonlinearProblems,ShanghaiJiaoTongUniversity,1992(Ph.D. dissertation).

[9]S.J.Liao,BeyondPerturbation:AnIntroductiontoHomotopyAnalysisMethod,ChapmanHall/CRCPress,BocaRaton,2003.

[10]S.Abbasbandy,TheapplicationofhomotopyanalysismethodtosolveageneralizedHirotaSatsumacoupledKdVequation,Phys.Lett.A361 (2007)478–483.

[11]S.Abbasbandy,Theapplicationofhomotopyanalysismethodtononlinearequationsarisinginheattransfer,Phys.Lett.A360(2006)109–113. [12]V.Singh,R.Pandey,O.Singh,NewstablenumericalsolutionsofsingularintegralequationsofAbeltypebyusingnormalizedBernstein

polynomials,Appl.Math.Sci.3(2009)241–255.

[13]F.Mirzaee,NumericalsolutionofFredholmfuzzyintegralequationsofthesecondkindusinghybridofblock-pulsefunctionsandTaylor series,AinShamsEng.J.5(2014)631–636.

[14]F.Mirzaee,M.Paripour,M.KomakYari,ApplicationoftriangularanddeltabasisfunctionstosolvelinearFredholmfuzzyintegralequation ofthesecondkind,Arab.J.Sci.Eng.39(2014)3969–3978.

[15]F.Mirzaee,M.Paripour,M.KomakYari,ApplicationofhatfunctionstosolvelinearFredholmfuzzyintegralequationofthesecondkind,J. Intell.FuzzySyst.(2014)(inpress).

[16]M.Ma,M.Friedman,A.Kandel,Dualityinfuzzylinearsystems,FuzzySetsSyst.109(2000)55–58.

[17]S.Gal,Approximationtheoryinfuzzysetting,in:HandbookofAnalytic-computationalMethodsinAppliedMathematics,Chapman/CRC, BocaRaton,NewYork,2000.

[18]G.A.Anastassiou,FuzzyMathematics:ApproximationTheory,Springer,Heidelberg,2010. [19]H.Wu,ThefuzzyRiemannintegralanditsnumericalintegration,FuzzySetsSyst.110(2000)1–25. [20]Wu.Hc,TheimproperfuzzyRiemannintegralanditsnumericalintegration,Inform.Sci.11(1999)109–137. [21]A.D.Polyanin,A.V.Manzhirov,HandbookofIntegralEquations,CRCPress,BocaRaton,1998.