ebook img

Connected quandles associated with pointed abelian groups PDF

33 Pages·2013·0.46 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Connected quandles associated with pointed abelian groups

Pacific Journal of Mathematics CONNECTED QUANDLES ASSOCIATED WITH POINTED ABELIAN GROUPS W. EDWIN CLARK, MOHAMED ELHAMDADI, XIANG-DONG HOU, MASAHICO SAITO AND TIMOTHY YEATMAN Volume264 No.1 July2013 PACIFICJOURNALOFMATHEMATICS Vol.264,No.1,2013 dx.doi.org/10.2140/pjm.2013.264.31 CONNECTED QUANDLES ASSOCIATED WITH POINTED ABELIAN GROUPS W. EDWIN CLARK, MOHAMED ELHAMDADI, XIANG-DONG HOU, MASAHICO SAITO AND TIMOTHY YEATMAN A quandle is a self-distributive algebraic structure that appears in quasi- groupandknottheories. Foreachabeliangroup Aand c∈ A,wedefinea quandleG(A,c)on(cid:90) ×A. Thesequandlesaregeneralizationsofaclassof 3 nonmedialLatinquandlesdefinedbyV.M.Galkin,sowecallthemGalkin quandles. Each G(A,c) is connected but not Latin unless A has odd or- der. G(A,c) is nonmedial unless 3A = 0. We classify their isomorphism classes in terms of pointed abelian groups and study their various proper- ties. AfamilyofsymmetricconnectedquandlesisconstructedfromGalkin quandles, and some aspects of knot colorings by Galkin quandles are also discussed. 1. Introduction Sets with certain self-distributive operations called quandles have been studied since the 1940s in various areas. They have been studied, for example, as an algebraicsystemforsymmetries[Takasaki1943],asquasigroups[Galkin1988], andinrelationtomodules[Nelson2003]. Thefundamentalquandlewasdefined inamannersimilarto thefundamentalgroup[Joyce1982;Matveev1982], which madequandlesanimportanttoolinknottheory. Algebraichomologytheoriesfor quandlesweredefined [Carteretal.2003b; Fennetal. 1995]anddevelopedand investigated([LitherlandandNelson2003;Mochizuki2011;Niebrzydowskiand Przytycki2009;2011;Nosaka2011],forexample),andextensionsofquandlesby cocycleshavebeenstudied[AndruskiewitschandGraña2003;Carteretal.2003a; Eisermann2007b]andappliedtovariouspropertiesofknotsandknottedsurfaces (see[Carteretal.2004]andreferencestherein). Beforealgebraictheoriesofextensionsweredeveloped, Galkin[1988]defineda familyofquandlesthatareextensionsofthe3-elementconnectedquandle R ,and 3 wecallthemGalkinquandles. EventhoughthedefinitionofGalkinquandlesisa M.S.wassupportedinpartbyNSFgrantDMS0900671. MSC2010: 57M25. Keywords: quandles,pointedabeliangroups,knotcolorings. 31 32 CLARK,ELHAMDADI,HOU,SAITOANDYEATMAN specialcaseofacocycleextensiondescribedin[AndruskiewitschandGraña2003], theyhavecuriouspropertiessuchastheexplicitandsimpledefiningformula,close connectionsto dihedralquandles,and thefactthattheyappearinthe listofsmall connectedquandles. Inthispaper,wegeneralizeGalkin’sdefinitionanddefineafamilyofquandles thatareextensionsof R ,characterizetheirisomorphismclasses,andstudytheir 3 properties. The definition is given in Section 3 after a brief review of necessary materialsinSection2. Isomorphismclassesarecharacterizedbypointedabelian groupsinSection4. VariousalgebraicpropertiesofGalkinquandlesareinvestigated inSection5,andtheirknotcoloringsarestudiedinSection6. 2. Preliminaries In thissection we briefly review somedefinitions and examplesof quandles. More details can be found, for example, in [Andruskiewitsch and Graña 2003; Carter etal.2004;Fennetal.1995]. A quandle X is a set with a binary operation (a,b) (cid:55)→ a ∗b satisfying the followingconditions. (1) (Idempotency) Foranya∈ X,a∗a=a. (2) (Invertibility) Foranyb,c∈ X, thereisauniquea∈ X suchthata∗b=c. (3) (Rightself-distributivity) Foranya,b,c∈ X, wehave(a∗b)∗c=(a∗c)∗(b∗c). Aquandlehomomorphismbetweentwoquandles X,Y isamap f : X →Y such that f(x∗ y)= f(x)∗ f(y),where∗ and∗ denotethequandleoperationsof X X Y X Y andY,respectively. Aquandleisomorphismisabijectivequandlehomomorphism, andtwoquandlesareisomorphicifthereisaquandleisomorphismbetweenthem. Typicalexamplesofquandlesincludethefollowing. • Anynonemptyset X withtheoperationx∗y=x foranyx,y∈X isaquandle calledthetrivialquandle. • Agroup X =G withtheoperationofn-foldconjugation,a∗b=b−nabn,isa quandle. • Letn beapositiveinteger. Fora,b∈(cid:90) (integersmodulon),define n a∗b≡2b−a (modn). Then ∗ defines a quandle structure called the dihedralquandle R . This set n canbeidentifiedwiththesetofreflectionsofaregularn-gonwithconjugation asthequandleoperation. CONNECTEDQUANDLESASSOCIATEDWITHPOINTEDABELIANGROUPS 33 • Any(cid:90)[T,T−1]-module M isaquandlewitha∗b=Ta+(1−T)bfora,b∈M. ThisiscalledanAlexanderquandle. AnAlexanderquandleisalsoregarded asapair(M,T),where M isanabeliangroupand T ∈Aut(M). Let X beaquandle. TherighttranslationR : X → X bya∈ X isdefinedby a R (x)=x∗aforx∈X. Similarly,thelefttranslationL isdefinedbyL (x)=a∗x. a a a ThenR isapermutationof X byAxiom(2). ThesubgroupofSym(X)generated a bythepermutationsR ,a∈ X,iscalledtheinnerautomorphismgroupof X and a isdenotedbyInn(X). Welistsomedefinitionsofcommonlyknownpropertiesof quandlesbelow. • AquandleisconnectedifInn(X)actstransitivelyon X. • ALatinquandleisaquandlesuchthatforeacha∈ X,thelefttranslationL a isabijection. Thatis,themultiplicationtableofthequandleisaLatinsquare. • Aquandleisfaithfulifthemappinga(cid:55)→R isaninjectionfrom X toInn(X). a • A quandle X is involutory, or a kei, if the right translations are involutions: R2 =idforalla∈ X. a • Theoperation∗¯ on X definedbya ∗¯ b=R−1(a)isaquandleoperation,and b (X,∗¯)iscalledthedualquandleof(X,∗). If(X,∗¯)isisomorphicto(X,∗), then(X,∗)iscalledself-dual. • Aquandle X ismedialif(a∗b)∗(c∗d)=(a∗c)∗(b∗d)foralla,b,c,d∈X. It is also called abelian. It is known and easily seen that every Alexander quandleismedial. Acoloringofanorientedknotdiagrambyaquandle X isamapC:A→X from thesetofarcsAofthediagramto X suchthattheimageofthemapsatisfiesthe relationdepictedinFigure1ateachcrossing. Moredetailscanbefoundin[Carter et al. 2004; Eisermann 2007a], for example. A coloring that assigns the same elementof X forallthearcsiscalledtrivial,andotherwisenontrivial. Thenumber ofcoloringsofaknotdiagrambyafinitequandleisknowntobeindependentofthe choiceofadiagram,andhenceisaknotinvariant. Acoloringbyadihedralquandle R for a positive integer n > 1 is called an n-coloring. If a knot is nontrivially n colored by a dihedral quandle R for a positive integer n > 1, then it is called n n-colorable. InFigure2,anontrivial3-coloringofthetrefoilknot(3 inacommon 1 notationinaknottable[ChaandLivingston2011])isindicated. Thisispresented y x x∗y Figure1. Acoloringruleatacrossing. 34 CLARK,ELHAMDADI,HOU,SAITOANDYEATMAN 0 1 2 0 1 Figure2. Trefoilastheclosureofσ3. 1 inaclosedbraidform. Eachcrossingcorrespondstoastandardgeneratorσ ofthe 1 2-strand braid group, and σ3 represents three crossings together as in the figure. 1 Thedottedlineindicatestheclosure;see[Rolfsen1976]formoredetailsofbraids. Thefundamentalquandleisdefinedinamannersimilartothefundamentalgroup [Joyce1982;Matveev1982]. Apresentationofaquandleisdefinedinamanner similartogroupsaswell,andapresentationofthefundamentalquandleisobtained from a knot diagram (see, for example, [Fenn and Rourke 1992]), by assigning generators to arcs of a knot diagram, and relations corresponding to crossings. The set of colorings of a knot diagram K by a quandle X is then in one-to-one correspondence with the set of quandle homomorphisms from the fundamental quandleof K to X. 3. DefinitionandnotationforGalkinquandles Let A beanabeliangroup,alsoregardednaturallyasa(cid:90)-module. Letµ:(cid:90) →(cid:90), 3 τ :(cid:90) → A be functions. These functions µ and τ need not be homomorphisms. 3 Defineabinaryoperationon(cid:90) ×A by 3 (x,a)∗(y,b)=(cid:0)2y−x,−a+µ(x−y)b+τ(x−y)(cid:1), x,y∈(cid:90) , a,b∈ A. 3 Proposition 3.1. For any abelian group A, the operation ∗ defines a quandle structureon(cid:90) ×Aifµ(0)=2,µ(1)=µ(2)=−1,andτ(0)=0. 3 Galkin[1988,p.950]gavethisdefinitionfor A=(cid:90) . Thepropositiongeneralizes p hisresulttoanyabeliangroup A. Fortheproof,weexaminetheaxioms. Lemma3.2. (A) Theoperationisidempotent—thatis,itsatisfiesAxiom(1)—if andonlyif(µ(0)−2)a=0foranya∈ A,andτ(0)=0. (B) Theoperationasarightactionisinvertible—thatis,itsatisfiesAxiom(2). Proof. Directcalculations. (cid:3) CONNECTEDQUANDLESASSOCIATEDWITHPOINTEDABELIANGROUPS 35 Lemma3.3. Theoperation∗on(cid:90) ×Aisrightself-distributive—thatis,itsatisfies 3 Axiom(3)—ifandonlyifµ,τ satisfythefollowingconditionsforany X,Y ∈(cid:90) 3 andb,c∈ A: (4) µ(−X)b=µ(X)b, (5) (cid:0)µ(X+Y)+µ(X−Y)(cid:1)c=(µ(X)µ(Y))c, (6) τ(X+Y)+τ(Y −X)=τ(X)+τ(−X)+µ(X)τ(Y). Proof. Rightself-distributivity,thatis, ((x,a)∗(y,b))∗(z,c)=((x,a)∗(z,c))∗((y,b)∗(z,c)) for x,y,z∈(cid:90) anda,b,c∈ A,issatisfiedifandonlyif 3 µ(x−y)b=µ(y−x)b, µ(2y−x−z)c=(cid:0)−µ(x−z)+µ(y−x)µ(y−z)(cid:1)c, −τ(x−y)+τ(2y−x−z)=−τ(x−z)+µ(y−x)τ(y−z)+τ(y−x). Thisisseenbyequatingthecoefficientsofb andc andtheconstantterm. Forthe equivalence of the first equation with (4), set X =x−y. For the equivalence of thesecondwith(5),set X = y−x and Y =z−y. Fortheequivalenceofthelast with(6),set X = y−x andY = y−z. (cid:3) ProofofProposition3.1. Assumetheconditionsstated. ByLemma3.2,Axioms(1) and (2) are satisfied under the specifications µ(0) = 2,µ(1) = µ(2) = −1, and τ(0)=0. If X =0orY =0,then(5)(togetherwith(4))becomesatautology. If X−Y =0 or X +Y =0, then (5) reduces to µ(2X)+2=µ(X)2, which is satisfied by the above specifications. For R , if X +Y (cid:54)=0 and X −Y (cid:54)=0, then either X =0 or 3 Y =0. Hence (5) is satisfied. For (6), it is checked similarly, for the two cases [X =0orY =0]and[X−Y =0or X+Y =0]. (cid:3) Definition3.4. Let A beanabeliangroup. Thequandledefinedby∗on(cid:90) ×A by 3 Proposition3.1, (x,a)∗(y,b)=(cid:0)2y−x,−a+µ(x−y)b+τ(x−y)(cid:1), x,y∈(cid:90) , a,b∈ A, 3 withµ(0)=2,µ(1)=µ(2)=−1,andτ(0)=0,iscalledtheGalkinquandleand denotedby G(A,τ). Sinceτ isspecifiedbythevaluesτ(1)=c andτ(2)=c wherec ,c ∈ A,we 1 2 1 2 alsodenoteitby G(A,c ,c ). 1 2 Example3.5. TheGalkinquandle G((cid:90) ,0,1)is(cid:90) ×(cid:90) asasetwiththequandle 2 3 2 operationdefinedasabovewithµ(0)=2,µ(1)=µ(2)=−1,τ(0)=τ(1)=0,and 36 CLARK,ELHAMDADI,HOU,SAITOANDYEATMAN τ(2)=1. Thus,(0,1)∗(1,0)=(2,−1+µ(2)0+τ(2))=(2,0)and(2,0)∗(1,1)= (0,0+µ(1)1+τ(1))=(0,1),forexample. Lemma 3.6. For any abelian group A and c ,c ∈ A, the quandles G(A,c ,c ) 1 2 1 2 andG(A,0,c −c )areisomorphic. 2 1 Proof. Letc=c −c . Defineη:G(A,c ,c )→G(A,0,c),asamapon(cid:90) ×A,by 2 1 1 2 3 η(x,a)=(x,a+β(x))whereβ(0)=β(1)=0andβ(2)=−c . Thisηisabijection, 1 and we show that it is a quandle homomorphism. We compute η((x,a)∗(y,b)) andη(x,a)∗η(y,b)for x,y∈(cid:90) anda,b∈ A. 3 Ifx=y,thenµ(x−y)=2andτ(x−y)=0forbothG(A,c ,c )andG(A,0,c), 1 2 sothat η((x,a)∗(x,b))=η(x,2b−a)=(x,2b−a+β(x)), η(x,a)∗η(x,b)=(x,a+β(x))∗(x,b+β(x))=(cid:0)x,2(b+β(x))−(a+β(x))(cid:1) =(x,2b−a+β(x)), asdesired. If x−y=1∈(cid:90) ,thenµ(x−y)=−1forbothG(A,c ,c )andG(A,0,c)and 3 1 2 τ(x−y)=c for G(A,c ,c )butτ(x−y)=0for G(A,0,c),sothat 1 1 2 η((x,a)∗(y,b))=η(2y−x,−a−b+c )=(2y−x,−a−b+c +β(2y−x)), 1 1 η(x,a)∗η(y,b)=(x,a+β(x))∗(y,b+β(y)) =(cid:0)2y−x, −(a+β(x))−(b+β(y))(cid:1). Thetwoexpressionsareequalifandonlyifβ(x)+β(y)+β(2y−x)=−c ,which 1 istruesince x (cid:54)= y impliesthatexactlyoneof x,y,2y−x is2∈(cid:90) . 3 If x−y=2∈(cid:90) ,thenµ(x−y)=−1forbothG(A,c ,c )andG(A,0,c)and 3 1 2 τ(x−y)=c for G(A,c ,c )butτ(x−y)=c −c =c for G(A,0,c),sothat 2 1 2 2 1 η((x,a)∗(y,b))=η(2y−x,−a−b+c )=(2y−x,−a−b+c +β(2y−x)), 2 2 η(x,a)∗η(y,b)=(x,a+β(x))∗(y,b+β(y)) =(cid:0)2y−x,−(a+β(x))−(b+β(y))+c −c (cid:1) 2 1 =(cid:0)2y−x,−a−b−β(x)−β(y)+(c −c )(cid:1), 2 1 andagaintheseareequalforthesamereasonasabove. (cid:3) Notation. Since,byLemma3.6,anyGalkinquandleisisomorphicto G(A,0,c) foranabeliangroup A andc∈ A,wedenote G(A,0,c)by G(A,c)forshort. Anyfiniteabeliangroupisaproduct(cid:90) ×···×(cid:90) ,wherethepositiveintegers n1 nk nj satisfynj|nj+1 for j=1,...,k−1. Inthiscase,anyelementc∈Aiswrittenin avectorform[c ,...,c ],wherec ∈(cid:90) . ThenthecorrespondingGalkinquandle 1 k j nj isdenotedby G(A,[c ,...,c ]). 1 k CONNECTEDQUANDLESASSOCIATEDWITHPOINTEDABELIANGROUPS 37 Remark3.7. WenotethatthedefinitionofGalkinquandlesinducesafunctor. Let Ab denote the category of pointed abelian groups; its objects are pairs (A,c), 0 where A is an abelian group and c ∈ A, and its morphisms f :(A,c)→(B,d) aregrouphomomorphisms f : A→ B suchthat f(c)=d. LetQbethecategory of quandles consisting of quandles as objects and quandle homomorphisms as morphisms. F Thenthecorrespondence(A,c)(cid:55)→G(A,c)definesafunctorF:Ab →Q. It 0 iseasytoverifythatifamorphism f :(A,c)→(B,d)isgiven,thenthemapping F(f)(x,a)=(x, f(a))with(x,a)∈G(A,c)=(cid:90) ×A isahomomorphismfrom 3 G(A,c)to G(B,d)andsatisfiesF(gf)=F(g)F(f)andF(id(A,c))=idG(A,c). Remark3.8. AreaderwillwondertowhatextentDefinition3.4ofaGalkinquandle canbegeneralized. Wetriedseveralgeneralizations. Forexample,ifoneattempts toreplace3byanarbitraryprime p inDefinition3.4,thenLemma3.3stillholds. Inthiscasefor p>3,weproveinLemma5.14thatµ(x)=2forall x ∈(cid:90) ,and p computer experiments indicate that one almost always obtains a quandle if and onlyifτ =0,inwhichcasethequandleobtainedissimplyaproductofdihedral quandles. We have also attempted to replace −x+2y by the Alexander quandle operationtx+(1−t)y inboththeleftandrightcoordinates,buthaveneitherbeen successfulinfindinginterestingnewquandles,norbeenabletoprovethatnosuch generalizations exist. We note that if a generalization for p >3 exists, then any such quandles will be less dense than Galkin quandles, since multiples of 3 are morenumerousthanmultiplesof p when p>3. 4. Isomorphismclasses InthissectionweclassifyisomorphismclassesofGalkinquandles. Lemma4.1. Let Abeanabeliangroup,andleth:A→A(cid:48)beagroupisomorphism. ThenGalkinquandlesG(A,τ)andG(A(cid:48),hτ)areisomorphicasquandles. Proof. Define f : G(A,τ) → G(A(cid:48),hτ), as a map from (cid:90) × A to (cid:90) × A(cid:48), 3 3 by f(x,a) = (x,h(a)). This f is a bijection, and we show that it is a quandle homomorphismbycomputing f((x,a)∗(y,b))and f(x,a)∗ f(y,b)forx,y∈(cid:90) 3 anda,b∈ A: f((x,a)∗(y,b))= f(2y−x,−a+µ(x−y)b+τ(x−y)) =(cid:0)2y−x,h(−a+µ(x−y)b+τ(x−y))(cid:1), f(x,a)∗ f(y,b)=(x,h(a))∗(y,h(b)) =(cid:0)2y−x,−h(a)+µ(x−y)h(b)+hτ(x−y)(cid:1). Theequality f((x,a)∗(y,b))= f(x,a)∗ f(y,b)followsfromthefactsthat h is agrouphomomorphismandµ(x−y)isaninteger. (cid:3) 38 CLARK,ELHAMDADI,HOU,SAITOANDYEATMAN Lemma 4.2. Let c,d,n be positive integers. If gcd(c,n) = d, then G((cid:90) ,c) is n isomorphictoG((cid:90) ,d). n Proof. If A = (cid:90) , then Aut(A) = (cid:90)∗ = unitsof(cid:90) , and the divisors of n are n n n representativesoftheorbitsof(cid:90)∗ actingon(cid:90) . (cid:3) n n Thus we may choose the divisors of n for the values of c for representing isomorphismclassesof G((cid:90) ,c). n Corollary4.3. If Aisavectorspace(elementary p-group),thenthereareexactly twoisomorphismclassesofGalkinquandlesG(A,τ). Proof. If A isavectorspacecontainingnonzerovectorsc andc ,thenthereisa 1 2 nonsingularlineartransformation h of A suchthat h(c )=c . That G(A,0)isnot 1 2 isomorphicto G(A,c)ifc(cid:54)=0followsfromLemma4.5. (cid:3) For distinguishing isomorphismclasses, cycle structures of the right action are useful,andweusethefollowinglemmas. Lemma4.4. Foranyabeliangroup A,theGalkinquandleG(A,τ)isconnected. Proof. Recallthattheoperationisdefinedbytheformula (x,a)∗(y,b)=(cid:0)2y−x,−a+µ(x−y)b+τ(x−y)(cid:1), with µ(0)=2, µ(1)=µ(2)=−1, and τ(0)=0. If x (cid:54)= y, then (x,a)∗(y,b)= (2y−x,−a−b+c )=(z,c),wherei =1or2and x,y∈(cid:90) anda,b∈ A. Note i 3 that{x,y,2y−x}=(cid:90) if x (cid:54)=y. Inparticular,forany(x,a)and(z,c)with x (cid:54)=z, 3 thereis(y,b)suchthat(x,a)∗(y,b)=(z,c). Forany(x,a )and(x,a )wherex∈(cid:90) anda ,a ∈A,take(z,c)∈(cid:90) ×Asuch 1 2 3 1 2 3 thatz(cid:54)=x. Thenthereare(y,b ),(y,b )suchthatx(cid:54)=y(cid:54)=zand(x,a )∗(y,b )= 1 2 1 1 (z,c)and(z,c)∗(y,b )=(x,a ). Hence G(A,τ)isconnected. (cid:3) 2 2 Lemma4.5. Thecyclestructureofarighttranslationin G(A,τ),whereτ(0)= τ(1)=0andτ(2)=c,consistsof1-cycles,2-cycles,and2k-cycles,wherek isthe orderofcinthegroup A. Sinceisomorphicquandleshavethesamecyclestructureofrighttranslations, G(A,c)andG(A,c(cid:48))forc,c(cid:48)∈ Aarenotisomorphicunlesstheordersofcandc(cid:48) coincide. Proof. Let τ(0) = 0, τ(1) = 0, and τ(2) = c. Then by Lemma 4.4, the cycle structureofeachcolumnisthesameasthe cyclestructureoftherighttranslation by(0,0),thatis,ofthepermutation f(x,a)=(x,a)∗(0,0)=(−x,−a+τ(x)). Weshowthatthispermutationhascyclesoflengthonly1,2andtwicetheorder ofc in A. Since f(0,a)=(0,−a)fora∈ A,a(cid:54)=0,wehave f2(0,a)=(0,a),so that(0,a)generatesa2-cycle,ora1-cycleif2a=0. Nowfrom f(1,a)=(2,−a) and f(2,a)=(1,−a+c) fora∈ A, by induction it is easy to see that for k >0, CONNECTEDQUANDLESASSOCIATEDWITHPOINTEDABELIANGROUPS 39 f2k(1,a)=(1,a+kc)and f2k(2,a)=(2,a−kc). Inthecaseof(1,a),a(cid:54)=0,the cyclecloseswhena+kc=a in A. Thesmallestk forwhichthisholdsistheorder ofc,inwhichcasethecycleisoflength2k. Acyclebeginningat(2,a)similarly hasthissamelength. (cid:3) Proposition4.6. Letn beapositiveinteger. Let A=(cid:90) andc ,c(cid:48) ∈(cid:90) fori =1,2. n i i n TwoGalkinquandles G(A,c ,c )and G(A,c(cid:48),c(cid:48))areisomorphicifandonlyif 1 2 1 2 gcd(c −c ,n)=gcd(c(cid:48) −c(cid:48),n). 1 2 1 2 Proof. Ifgcd(c −c ,n)=gcd(c(cid:48) −c(cid:48),n),thentheyareisomorphicbyLemmas3.6 1 2 1 2 and4.2. Thecyclestructuresaredifferentifgcd(c −c ,n)(cid:54)=gcd(c(cid:48) −c(cid:48),n)by 1 2 1 2 Lemma4.5,andhencetheyarenotisomorphic. (cid:3) Remark 4.7. The cycle structure is not sufficient for noncyclic groups A. For example,let A=(cid:90) ×(cid:90) . Then G(A,[1,0])and G(A,[0,2])havethesamecycle 2 4 structure for right translations, with cycle lengths {2,2,4,4,4,4} in a multiset notation,yettheyareknownnottobeisomorphic. (InthenotationofExample4.12 below, G(A,[1,0])=C[24,29]and G(A,[0,2])=C[24,31].) Wenotethatthere isnoautomorphismof A carrying[1,0]to[0,2]. Moregenerally,theisomorphismclassesofGalkinquandlesarecharacterizedas follows. Theorem 4.8. Suppose A,A(cid:48) are finite abelian groups. Two Galkin quandles G(A,τ) and G(A(cid:48),τ(cid:48)) are isomorphic if and only if there exists a group isomor- phismh: A→ A(cid:48) suchthathτ =τ(cid:48). OneimplicationintheproofofTheorem4.8isLemma4.1. Fortheother,first weprovethefollowingtwolemmas. Wewilluseawellknowndescriptionofthe automorphismsofafiniteabeliangroup,whichcanbefoundin[HillarandRhea 2007;Ranum1907]. Lemma 4.9. Let A be a finite abelian p-group and let f : pA → pA be an automorphism. Then f canbeextendedtoanautomorphismof A. Proof. Let A=(cid:90)n1 ×···×(cid:90)nk. Then p1 pk       px px x 2 2 2 . . . (7) f  .. = P .. ,  .. ∈(cid:90)n2 ×···×(cid:90)nk,       p2 pk px px x k k k where  P P ··· P  22 23 2k  pP32 P33 ··· P3k (8) P = . . . ,  .. .. ..    pk−2P pk−3P ··· P k2 k3 kk

Description:
264, No. 1, 2013 dx.doi.org/10.2140/pjm.2013.264.31. CONNECTED QUANDLES ASSOCIATED. WITH POINTED ABELIAN GROUPS. W. EDWIN CLARK, MOHAMED ELHAMDADI, XIANG-DONG HOU,. MASAHICO SAITO AND TIMOTHY YEATMAN. A quandle is a self-distributive algebraic structure
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.