SOME ABSTRACT PROPERTIES OF SEMIGROUPS APPEARING IN SUPERCONFORMAL THEORIES

合集下载

Semigroup-controlled asymptotic dimension

Semigroup-controlled asymptotic dimension

a rX iv:mat h /68736v1[mat h.MG ]29Aug26SEMIGROUP-CONTROLLED ASYMPTOTIC DIMENSION J.HIGES Abstract.We introduce the idea of semigroup-controlled asymptotic dimension.This notion generalizes the asymptotic dimension and the asymptotic Assouad-Nagata dimension in the large scale.There are also semigroup controlled dimensions for the small scale and the global scale.Many basic properties of the asymptotic dimension theory are satisfied by a semigroup-controlled asymptotic dimension.We study how these new dimensions could help in the understanding of coarse embeddings and uniform embeddings.In particular we have introduced uncountable many invariants under quasi-isometries and uncountable many bi-Lipschitz invariants.Hurewicz type theorems are generalized and some applications to geometric group theory are shown.Contents 1.Introduction and preliminaries 12.Control semigroups 33.Semigroup-controlled dimensions:basic properties rge scale and small scale dimensions 75.Non equivalent semigroup-controlled dimensions 106.Maps between metric spaces and dimension 147.Hurewicz type theorems 17References 191.Introduction and preliminaries The aysmptotic dimension was introduced by Gromov in [8]as one im-portant coarse invariant in the study of geometric group theory.The idea behind its definition is to analyze a metric space as a large scale object.2J.HIGESAn analogous concept to the asymptotic dimension but for small scale ob-jects would be the uniform dimension introduced by Isbell in[10].These two dimensions suggest the idea of seeing a metric space as a global object and then we would get a global definition of dimension.In[4](see also[12]) it was given such definition and it was analyzed some properties of zero dimensional spaces.The following notions associated to a cover U={U s}s∈S of a metric space (X,d)are standard concepts in the theory of metric spaces.They are used in many equivalent definitions of asymptotic dimension.Let U={U s}s∈S be a cover of a metric space(X,d X),not necessarily open.Associated to this cover there is a family a natural family of functions {f s}s∈S with f s:X→R+defined by:f s(x):=d X(x,X\U s).With such functions we can define:-Local Lebesgue number L x(U)of U at x∈X:L x(U):=sup{f s(x)|S∈S}.-Global Lebesgue number L(U):=inf{L x(U)|x∈X}.-Local s-multiplicity s−m U(x)of U at x∈X is defined as the numberof elements of U that intersect B(x,s).-Global s-multiplicity s−m(U):=sup{s−m U(x)|x∈X}.If s=0then the0-multiplicity will be called multiplicity of U and it will benoted by m(U).Given a family of subsets U of a metric space(X,d X)it is said that U is C-bounded with C>0if diam(U)≤C for every U∈U.If U is C-bounded for some C>0it is said that U is uniformly bounded and if d X(U,V)>s for every U,V∈U,it is said also that U is s-disjoint with s>0.A definition of asymptotic dimension could be the following:Definition 1.1.We will say that a metric space(X,d)has asymptotic dimension at most n(notation:asdimX≤n)if there is an s0such that for every s≥s0there exists an uniformly cover(colored cover)U= n+1i=1U i so that each U i is s-disjoint.From this definition we can deduce that there exist a function f:R+→R+with lim x→∞f(x)=∞such that each s-disjoint n+1colored cover is f(s)-bounded.If we restricted the range of functions f allowed to the linear ones(or asymptotically linear ones)we would get the notion of asymptotic Assouad-Nagata dimension,also called linear controlled asymptotic dimen-sion or asymptotic dimension with Higson property.Many results have appeared recently associated to this dimension,see[3],[4],[6],[7]and[9], for example.Here is the definition:Definition 1.2.We will say that a metric space(X,d)has asymptotic Assouad-Nagata dimension at most n(notation:asdimX≤n)if there is a s0>0and a C>0such that for every s≥s0there exist a Cs-bounded cover(colored cover)U= n+1i=1U i so that each U i is s-disjoint.SEMIGROUP-CONTROLLED ASYMPTOTIC DIMENSION3 If we just change the condition of s≥s0by s≤s0we will get the notion of capacity dimension introduced by Buyalo in[5].The asymptotic Assouad-Nagata dimension and the capacity dimension are just the large scale and the small scale versions of the Assouad-Nagata dimension introduced by Assouad in[1].The Assouad-Nagata dimension is a bi-Lipschitz invariant. Lang and Schlichenmaier proved in[11]that the Assouad-Nagata dimension is in fact a quasisymmetry invariant.Many other interesting properties of the Assouad-Nagata dimension appeared in that work.The relations among the capacity dimension,the asymptotic Assouad-Nagata dimension and the Assouad-Nagata dimension were showed in[3]. In this paper we generalize all these notions using the concept of semigroup-controlled asymptotic dimension,semigroup-controlled small scale dimen-sion and semigroup-controlled global dimension.The idea is to modify the range of functions allowed to control the size of the colored covers in def-inition1.2.These functions will have a semigroup structure.In section2 we will study some basic properties of this kind of semigroups.In the next section we will define the notion of semigroup-controlled dimension and we will show that many basic properties of asymptotic dimension(see[2])are satisfied for these new dimensions.Section4is dedicated to relate the large scale,small scale and global dimensions following the same ideas of[3].In section5we will prove that this generalization is not trivial i.e.we have introduced uncountable many different dimensions.The last two sections are dedicated to important topics.In section6we will study the types of maps between metric spaces for which these new dimensions are invariant. In particular we will prove that all these dimensions are quasi-isometric in-variants in the large scale theory or bi-Lipschitz invariants in the global case. Last section generalizes the results of[9]about the Hurewicz type theorem. Some applications to geometric group theory are obtained as corollaries.2.Control semigroupsIn this work we will consider properties P(s)depending on positive real numbers s∈R+.We will say that P(s)is satisfied in a neighborhood of∞(respectively in a neighborhood of0)if there is a s0such that P(s)is true for every s≥s0(resp.s≤s0).Properties that are satisfied in a neighborhood of∞will be called large scale properties or asymptotic properties,properties that are satisfied in a neighborhood of0will be called small scale properties.If a property P(s) is satisfied in a neighborhood of∞and in a neighborhood of0we will say that P(s)is a global property.In this work we will usually give proofs and statements for the large scale case.The small scale case and the global case can be usually done using a dual reasoning and in many cases they will be left to the reader.Next definition describes the type of functions that are going to control the dimension.4J.HIGESDefinition2.1.Let f:R+→R+be an increasing continuous functiona.We will say that it is a large scale(or asymptotic)dim-control func-tion if f(x)≥x in a neighborhood of∞and f(∞)=∞i.e.lim x→∞f(x)=∞.b.If we require that f(x)≥x in a neighborhood of0and f(0)=0wewill say that f is a small scale dim-control function.c.If a function is a small scale dim-control function and a large scaledim-control function then we will say that it is a global dim-controlfunction.If a large scale dim-control function f is equal to some linear function in a neighborhood of∞we will say that it is a large scale(or asymptotically)lin-ear dim-control function.Analogously it can be defined the notion of small scale linear dim-control function and global linear dim-control function. Now we define the notion of large scale(or asymptotic)control semigroup. This notion is the main concept of this paper.Definition2.2.Let S be a set of asymptotic dim-control functions.We will say that S is a large scale(or asymptotic)control semigroup if the following properties are satisfied:(1)Every asymptotically linear dim-control function is in S(linear con-dition).(2)For every pair of functions g1,g2∈S we have g1◦g2∈S(semigroupcondition).We define the notion of small scale and global control semigroup for a set of small scale(resp.global)dim-control function analogously.We will note it byξ(resp.¯S)Remark2.3.In this work we will refer to asymptotic control semigroup just as control semigroups unless it were necessary to remark the large scale condition.Here are the main examples of control semigroups.Example2.4.The set of all asymptotic dim-control functions is a control semigroup.It will be called the uniform control semigroup and we will note it by U(u and¯U for the small scale and the global case respectively)The set of all asymptotically linear dim-control functions is a control semigroup.It will be called the Nagata(or the linear)control semigroup and it will be noted by N(n and¯N for the small scale and large scale case respectively).Remark2.5.Let K be a set of asymptotic dim-control functions.From the theory of semigroups we know that such set generates a control semigroup L(K)i.e.the intersection of all control semigroups S so that K⊂S. It will be called the control semigroup generated by K.Note that N= L({x}).We have also obtained many examples of control semigroups using this procedure.SEMIGROUP-CONTROLLED ASYMPTOTIC DIMENSION5 In the set of all control semigroups we can define a partial order.Definition2.6.Given two control semigroups S1and S2,we will say that S2isfiner than S1(notation:S1 S2)if for every asymptotic dim-control function f∈S2there is a dim-control function g∈S1such that f≤g in a neighborhood of∞.If S1 S2and S2 S1we will say that both control semigroups are equivalent and we will note it by S2≈S1.The following result is easy to check and it is left as an exercise. Proposition2.7.Let S be a control semigroup then:(1)U S N.(2)L(K∪S) S with K is any set of asymptotic dim-control functions. Remark 2.8.As a consequence of the second statement in the previous proposition we obtain that L(S2∪S1) S2and L(S2∪S1) S1for every pair of control semigroups.Then the set of all control semigroups with the relation is a directed set.Now we show an example of two different equivalent control semigroups. In the following section we could see how this example allows us to give another definition of asymptotic Assouad-Nagata dimension:Example2.9.Let{C i}n i=1be afinite set of constants with C i≥1.For eachfinite set of those constants we can define an asymptotic dim-controlfunction f{Ci}as a continuous piecewise linear function built with linearfunctions of slope C i such that f{Ci}(∞)=∞.It can be easily checkedthat the set of all functions of the form f{Ci}is a control semigroup.Wewill note it by PL.Clearly we have PL N.Take a f{Ci}∈PL.LetC be the maximum of the C i then the linear function f(x)=Cx satisfiesf(x)≥f{Ci}(x)for all x in a neighborhood of∞.So we have proved thatN L and both semigroups are equivalent.The following is an easy consequence of the semigroup theory and it will be used later.Proposition2.10.Let K be a set of dim-control functions.If g is a dim-control function g of L(K)then there exist afinite sequence of dim-control functions{f i}n i=1such that:g=f1◦f2◦...◦f nwhere each f i belongs to N∪K.3.Semigroup-controlled dimensions:basic properties Now,using the notion of control semigroup we can give the definition of semigroup-controlled asymptotic dimension.It generalizes the notions of Assouad-Nagata asymptotic dimension and asymptotic dimension.Definition3.1.Let S be a control semigroup.We will say that a metric space(X,d)has S-controlled asymptotic dimension at most n(notation:6J.HIGESasdim S X≤n)if there is an f∈S such that for every s in some neighbor-hood of∞there exist a cover(colored cover)U= n+1i=1U i so that each U i is s-disjoint and f(s)-bounded.A metric space is said to have S-controlled asymptotic dimension n if it has S-controlled asymptotic dimension at most n and for every k<n it does not happen that asdim S X≤k.Remark3.2. a.The function f∈S for which asdim S X≤n will be called(n,S)-dimensional control function of X.Such notion willbecome very important in the last section.b.The small scaleξ-controlled dimension will be noted by smdimξX.c.Special remark is needed for the definition of global dimension.Thestrictly analogous definition to the Assouad-Nagata dimension wouldbe the following:(X,d)has¯S-controlled global dimension at most n(notation:dim¯S X≤n)if there is a f∈¯S such that for every s there is a cover(colored cover)U= n+1i=1U i so that each U i is s-disjoint andf(s)-bounded.We will say that two control semigroups S2and S1are dim-equivalent(notation:asdim S2≡asdim S1)if for every metric space(X,d)we haveasdim S2X=asdim S1X.Next proposition and corollaries justify our definition of . Proposition3.3.Let S1and S2be two control semigroups.If S1 S2thenfor every metric space X we have asdim S1X≤asdim S2X.Proof.Given f∈S2take g∈S1such that g(s)≥f(s)in a neighborhood of ∞,then any f(s)-bounded family U of subsets of X is g(s)-bounded with s in a neighborhood of∞. Corollary3.4.Let S be a control semigroup.For every metric space X we have asdim U X≤asdim S X≤asdim N X.Corollary3.5.If two control semigroups are equivalent then they are dim-equivalent,i.e.they define the same asymptotic dimension.Example3.6.Note that dim¯U X and asdim U X are the uniform dimension defined in[4]and the asymptotic dimension of[8]respectively.We have also that asdim N X,smdim n X and dim¯N X are the asymptotic dimension with Higson property(see for example[13]),the capacity dimension(see[5])and the Assouad Nagata dimension(introduced in[1])respectively.Note that applying3.5to the example2.9we get another way of defining the Assouad-Nagata asymptotic dimension.The following proposition covers some basic properties of a semigroup-controlled asymptotic dimension.The proofs are highly similar to that ones for Assouad Nagata dimension and asymptotic Assouad Nagata dimension.SEMIGROUP-CONTROLLED ASYMPTOTIC DIMENSION7 We quote between brackets the works where the analogous proofs can be founded.Proposition3.7.Let(X,d X)and(Y,d Y)be metric spaces and S a control semigroup.Then it is satisfied:(1)asdim S A≤asdim S X for every A⊂X.(This is trivial)(2)asdim S X×Y≤asdim S X+asdim S Y([9]or[11])(3)If X=A∪B then asdim S X=max{asdim S A,asdim S B}([9]or[11]).(4)The following conditions are equivalent:([2])a.asdim S X≤n.b.There is a f∈S such that for every s in some neighborhood of∞there exist a cover U with s−m(U)≤n+1and f(s)-bounded.c.There is a g∈S such that for every s in some neighborhoodof∞there exist a cover U with m(U)≤n+1,L(U)≥s andg(s)-boundedd.There is a h∈S such that for everyǫin some neighborhoodof0there is a mapǫ-Lipschitz p:X→K n with K n a n-dimensional simplicial complex such that the family p−1(st v)ish(1/ǫ)-bounded.Remark 3.8.Note that in the proof of the third property it is used the semigroup condition.For the second and fourth properties it is necessary that given g1,g2∈S there exists a g3∈S such that g1+g2≤g3in a neighborhood of∞.Using the semigroup condition and the linear condition define g3=2·g1◦g2if g1(x)>x and g2(x)>x in a neighborhood of∞.rge scale and small scale dimensionsThe aim of this section is to study how the large scale,small scale and the global dimensions are related.Many of these results are based on[3]. Given a global control semigroup we can see it just as a large scale(or small scale)control semigroup.This is the idea behind next definition.Definition4.1.Let¯S be a global control semigroup we will define the large scale(resp.small scale)truncated semigroup of¯S as the semigroup of all functions g for which there exist a dim-control function f∈¯S with g(x)=f(x)in a neighborhood of∞(resp.in a neighborhood of0).We will note it by:T runc∗∗(¯S)(resp.T runc∗∗(¯S)).Now we present some kind of inverse operation of truncation.Given a small scale control semigroupξand a large scale control semigroup S we want to create a global control semigroup¯S associated to those ones.Definition4.2.We define the linked set ofξand S as the set of all contin-uous increasing functions g for which there exist a small scale dim-control function g1with g1∈ξand a large scale dim-control function g2with g2∈S such that g(x)=g1(x)in a neighborhood of0and g(x)=g2(x)in a neigh-borhood of∞.It will be noted by Link(ξ,S).8J.HIGESClearly T runc∗∗(¯S)and T runc∗∗(¯S)are small scale and large scale control semigroups respectively.Next proposition shows we have the same property for linked sets.Proposition4.3.Let Link(ξ,¯S)be a linked set then it is a global control semigroup.Proof.The linear condition is trivial.Now let f,g be two dim-control func-tions of Link(ξ,S)and let(f1,f2),(g1,g2)its small scale and large scale associated functions.we have that f(g(x))=f1(g1(x))in a neighborhood of0and f(g(x))=f2(g2(x))in a neighborhood of∞then as f1◦g1∈ξand f2◦g2∈S the semigroup condition is satisfied. The relationship between truncation and linking is given in the following result:Proposition4.4.Let¯S be a global control semigroup then:Link(T runc∗∗(¯S),T runc∗∗(¯S))≈¯S.Conversely T runc∗∗(Link(ξ,S))≈S and T runc∗∗(Link(ξ,S))≈ξ. Proof.Let P be the global control semigroup Link(T runc∗∗(¯S),T runc∗∗(¯S)). Clearly by¯S⊂P we have P ¯S.Now let g be a dim-control function in P.There exist two dim-control functions g1,g2∈¯S such that g(x)=g1(x)in a neighborhood of0and g(x)=g2(x)in a neighborhood of∞.Let x1≤x2be two positive numbers such that g(x)=g1(x)if x≤x1and g(x)=g2(x)if x≥x2.Let M be the maximum of g in[x1,x2]and let m be the minimum of g1◦g2in the same interval.Suppose m<M then there exist a C>1such that Cm≥M. Define the function g′(x)=C(g1(g2(x))).We have g′(x)≥g(x)for every x and then¯S P.The converse is obvious. Next definition is the key to connect large scale,small scale and global dimensions,see[3].Definition4.5.Let¯S be a global control semigroup and let X be any metric space.We will say that(X,d)has¯S-microscopic controlled dimension at most n and we will note by m−dim¯S X≤n if the metric space(X,d′1= min(d,1))has¯S-controlled dimension at most n.In a similar way we will say that a metric space(X,d)has¯S-macroscopic controlled dimension at most n(M−dim S X≤n)if dim¯S(X,d′′1)≤n with d′′=max(1,d). Lemma 4.6.Let(X,d)be a metric space and let¯S be a global control semigroup.If for every c>0we define the metrics d′c=min(c,d)and d′′c= max(c,d)then m−dim¯S X=dim¯S(X,d′c)and M−dim¯S X=dim¯S(X,d′′c). The proof of this result is given in[3].In such proof the authors used that the Assouad-Nagata dimension is invariant under Lipschitz functions. We give another proof without using this fact.SEMIGROUP-CONTROLLED ASYMPTOTIC DIMENSION9 Proof.We just do the proof for the microscopic case.The macroscopic case is similar.Suppose without loss of generality that c<1.Let s be any positive number.If s≥c then pick the cover U=X.Assume that s<c then there is a cover U with s-Lebesgue number in(X,d′1),multiplicity at most m−dim¯S X+1and f(s)-bounded for some dim-control function f∈¯S. But as d′1(x,y)=d′c(x,y)=d(x,y)if d(x,y)≤c then such cover satisfy L(U)≥s in(X,d′c)and it is f(s)-bounded.We have proved dim¯S(X,d′c)≤m−dim S X.The remaining case is similar. Corollary4.7.Let¯S1be a global control semigroup.If X is a boundedmetric space then m−dim¯S1X=dim¯S1X.If X is a discrete metric spacethen M−dim¯S1X=dim¯S1X.Next lemma shows how the microscopic dimension of a global control semigroup is greater or equal than the semigroup-controlled asymptotic di-mension associated to the large scale truncated semigroup.Lemma4.8.Let X be a metric space and let¯S be a global control semigroup then the following properties are equivalent:(1)M−dim¯S X≤n(2)There is a function f∈¯S such that for all s in a neighborhood of ∞there is a colored cover U= n+1i=1U i with each U i s-disjoint andf(s)-bounded.The proof is almost equal to the proof of Lemma2.7.of[3].It will be left to the reader.Using the same reasoning we can get the analogous Lemma for the mi-croscopic case:Lemma4.9.Let X be a metric space and¯S a global control semigroup then the following properties are equivalent:(1)m−dim¯S X≤n(2)There is a function f∈¯S such that for all s in a neighborhood of 0there is a colored cover U= n+1i=1U i with each U i s-disjoint andf(s)-bounded.Next lemma could be considered some kind of converse of the previous ones.Lemma4.10.Let¯S be a global control semigroup.For every metric space X we have dim¯S X≤n if and only if m−dim¯S X≤n and M−dim¯S X≤n. Proof.The necessary condition is obvious by lemma4.8.Let us prove the sufficient condition.Suppose m−dim¯S X≤n and M−dim¯S X≤n. Without loss of generality we can assume that the f associated to the bounds of the microscopic covers and the macroscopic covers is the same,if not take the composition.Let s be a positive real number.We want tofind a dim-control function g∈¯S and a colored covering U s-disjoint and g(s)-bounded. It is clear that if s>1or f(s)<1the result is obvious.Assume that s≤110J.HIGESand f(s)≥1.Pick s0=f−1(1)and define the function g(x)=f(2f(x)).So if s0≤s≤1take a colored covering U of(X,d′′1)so that it is2f(s)-disjoint and f(2(f(s)))-bounded. Combining all the results of this section we get that the global dimension can be obtained just studying the dimension in a neighborhood of0and in a neighborhood of∞.Theorem4.11.Let¯S be a global control semigroup.For every metric spaceX we have that smdim T runc∗∗(¯S)X=m−dim¯S X and asdim T runc∗∗(¯S)X=M−dim S X.In the other hand givenξa small scale control semigroup and S a largescale control semigroup then:dim Link(ξ,¯S)X=max{smdimξX,asdim S X}.Proof.By the two previous lemmas we get M−dim¯S X≥asdim T runc∗∗(¯S)X. Suppose asdim T runc∗∗(¯S)X≤n.Then for every s≥s0there is a cover U s= n+1i=1U s i that is s-disjoint and f(s)-bounded with f∈T runc∗∗(¯S). That means that there is a f1∈¯S such that f(x)=f1(x)if x≥x1for somex1.Let s′0be the maximum of x1and s0.We have that for every s≥s′0 there is a cover U s= n+1i=1U s i s-disjoint and f1(s)-bounded.Applying4.8 we get the result.The microscopic case is analogous.For the second statement just note:dim Link(ξ,S)X=max{m−dim Link(ξ,S)X,M−dim Link(ξ,S)X}=max{smdim T runc∗∗(Link(ξ,S))X,asdim T runc∗∗(Link(ξ,S))X}So the result follows from the second statement of4.4and3.3.5.Non equivalent semigroup-controlled dimensionsLet SMDIM)be the quotient set of all large scale(resp.small scale)control semigroups with the equivalence relation≡de-fined in section3.In this section we will estimate the cardinality of SMDIM.LetΩbe the set of all the countable ordinals(Ω0)union thefirst uncount-able ordinal.This set is uncountable and it has a natural well order.Wewill prove that there exist two order preserving maps i L:Ω→SMDIM.As usual we will prove the results for the large scalecase.The small scale case will be left to the reader.As afirst step we will estimate the cardinality of the set of all largescale(resp.small scale)control semigroups modulo the equivalence relation≈.They will be noted by ASDIM and SMDIM.The next lemmas are necessary to show that there exist at least countablemany semigroups.SEMIGROUP-CONTROLLED ASYMPTOTIC DIMENSION 11Lemma 5.1.Let g :[a i ,a i +1]→[g (a i ),g (a i +1)]be an increasing continuous function defined in some interval and let f (a i ),f (a i +1)be any pair of points that satisfies f (a i )≥g (a i ),f (a i +1)≥g (a i +1)then there is a continuous increasing function ¯f defined in the same interval such that ¯f ≥g with ¯f(a i )=f (a i )and ¯f (a i +1)=f (a i +1).Proof.Define the function f aux (x )=g (x )+f (a i )−g (a i ).This is an in-creasing continuous function in the interval.If f aux (a i +1)=g (a i +1)then f a ux =¯f.Otherwise let z be the greatest point in [a i ,a i +1]such that the segmenta i +1−z (x −z )+f aux (z )if x ∈(z,a i +1]It is clear that this function satisfies the requirements of the lemma. Lemma 5.2.Let f be a large scale dim-control function such that it is strictly bigger than the identity in a neighborhood of ∞.Then there exist a large scale dim-control function g such that for every n ∈N g (x )>f n (x )in some neighborhood of ∞.Proof.As the function is strictly bigger than the identity we have that there is an increasing sequence x n →∞such that if x ≥x n then f n (x )>f i (x )for all i ≤n −ing 5.1define in each interval [x n ,x n +1]a function g n so that g n (x n )=f n (x n ),g n (x n +1)=f n +1(x n +1)and g n (x )≥f n (x ).Paste all these functions and we get g the required function. Using a similar reasoning we can give the following lemma:Lemma 5.3.Let f be a small scale dim-control function such that it is strictly bigger than the identity in a neighborhood of 0.Then there exist a small scale dim-control function g such that for every n ∈N g (x )>f n (x )in some neighborhood of 0.Next lemma shows that there are at least countable many non equivalent large scale control semigroups.Lemma 5.4.There exist a sequence {S i }i ∈N of large scale control semi-groups with S i ≺S i −1and S i =L ({f i }∪S i −1)where f i is an asymptotic dim-control function such that for every g ∈S i −1f i (x )>g (x )in a neigh-borhood of ∞.Proof.Take S 1=N .Let f :R +→R +be the dim-control function defined by:f (x )= x 2if x ∈[1,∞)x otherwise.It is clear that for every asymptotically linear dim-control function g there is a point x 0∈R +such that f (x )>g (x )if x ≥x 0then we have S 2=L ({f }∪N )≺N =S ∞.12J.HIGESSuppose we have constructed a sequence of control semigroups with S n≺...≺S1so that for each control semigroup S i there is an asymptotic dim-control function f i that S i=L({f i}∪S i−1)and for every dim-control func-tion g of S i−1there exists a x0∈R+such that f i(x)>g(x)if x>x0.Now apply lemma5.2to f n in order to get a dim-control function f n+1so that for every j∈N,f n+1(x)>f j n(x)if x>x0for some x0.We claim that f n+1 satisfies the same property for all g∈S n.Let g be a dim-control function in S ing2.10we have that:g=f1◦f2◦...◦f pFor some functions f i in{f n}∪S n−1.For every f i there exist an x i and a j i such that f j i n(x)>f i(x)if x≥x i.Let x′0be the maximum of all x i then we have that g(x)<f p i=1j i n(x).We have obtained that g(x)<f j n(x)for some j if x≥x0and by the method we have built f n+1we have f n+1(x)>g(x) in a neighborhood of∞.Note that for getting S2we just need an asymptotic dim-control func-tion f that were strictly greater than any asymptotically linear dim-control function in a neighborhood of∞.Doing a dual reasoning we can get:Lemma5.5.There exist a sequence{ξi}i∈N small scale control semigroups withξi≺ξi−1andξi=L({f i}∪ξi−1)where f i is a dim-control function such that for every g∈ξi−1f i(x)>g(x)in a neighborhood of0.We have proved that the sets SMDIM and ASDIM are at least count-able.The lemmas of above suggest the following definition:Definition5.6.We will say that a large scale(small scale)control semigroup S(resp.ξ)is mono-bounded at∞(respectively at0)if there exist an asymptotic dim-control function f such that for every g∈S(resp g∈ξ)we have f(x)>g(x)in a neighborhood of∞(resp.in a neighborhood of0). The function f will be called the bound function of S(ofξ)at∞(resp.at 0).Using the sequence generated in5.4we can build the control semigroup generated by such sequence.Such semigroup will be mono-bounded and then we can begin again a similar process as in5.4.This is the idea of the next two lemmas.Lemma5.7.Let{S i}i∈N be a sequence of large scale control semigroups mono-bounded at∞such that S i≺S i−1and S i=L({f i}∪S i−1)where f i is a bound function of S i−1then the large scale control semigroup given by L( ∞i=1¯S i)is mono-bounded at∞.Proof.The reasoning is similar to the previous ones. Firstly we note that S:=L( ∞i=1S i)= ∞i=1S i.We will prove that S is mono-bounded at∞.We have that there is a sequence of points{a i}i∈N with a i→∞and a i+1>a i such that f i+1(x)>。

台湾国立交通大学

台湾国立交通大学

台湾国⽴交通⼤学数学视频数学视频Calculus I 台湾国⽴交通⼤学 Michael Fuchs⽼師 36集(点击进⼊我的淘宝店)Calculus II 台湾国⽴交通⼤学 Michael Fuchs⽼師 29集(点击进⼊我的淘宝店)Chapter1 Functions and Model1-5 Exponential Functions1-6 Inverse Functions and LogarithmsChapter2 Limits and Derivatives2-2 The Limit of a Function2-4 The Precise Definition of a Limit2-3 Calculating Limits Using the Limit Laws2-6 Limits at Infinity; Horizontal Asymptotes2-5 Continuity2-8 Derivatives2-9 The Derivative as a FunctionChapter3 Differentiation Rules3-1 Derivatives of Polynomials and Exponential Functions3-2 The Product and Quotient Rules3-4 Derivatives of Trigonometric Functions3-5 The Chain Rule3-6 Implicit Differentiation3-8 Derivatives of Logarithmic Functions3-10 Related Rates3-7 Higher Derivatives3-11 Linear Approximations and DifferentialsChapter4 Applications of Differation4-1 Maximum and Minimum Values4-2 The Mean Value Theorem4-3 How Derivatives Affect the Shape of a Graph4-4 Indeterminate Forms a nd L’Hospital’s Rule4-7 Optimization Problems4-5 Summary of Curve Sketching4-10 AntiderivativesChapter5 Integrals5-1 Areas and Distances5-2 The Definite Integral5-3 The Fundamental Theorem of Calculus5-4 Indefinite Integrals and the Total Change Theorem5-5 The Substitution Rule5-6 The Logarithm Defined as an IntegralChapter6 Applications of Integration6-1 Areas between Curves6-2 Volumes6-3 Volumes be Cylindrical ShellsChapter7 Techniques of Integration7-1 Integration by Parts7-2 Trigonometric Integrals7-3 Trigonometric Substitution7-4 Integration of Rational Functions by Partial Fractions7-8 Improper Integrals7-7 Approximate IntegrationChapter8 Further Applications of Integration8-1 Arc Length8-2 Area of a Surface of RevolutionChapter10 Parametric Equations and Polar Coordinates10-1 Curves Defined by Parametric Equations10-2 Calculus with Parametric Curves10-3 Polar Coordinates10-4 Areas and Lengths in Polar Coordinates微积分(⼀) 台湾国⽴交通⼤学莊重⽼師 24集(点击进⼊我的淘宝店)微积分(⼆)台湾国⽴交通⼤学莊重⽼師 24集(点击进⼊我的淘宝店)課程章節第⼀章Functions and Model第⼆章Limits and derivatives第三章Differentiation Rules第四章The Properties of Gases第五章Integrals第六章Applications of Integration第七章Techniques of Integration第⼋章Further Applications of Integration第⼗章Parametric Equations and Polar Coordinates第⼗⼀章Infinite Sequences and Series第⼗⼆章Vectors and the Geometry of Space第⼗三章Vector Functions第⼗四章Partial Derivatives第⼗五章Multiple Integrals⾼等微积分(⼀)台湾国⽴交通⼤学⽩啟光⽼師 29集(点击进⼊我的淘宝店)⾼等微积分(⼆) 台湾国⽴交通⼤学 ⽩啟光⽼師 27集(点击进⼊我的淘宝店)第⼀章The Real and Complex Number SystemsFields Axioms, Order Axioms Completeness Axioms第⼆章Basic TopologyCardinality of SetsMetric SpacesCompact SetsConnected Sets第三章Numerical Sequences and SeriesConvergent SequencesCauchy SequencesUpper and Lower LimitsSeries of Nonnegative TermsThe Root and Ratio TestAbsolute Convergence, Rearrangements第四章ContinuityLimits of Functions and Continuous FunctionsContinuity and CompactnessContinuity and Connectednessdiscontinuities, Infinite Limits and Limits at Infinity第五章Differentiation The Derivative of a Real Function, Mean Value TheoremL’Hopital’s RuleTaylor’s TheoremDifferentiation of Vector-valued Functions第六章The Riemann-Stieltjes Integral Definition and Existence of the IntegralProperties of the IntegralIntegration and DifferentiationIntegration and Differentiation第六章The Riemann-Stieltjes IntegralIntegration and Differentiation第七章Sequence and Series of FunctionsSequence and Series of Functions --- the Main ProblemUniform Convergence and ContinuityUniform Convergence and IntegrationUniform Convergence and DifferentiationEquicontinuous Family of FunctionsThe Stone-Weierstrass Theorem第⼋章Some Special FunctionsPower seriesSome Special FunctionsFourier SeriesThe Gamma Function第九章Functions of several variablesFunction of Several VariablesFunction of Several Variables:DifferentiationFunction of Several Variables:DifferentiationThe Inverse Function TheoremThe Implicit Function TheoremThe Rank TheoremDeterminantsDifferentiation of Integrals偏微分⽅程(⼀) 台湾国⽴交通⼤学林琦焜⽼师 3.8GB (点击进⼊我的淘宝店)偏微分⽅程(⼆) 台湾国⽴交通⼤学林琦焜⽼师 3.4GB (点击进⼊我的淘宝店)内容纲要第⼀章 The Single First-Order Equation1-1 Introduction Partial differential equations occur throughout mathematics. In this part we will give some examples1-2 Examples1-3 Analytic Solution and Approximation methods in a simple example 1-st order linear example1-4 Quasilinear Equation The concept of characteristic1-5 The Cauchy Problem for the Quasilinear-linear Equations1-6 Examples Solved problems1-7 The general first-order equation for a function of two variables characteristic curves, envelope1-8 The Cauchy Problem characteristic curves, envelope1-9 Solutions generated as envelopes第⼆章Second-Order Equations: Hyperbolic Equations for Functions of Two Independent Variables2-1 Characteristics for Linear and Quasilinear Second-Order Equations Characteristic2-2 Propagation of Singularity Characteristic curve and singularity2-3 The Linear Second-Order Equation classification of 2nd order equation2-4 The One-Dimensional Wave Equation dAlembert formula, dimond law, Fourier series2-5 System of First-Order Equations Canonical form, Characteristic polynominal2-6 A Quasi-linear System and Simple Waves Concept of simple wave第三章 Characteristic Manifolds and Cauchy Problem3-1 Natation of Laurent Schwartz Multi-index notation3-2 The Cauchy Problem Characteristic matrix, characteristic form3-3 Real Analytic Functions and the Cauchy-Kowalevski Theorem Local existence of solutions of the non-characteristic 3-4 The Lagrange-Green Identity Gauss divergence theorem3-5 The Uniqueness Theorem of Ho ren Uniqueness of analytic partial differential equations3-6 Distribution Solutions Introdution of Laurent Schwartzs theory of distribution (generalized function)第四章 The Laplace Equation4-1 Greens Identity, Fundamental Solutions, and Poissons Equation Dirichlet problem, Neumann problem, spherical symmetry, mean value theorem, Poisson formula4-2 The Maximal Principle harmonic and subharmonic functions4-3 The Dirichlet Problem, Greens Function, and Poisson Formula Symmetric point, Poisson kernel4-4 Perrons method Existence proof of the Dirichlet problem4-5 Solution of the Dirichlet Problem by Hilbert-Space Methods Functional analysis, Riesz representation theorem, Dirichlet integra第五章 Hyperbolic Equations in Higher Dimensions5-1 The Wave Equation in n-Dimensional Space(1) The method of sphereical means(2) Hadmards method of descent(3) Duhamels principle and the general Cauchy problem(4) mixed problem5-2 Higher-Order Hyperbolic Equations with Constant Coefficients(1) Standard form of the initial-value problem(2) solution by Fourier transform,(3) solution of a mixed problem by Fourier transform5-3 Symmetric Hyperbolic System(1) The basic energy inequality(2)Finite difference method(3) Schauder method第六章 Higher-Order Elliptic Equations with Constant Coefficients6-1 The Fundamental Solution for Odd n Travelling wave6-2 The Dirichlet Problem Lax-Milgram theorem, Garding inequality6-3 Sobolev Space Weak solution and Hibert space第七章 Parabolic Equations7-1 The Heat Equation Self-Similarity, Heat kernel, maximum principle7-2 The Initial-Value Problem for General Second-Order Parabolic Equations(1) Finite difference and maximum principle(2) Existence of Initial Value Problem第⼋章 H. Lewys Example of a Linear Equation without Solutions8-1 Brief introduction of Functional Analysis Hilbert and Banach spaces, projection theorem, Leray-Schauder theorem8-2 Semigroups of linear operator Generation, representation and spectral properties8-3 Perturbations and Approximations The Trotter theorem8-4 The abstract Cauchy Problem Basic theory8-5 Application to linear partial differential equations Parabolic equation, Wave equation and Schrodinger equation8-6 Applications to nonlinear partial differential equations KdV equation, nonlinear heat equation, nonmlinear Schrodinger equation变分学导论应⽤数学系林琦焜⽼师台湾国⽴交通⼤学 2GB (点击进⼊我的淘宝店)内容纲要第⼀章变分学之历史名题1.1 Bernoulli 最速下降曲线1.2 最⼩表⾯积的迴转体1.3 Plateau问题(最⼩曲⾯)1.4 等周长问题1.5 古典⼒学之问题第⼆章 Euler- Lagrange⽅程2.1 变分之原理2.2 折射定律与最速下降曲线2.3 ⼴义座标2.4 Dirichlet 原理与最⼩曲⾯2.5 Lagrange乘⼦与等周问题2.6 Euler-Lagrage ⽅程之不变量2.7 Sturm-Liouville问题2.8 极值(积分)问题第三章 Hamilton系统3.1 Legendre变换3.2 Hamilton⽅程3.3 座标变换与守恒律3.4 Noether定理3.5 Poisson括号第四章数学物理⽅程4.1 波动⽅程4.2 Laplace与Poisson⽅程4.3 Schrodinger ⽅程4.4 Klein-Gordon ⽅程4.5 KdV ⽅程4.6 流体⼒学⽅程 课程书⽬变分学导论 (Lecture note by Chi-Kun Lin).向量分析台湾国⽴交通⼤学林琦焜 3.3GB (点击进⼊我的淘宝店)向量分析主要是要谈”梯度、散度与旋度”这三个重要观念,⽽对应的则是⽅向导数、散度定理、与Stokes定理因此重⼼就在於如何釐清线积分、曲⾯积分以及他们所代表的物理意义。

逆半群同余的对偶刻画

逆半群同余的对偶刻画

五邑大学学报(自然科学版)JOURNAL OF WUYI UNIVERSITY (Natural Science Edition )第34卷第2期2020年5月Vol.34No.2May 2020文章编号:1006-7302(2020)02-0001-05逆半群同余的对偶刻画赵雪欣,陈芬芬,高连飞,谢祥云(五邑大学数学与计算科学学院,广东江门529020)摘要:本文给出了逆半群上同余的核迹对偶刻画,并在此基础上给出了最小群同余和最大幂等分离同余的对偶刻画及其相关性质.本文的主要结论是:设S 为逆半群,(,)N τ是S 上的同余对,则关系(,)N ρτ是S 上的一个同余;min στ=是S 上的最小群同余当E E τ=⨯;max μτ=是S 上的最大幂等分离同余当1E τ=.关键词:逆半群;核迹同余;最小群同余;最大幂等分离同余中图分类号:O152.7文献标志码:AA Dual Characterization of Congruences on Inverse Semigroups ZHAO Xue-xin,CHEN Fen-fen ,GAO Lian-fei,XIE Xiang-yun(School of Mathematics and Computational Science,Wuyi University,Jiangmen 529020,China)Abstract:In this paper,the kernel-trace dual characterization of congruences on inverse semigroups isobtained.On this basis,the dual characterization of the minimum group congruence and the maximumidempotent-separating congruence and the related properties are given.The main conclusions of this paperare as follows:let S be an inverse semigroup,(,)N τbe a congruence pair of S ,then the relation(,)N ρτis a congruence on S .min στ=is the minimum group congruence on Sif E E τ=⨯,and max μτ=is the maximum idempotent-separating congruence on S if 1E τ=.Key words:Inverse semigroups;Kernel-trace congruence;Minimum group congruence;Maximum idempotent-separating congruenceVagner [1]在1952年首次提出逆半群理论,随后Preston [2-4]也提出这个概念.Vagner [5]最初把逆半群称为“广义群”,无论是Vagner 还是Preston ,最初提出逆半群的动机是研究集合上的部分一一映射所构成的半群.逆半群最早的结果之一是表示定理(类似于群论中的Cayley 定理),即每个逆半群都具有忠实表示作为部分一一映射的逆半群.由于逆半群的理论与群的理论有许多相似之处,所以促使了很多学者对逆半群上的同余关系进收稿日期:2019-12-30基金项目:国家自然科学基金资助项目(11801081);广东省自然科学基金资助项目(2014A030313625,2018A030313063);广东省普通高校特色创新类项目(2018KTSCX234);广东省教学团队项目(粤教高函【2018】179号);安徽省高校自然科学研究项目(KJ2018A0329);研究生示范课建设项目(2016SFKS_40).作者简介:赵雪欣(1995—),广东江门人,在读硕士生,研究方向为半群的代数理论;谢祥云,教授,博士,硕士生导师,通信作者,研究方向为序半群的代数理论、模糊代数、粗糙集理论.五邑大学学报(自然科学版)2020年2行研究学习,特别是核—迹方法对同余的刻画.1961年,Munn [6]首次提出逆半群的同余σ,并给出其刻画;1964年Howie [7]给出了最小群同余和最大幂等分离同余的刻画;1966年Lallement [8]给出对b H μ=的研究.1974年Scheiblich [9]首次提出核与迹;1975年和1978年Green [10]和Petrich [11]也对其做了更深入的研究.1981年Petrich 和Reilly [12]将相同的想法应用于完全单半群.本文基于Petrich [11]和Howie [7,13]对逆半群核迹同余的研究,给出了核迹同余的一种对偶刻画,并在此基础上研究了S 上的最小群同余和最大幂等分离同余的相应理论和性质.1预备知识本节的基本知识主要来源于文献[13].在一个半群S 中,定义Green 关系如下:1,,L a x y S x b a b a yb ⇔∃∈==();1R (,),a b u v S au b bv a ⇔∃∈==;H =L R ∧;D =L R ∨.设a S ∈,元素a 称为正则的,如果存在x S ∈,使得axa a =.如果S 中每一个元素都是正则的,则S 称为正则的.半群S 称为逆半群如果S 中每个元均有逆元存在且幂等元可交换.引理1[13]145设S 是一个半群,则下列命题等价1)S 是一个逆半群;2)S 是正则的,并且它的幂等元可换;3)每一个L 类和每一个R 类只包含一个幂等元;4)S 中的每一个元素有唯一逆元.引理2[13]146设S 是一个逆半群,()E S 是其幂等元构成的半格.那么1)对任意,a b S ∈,111()ab b a ---=;2)对任意a S ∈,()e E S ∈,1aea -和1a ea -都是幂等元.设ρ是逆半群S 上的一个同余关系,()E S 是S 的幂等元构成的半格.ρ限制在()E S 上是()E S 的一个同余关系,我们称为ρ的迹,写作tr τρ=.每一个τ类e τ等于()e E S ρ .同余关系τ称作正规的,如果11 () e f a S a ea a fa ττ--⇒∀∈.我们知道,设ρ是逆半群S 上的一个同余关系,那么S ρ/是一个群当且仅当tr ()()E S E S ρ=⨯.设ρ是逆半群S 上的同余关系,S 的幂等元集是()E S ,那么ρ的核定义为Ker e EN ep ρ∈== .文献[13]中已经证明逆半群S 的核N 是S 的完全逆子半群且是正规的,且核Ker ρ与迹tr ρ有如下关系:1)1ker ,(,)tr ker ae e a a a ρρρ-∈∈⇒∈;2)11ker (,)tr a aa a a ρρ--∈⇒∈.若N 是S 上的正规子半群,τ是()E S 上的正规同余,则S 上的同余对(,)N τ定义如下[13]:1)1,(,)ae N e a a a N τ-∈∈⇒∈;2)11(,)a N aa a a τ--∈⇒∈.定理1[13]157设S 为逆半群,()E S 是S 上幂等元构成的半格,若ρ为S 上的同余,则(ker ,tr )ρρ是一个同余对;反之,若(,)N τ是一个同余对,则关系111(,){(,):(,),}N a b S S a a b b ab N τρτ---=∈⨯∈∈是S 上的一个同余,并且(,)ker N N τρ=,(,)tr N τρτ=,(Ker ,tr )ρρρρ=.本文主要内容是在定理1的基础上给出它的对偶定理.第34卷第2期3赵雪欣等:逆半群同余的对偶刻画2主要结果在给出定理1的对偶刻画之前,我们有以下关于同余对的对偶定义:定义1S 是一个逆半群,()E S 是S 上幂等元构成的半格.设N 是S 上的正规子半群,τ是()E S 上的正规同余,称(,)N τ是S 的一个同余对如果满足条件:对任意的a S ∈和()e E S ∈,有C1)1,(,)ea N e aa a N τ-∈∈⇒∈;C2)11(,)a N a a aa τ--∈⇒∈.引理3设S 是一个逆半群,()E S 是S 上幂等元构成的半格,ρ是S 上的一个同余,则对任意的a S ∈,()e E S ∈,有1)1ker ,(,)ker ea e aa a ρτρ-∈∈⇒∈;2)11ker (,)tr a a a aa ρρ--∈⇒∈.证明1)设 ea f ρ,()f E S ∈,因为1() a aa a ea f ρρ-=,所以ker a ρ∈.2)设a e ρ∈,()e E S ∈,显然有1a e ρ-∈,于是有11,a a aa e ρ--∈,则11(,)tr a a aa ρ--∈.引理4设S 是一个逆半群,(,)N τ是S 的一个同余对,则对于任意的,a b S ∈,()e E S ∈,有1)若bea N ∈,1 e aa τ-,则ba N ∈.2)若11(,)aa bb τ--∈,1a b N -∈,则对每个()e E S ∈,有11(,)aea beb τ--∈.证明1)设bea N ∈,1 e aa τ-,则11()bea bb bea beb ba f ba N --===∈,其中1()f beb E S -=∈.因为1111( ) (())f beb baa b ba ba τ----==,所以根据定义1的C1),有ba N ∈.2)设11(,)aa bb τ--∈,1a b N -∈,则(mod )τ有11111111111111111111()() ()()()()()() ( )()() ()() (C2), ) aea a a a ee a a a aea aa aea aea bb aea aa bb ae a b a b ea ae b a a b ea a b N τττ--------------------==≡=≡∈因为和是正规的根据因为和是正规的111111111111 ()() (C2), )( ).a a be a be a aa beb aa a be N bb beb bb E beb ττ------------=≡∈≡=根据因为和是正规的因为是上的一个同余,所以,11(,)aea beb τ--∈.下面我们给出定理1逆半群同余的对偶刻画:定理2设S 为逆半群,()E S 是S 上幂等元构成的半格,若ρ为S 上的同余,则(ker ,tr )ρρ是一个同余对;反之,若(,)N τ是一个同余对,则关系111(,){(,):(,),}N a b S aa bb a b N τρτ---=∈∈∈是S 上的一个同余.此外(,)ker N N τρ=,(,)tr N τρτ=,(ker ,tr )ρρρρ=.证明""⇒.若ρ为S 上的同余,则由引理3,(ker ,tr )ρρ是一个同余对.""⇐.设(,)N τ是一个同余对,111(,){(,):(,),}N a b S aa bb a b N τρτ---=∈∈∈.因为N 是S 上的正规子半群,τ是()E S 上的正规同余,所以ρ是自反的和对称的.要证ρ是传递的,只需证明对任意的五邑大学学报(自然科学版)2020年4(,)a b ρ∈,(,)b c ρ∈,有(,)a c ρ∈.设(,)a b ρ∈,(,)b c ρ∈,则由11(,)aa bb τ--∈,11(,)bb cc τ--∈,有11(,)aa cc τ--∈.因为11,a b b c N --∈,于是有111()a bb c a ec N ---=∈,1e bb -=.由于11() bb e cc τ--=,根据引理41),有1a c N -∈,所以(,)a c ρ∈.因此ρ是一个等价关系.要证ρ是一个同余关系,只需证明()(,)(,),(,)c S a b ac bc ca cb ρρρ∀∈∈⇒∈∈.设(,)a b ρ∈,则有11(,)aa bb τ--∈,1a b N -∈,根据引理42),有111111()()()() (mod )ac ac a cc a b cc b bc bc τ------=≡=.因为N 是正规的,所以有111()()ac bc c a b c N ---=∈,因此(,)ac bc ρ∈.又因为111111()() (mod )ca ca caa c cbb c cb cb τ------=≡=,111111111()()()()ca cb a c cb a c c bb b a b b c cb N ---------===∈,所以(,)ca cb ρ∈.因此,(,)N τρρ=是S 上的一个同余关系.若a e ρ∈,()e E S ∈,则1 e aa τ-,ea N ∈,根据定义1C1),有a N ∈.因此,ker N ρ⊆.反之,若a N ∈,则1e a ea N -=∈,1e aa -=;由幂等元11ee aa --=有11 ee aa τ--,所以ker a e ρρ∈⊆.因此,(,)ker N N τρ=.设,()e f E S ∈,若(,)(,)N e f τρρ∈=,则11() ( )e ee ff f τ--==,因此tr ρτ⊆.反之,若 e f τ,则11() ( )ee e f ff τ--==,1()e f ef E S N -=∈⊆,所以()(,)()()tr e f E S E S ρρ∈⨯= .因此(,)tr N τρτ=.设(,)a b ρ∈,则11(,)a b ρ--∈,于是有11(,)aa bb ρ--∈.因为1aa -,1bb -是幂等元,所以有11(,)tr aa bb ρ--∈.由于11(,)a b b b ρ--∈,则有11()ker a b b b ρρ--∈⊆,因此(ker ,tr )ρρρρ⊆.反之,设(ker ,tr )(,)a b ρρρ∈,则有11(,)tr aa bb ρ--∈,1ker a b ρ-∈.因为1()a b ρ-是/S ρ的一个幂等元,所以()()()111111()()()()a b a b a b a bb a ρρρρ------==,于是有1111111 (mod )b bb b aa b aa bb a aa aa a aa a a ρ-------=≡≡≡==,因此(,)a b ρ∈,故(ker ,tr )ρρρρ=.性质1设S 是一个逆半群,()E S 是S 上幂等元构成的半格,令τ是()E S 上的正规同余,然后有:1)关系111min {(,): ,(())( ,)}a b S S a a b b e E S e a a ae be τττ---=∈⨯∃∈=是S 上以τ为迹的最小同余;2)关系11max {(,):(()) }a b S S e E S aea beb ττ--=∈⨯∀∈是S 上以τ为迹的最大同余.证明仿照文献[13]中相关定理的证明,省略.推论1若E E τ=⨯,则min {(,):(())}a b S S e E S ae be τ=∈⨯∃∈=.一般地,记min στ=,称为S 上的最小群同余,则/S σ是S 上的最大群同态像.于是,对于S 上使得/S γ是一个群的每一个同余γ,都存在一个同态://S S ζσγ→.容易验证,σ的核如下:定理3设S 是一个逆半群,()E S 是S 上幂等元构成的半格,令σ是S 上的最小群同余,则ker E σω=,且1{(,):}a b S S a b E σω-=∈⨯∈.一个同余称为幂等分离,如果它的迹是恒同余1的.根据性质12),我们有以下推论:推论2若1E τ=,则11max {(,):(())}a b S S e E S aea beb τ--=∈⨯∀∈=.一般地,记max μτ=,称为S 上的最大幂等分离同余.要确定ker μ,我们需要一个定义:在S 上设()E S 的中心化子为E ς,其中{:(())}E a S e E S ae ea ς=∈∀∈=.然后有下面的结论:第34卷第2期5赵雪欣等:逆半群同余的对偶刻画定理4设S 是一个逆半群,()E S 是S 上幂等元构成的半格,令μ是S 上的最大幂等分离同余,则ker E μς=,且111{(,):,}a b S S aa bb a b E μς---=∈⨯=∈.下面给出μ的进一步刻画:性质2设S 是一个逆半群,()E S 是S 上幂等元构成的半格,令μ是S 上的最大幂等分离同余,则b H μ=为S 上包含在H 里的最大同余.证明设(,)a b μ∈,则由定理4得11aa bb --=.因为11(,)a b μ--∈,所以有11a a b b --=,因此(,)H a b ∈,H μ⊆.下面证明μ是S 上包含在H 里的最大同余.考虑S 上的任意一个同余ρ,且H ρ⊆.假设(,)a b ρ∈,则11(,)a b ρ--∈,于是对任意的()e E S ∈,有11(,)H aea beb ρ--∈⊆.由S 上的每个H 类最多包含一个幂等元可知,对任意的()e E S ∈,有11aea beb --=,因此(,)a b μ∈.一个逆半群S 称为基本的,如果在S 上的最大幂等分离同余是恒同余1S 的.定理5设S 是一个逆半群,()E S 是S 上幂等元构成的半格,令μ是S 上的最大幂等分离同余,则/S μ是基本的,且/S μ上的幂等元所构成的半格与()E S 同构.参考文献[1]VAGNER V V.Generalized groups [J].Doklady AkademiĭNauk SSSR,1952(84):1119-1122.[2]PRESTON G B.Inverse semi-groups [J].Journal of the London Mathematical Society,1954,29(4):396-403.[3]PRESTON G B.Inverse semi-groups with minimal right ideals [J].Journal of the London Mathematical Society,1954,29(4):404-411.[4]PRESTON G B.Representations of inverse semi-groups [J].Journal of the London Mathematical Society,1954,29(4):411-419.[5]VAGNER V V.Theory of generalized heaps and generalized groups [J].MatematicheskiĭSbornik (NS),1953(32):545-632.[6]MUNN W D.A class of irreducible matrix representations of an arbitrary inverse semigroup [J].Proceedings ofthe Glasgow Mathematical Association,1961,5(1):41-48.[7]HOWIE J M.The maximum idempotent-separating congruence on an inverse semigroup [J].Proceedings of theEdinburgh Mathematical Society(2),1964,14(1):71-79.[8]LALLEMENT G.Congruences et équivalences de Green sur un demigroupe régulier [J].Comptes Rendus del’Acad émie des Sciences Paris(Sér A),1966,262:613-616.[9]SCHEIBLICH H E.Kernels of inverse semigroup homomorphisms [J].Journal of the Australian MathematicalSociety,1974,18(3):289-292.[10]GREEN D G.The lattice of congruences on an inverse semigroup [J].Pacific Journal of Mathematics,1975,57(1):141-152.[11]PETRICH M.Congruences on inverse semigroups [J].Journal of Algebra,1978,55(2):231-256.[12]PETRICH M,REILLY N R.The kernel-trace approach to congruences on completely simple semigroups [J].Studia Scientiarum Mathematicarum Hungarica,1981,16(4):103-114.[13]HOWIE J M.Fundamentals of semigroup theory [M].New York:Oxford University Press,1995.[责任编辑:韦韬]。

大小结构模型的稳定性和正则性

大小结构模型的稳定性和正则性
m a
= −μ=
β(a, P∗ )p∗ (a) da,
2
β
0,
(1.5) (1.6)
γ > 0.
These assumptions will suffice (and could actually be relaxed) to make the analysis of the linearized problem work. They are, however, generally not strong enough to prove global existence results for the nonlinear problem. In addition, for practical purposes several other biologically relevant assumptions (such as μ > 0) will have to be imposed on these functions. The population model treated here is equivalent to the one usually considered in the literature (see [1,2] and references therein) when the boundary condition (1.2) is replaced by
m
γ 0, P (t) p(0, t) =
0
β a, P (t) p(a, t) da,
t > 0,
(1.7)
and no population inflow from an external source takes place. Condition (1.2) incorporates the γ -term on the left of Eq. (1.7) in the birth rate β on the right. We prefer working with the boundary condition in the form of Eq. (1.2) to simplify the following developments. Local and global existence and uniqueness of solutions to this nonlinear problem have been analyzed in [2]. The model considered here reduces to the Gurtin–MacCamy (or McKendrick) nonlinear agestructured model if γ ≡ 1 (see [13]) and is a generalization of the simple problem treated in [10]. Similar physiologically structured population models have been studied intensively in the literature. Let us just mention the well-known works [20,22,25] for reference here. The main purpose of the present work is to investigate the linear stability of stationary solutions of the system (1.1)–(1.3) using semigroup techniques and spectral methods based on the characteristic equation. Linear semigroup methods were successfully developed to study the linear stability and regularity of solutions of linearized fluid flow problems where the underlying dynamics is driven by a one-dimensional mass transport equation (see [14,18,19]). The model equations treated in this work are similar in nature. Sophisticated semigroup methods have recently been used to obtain sharp regularity results for one-dimensional hyperbolic–elliptic fluid flow problems (see [15,17]) and to explain the phenomenon that the roots of the underlying characteristic equations (eigenvalues) are lined up along certain curves (see [16]). Earlier, quite deep semigroup results for an n-dimensional age-structured model with constant growth rate can be

Some properties of group-theoretical categories

Some properties of group-theoretical categories
A fusion category is said to be pointed if all its simple objects are invertible. A typical example of a pointed category is VecωG - the category of finite dimensional vector spaces over k graded by the finite group G. The morphisms in this category are linear transformations that respect the grading and the associativity constraint is given by the normalized 3-cocycle ω on G.
In Section 4 we first show that every group-theoretical category is graded by a certain double coset ring. As a consequence, we obtain a necessary and sufficient condition for a group-theoretical category to be nilpotent.
In Section 5 we give an explicit description of the simple objects in a grouptheoretical category (following Proposition 3.2 in [O2]; see Theorem 5.1) and of the group of invertible objects of a group-theoretical category, in group-theoretical terms.

高一英语学术研讨会组织单选题40题

高一英语学术研讨会组织单选题40题

高一英语学术研讨会组织单选题40题1. We are going to hold an academic seminar on _____.A.scienceB.historyC.mathD.literature答案:A。

本题考查学术研讨会常见主题名词辨析。

“science”意为“科学”,学术研讨会可以围绕科学领域的新发现、研究方法等展开;“history”意为“历史”,通常侧重于过去事件的研究;“math”意为“数学”,一般是关于数学理论和问题的讨论;“literature”意为“文学”,主要涉及文学作品和文学理论。

在这个语境中,通常学术研讨会会涉及科学领域的前沿研究等,所以选择“science”。

2. The seminar will focus on the topic of _____.A.artB.technologyC.geographyD.philosophy答案:B。

“art”是“艺术”,通常涉及绘画、音乐等艺术形式的讨论;“technology”意为“技术”,学术研讨会可以探讨新技术的发展和应用;“geography”意为“地理”,主要研究地球表面的自然和人文现象;“philosophy”意为“哲学”,侧重于对基本问题的思考和理论探讨。

这里说研讨会聚焦的主题,通常技术的发展变化快,容易成为研讨会的焦点,所以选“technology”。

3. Which of the following is a common theme for an academic seminar?A.musicB.biologyC.sportsD.fashion答案:B。

“music”是“音乐”,一般在音乐领域的活动中讨论;“biology”意为“生物学”,是学术研讨会常见的主题之一,可以探讨生物的结构、功能、进化等;“sports”意为“体育”,更多在体育活动或相关领域讨论;“fashion”意为“时尚”,主要在时尚界或相关领域被关注。

单演半群的几条性质(1)

第26卷第1期2004年3月 湘潭师范学院学报(自然科学版)Journal of Xiangtan Normal U niversity(N atural Science Edition) Vol.26N o.1Mar.2004单演半群的几条性质赵雨清(湖南科技大学数学与计算科学学院,湖南湘潭411201)摘 要:讨论了单演子半群的同态以及单演半群是零半群、1半群时的性质,得出了有限单演半群的几条性质。

关键词:单演半群;同态;零半群;1半群中图分类号:O152.7 文献标识码:A 文章编号:1671-0231(2004)01-0020-02文献[1]给出了单演子半群和单演半群的概念并研究了它们的结构和性质,讨论了单演子半群同构的充要条件。

作者在文献[1]的基础上讨论了单演子半群的同态和单演半群是零半群、幺半群的性质,得出了有限单演半群的几条性质。

1 有关概念和结论如果A 是半群S 的一个非空子集,那么S 的包含A 的子半群簇的交是S 的子半群,这个子半群用 A !表示。

它是由S 中所有的可以写成A 中有限个元素的乘积形式的元素组成的。

定义1 如果 A !=S ,则称A 是S 的一个生成集。

特殊地,当A ={a}时,a !={a ,a 2,a 3,∀∀}称 a !为由元素a 生成的S 的单演子半群,把 a !的阶叫做元素a 的阶。

定义2 如果对S 中的某个元素a ,有S = a !,则称S 为单演半群。

如果在序列a ,a 2,a 3,∀∀中不出现重复,则称 a !是一个无限单演半群,并且说元素a 有无限阶。

如果在a 的乘幂中出现重复,那么集合{x #N :(存在y #N )a x =a y ,x ∃y }和 {X #N:a m +x =a m}是非空的,并且分别有最小元m 和r 。

分别把m 和r 叫做单演半群 a !的指数和周期。

这时单演半群 a !称为有限单演半群,并常记为M(m ,r )。

特别地,M (1,r )是阶为r 的循环群。

C-lpp半直积英文

1.IntroductionSince 1960's,w hensemidirect pro ducts of sem igro ups wereintroduced by B.H.Neuman n,they hav e b een a subject of interest o fnumero us semigroup theo reticians and they have had a lo t o f significant applicatio ns in m any areas of semig roup theory.One o f the mo st frequently stated questio ns concerning semidirect product was the following :Under which co nditions a semidirect product of two semig ro ups has a given property?T he main aims o f this paper are to study semidirect pro ducts that have abundant properties.We g iv e here some kno wn results used in the seq uel.L em ma 1[3]L et a,e 2=e ∈S.T hen the fo llo wing statements areequiv alent:(1)aL *(R *)e(2)a=ae (a=ea)and for all x,y ∈S 1,ax =ay (x a =ya)implies thatex =ey(x e=y e)A semig ro up S is called a rpp (respectively,lpp)sem ig ro up if every L *-class (respectiv ely ,L *-class )of S contain s at least on e idem potent.A rpp sem ig ro up S is a right (respectiv ely,left)adequate sem igroup if theset o f idem potents,$E(S)$is a sem illatice.A sem ig ro up which is bo th left and rig ht adeq uate is called an adequate semig ro up.In [4],it is show nthat each L *-class(respectively,R *-class)o f S contains a uniqueidempo tent if S is a rig ht (respectiv ely,left)adequate semigroup.Fo r anelemen t a o f such a sem ig ro up,the idempo tent in the L *-class (respectively,R *-class)co ntainin g a will be denoted by a *(respectiv ely,a +)We call a rpp semigroup with central idempo tents a C -rpp sem ig ro up.C -lpp Sem ig ro ups are the dual semigroups o f C -rp p sem ig ro ups.Fo untain [5]has sho wn that a semigroup S is a C-rpp (respectiv ely,C-lpp)semigroup if and o nly if it is a strong sem illatice o f left (respectiv ely ,rig ht)cancellativ e mo noids.T hro ugho ut this chapter,we use the sam e definition and no tations as stated in [6].L et S ×θT be a sem idirect prod uct of two mo noids,S and T,where θ∶S →End(T )is a m ono id ho mom orphism.2.lpp S e midire ct P roductsLemm a 2Let S ×θT is rig ht adequate.Then the following statements are true:(1)S and T are rig ht adequate;(2)($e ∈E (S))($u ∈E(T))u e=u ;(3)(e,u)∈E (S ×θT)&e ∈E(S)and u ∈E(T)(4)($s ∈S)($t ∈T)(s ,t)L *(s *,t *).Proo f.(1)is easy to verify since S and T are both m ono ids.(2)(e,1)(1,u)=(1,u )(e,1)since (e,1)(1,u)∈E(S ×θT),which implies u e =u.(3)(,)∈(S ×θT)&e 2=e and u e u=u&e ∈E(S)and u ∈E(T)(by (2))(4)Since S ×θT is right adequate,there ex ists (e ,u )∈E (S ×θT )such that (e,u)L *(s,t).T hen sL *e=s *since S is rig ht adequate and t s*u =t multiplied by u which implies tu =t.L et x,y ∈T 1.Suppose that tx =ty .T hen (s ,t)(1,x)=(s ,t)(1,y)so that(e,u)(1,x)=(e,u)(1,y)T hat is ux=uy and thus tL *u=t *since T is right adequate.T heo rem 1S ×θT is right adequate if and o nly if the follow ingconditio ns ho ld:(1)S and T are right adequate;(2)($e ∈E(S))($u ∈E(T ))u e =u;(3)($s ∈S)($t ∈T )(s,t)L *(s *,t *).Pro of.By L em ma 2.1,w e need only to sho w the "if"part.B y (1),(3),S ×θT is rpp and by (2)and the pro of o f L em ma 2,it is clear that E (S ×θT)is a sem illatice and hence S ×θT is rig ht adequate.L em ma 3If the set of idempotents of S ×θT ,E(S ×θT)is cental,then (1)($e ∈E (S ))($t ∈T )t e=t ;(2)($s ∈S)($u ∈E(T))u s =u.Pro of.(e,1),(1,u )∈E (S ×θT)and (e,1)(1,t)=(1,t)(e,1),(1,u )(s,1)=(s ,1)(1,u)since E (S ×θT)is central,which imply t e =t and u s =u.T heo rem 2S ×θTis C-rpp if and on ly if (1)S and T are C-rpp ;(2)($e ∈E(S))($t ∈T)t e =t;(3)($s ∈S)($u ∈E(T))u s =u ;(4)($s ∈S)($t ∈T )(s,t)L *(s *,t *).Pro of.Necessity is obv io us.We only show sufficiency.By T heo rem 1,S ×θT is rig ht adequate.Let (e ,u )∈E (S ×θT ).It's clear that e ∈E (S )and u ∈E(T )since (2),(3)ho ld.Fo r all (s,t)∈S ×θT .(e,u )(s ,t)=(es ,u s t)=(es,ut)=(se,t e u)=(s,t)(e,u)T herefore,E(S ×θT )is central.T heo rem 3S ×θT is C-lp p if and only if (1)S and T are C-lpp ;(2)($e ∈E(S))($t ∈T)t e =t;(3)($s ∈S)($u ∈E(T))u s =u ;(4)S θ(M o n (T )where M o n (T)is the set of all mo nom orph ism s of T into itself.Pro of.Necessity .We only need to show (4)ho lds.Let t 1,t 2∈T and s ∈S .Suppose t 1s =t 2s .T hen (1,t 1)(s,1)=(1,t 2)(s,1).Since (s,1)R *(s +,1),w e have t 1=t 1s+=t 2s+=t 2.Hence S θ(M on(T)as required.Sufficiency.Sim ilar to the proo f of T heo rem 2,E (S ×θT)is central.We sho w S ×θT is lpp.For any (s ,t)∈S ×θT ,supp ose that(a,x )(b ,y)S ×θT and (a,x)(s,t)=(b,y)(s,t).T hen as=bs and x s t=y s t.T hus x s t +=y s t +and (x t +)s =(y t +)s .Fro m (4),S θ(Mo n(T)so that xt +=yt +.T hen (a,x)(s +,t +)=(+,x s ++)=(+,x +)=(+,y +)=(+,y s++)=(,y )(+,+)F ,C-lpp semidir ect productsWU J ingpe ng(Kunming University of Science and Tec hnology,Kunming,Yunna n,650093)【Abstract 】Necessary and sufficient co nditio ns fo r a sem idirect product o f mono ids being C-lpp are obtained in this paper.T he structure,particularly ,the -decom posability,of C -lpp semidirect products is also inv estigated.【K ey wor ds 】sem idirect pro ducts;C-lpp semigroups;σ-decom posable.C-l p p 半直积吴劲鹏(昆明理工大学数学系云南昆明650093)【摘要】本文得到了幺半群的半直积是C-l pp 的充分必要条件,研究了$C-$lpp 半直积的结构,特别是其同余可分解性。

具抽象边界条件的迁移方程的研究进展

具抽象边界条件的迁移方程的研究进展王胜华;黄伟【摘要】本文介绍了近年来国际上具抽象边界条件的迁移方程的研究进展,主要阐述了这类迁移方程解的构造性理论及应用等研究成果.【期刊名称】《上饶师范学院学报》【年(卷),期】2011(031)003【总页数】5页(P1-5)【关键词】迁移方程;抽象边界条件;构造性理论【作者】王胜华;黄伟【作者单位】上饶师范学院,江西上饶334001;南昌大学数学系,江西南昌330031【正文语种】中文【中图分类】O177.2迁移方程是研究在大块物质中,由于粒子(中子、辐射粒子、电子、种群细胞等)运动所产生的微观效应综合所致的宏观迁移现象规律的一种数学模型,它的精确数学表示是积分-微分型的迁移方程。

它涉及到物理学、化学、生物学和社会科学等众多学科。

另一方面,各类迁移方程的结构形式雷同,并在一定的条件下可以相互渗透和转化。

这说明脱离特定的对象来研究这类方程的一般数学理论是十分重要的。

仅就线性迁移方程而言,它所确定的迁移算子是一类无界、非自伴和豫解算子不紧的积-微分算子。

因此,研究这类算子不仅在应用上十分重要,而且对数学理论的发展也有着非常重要的意义。

1955年Lehner.J和Wing.G.M.在文献[1,2]中首先对无限平行板几何中最简单的中子迁移方程解的构造性理进行了研究,用现在分析和Peierls积分算子等方法,对这类方程相应的迁移算子的谱进行了系统的分析,得到了该迁移算子在平面上较完整的谱分布和这类迁移方程解的渐近展开等一系列结果。

1958年Jor2 gens在文献[3]中研究了任意有界凸体中最小速率大于零的一般中子迁移方程,利用泛函分析和半群理论等方法,证明了该方程相应的迁移算子产生的半群是紧的,从而利用谱映射原理得出了迁移算子的谱分析和这类迁移方程解的渐近展开式等一系列结果。

之后,国际上关于迁移方程解的构造性理论及应用研究取得了许多优秀的成果(部分见文献[4-12])。

fulltext1

Semigroup ForumDOI10.1007/s00233-011-9301-2R E S E A R C H A RT I C L EOn the translational hull of a type B semigroupChunhua Li·Li-min WangReceived:3December2010/Accepted:28February2011©Springer Science+Business Media,LLC2011Abstract In this paper,the translational hull of a type B semigroup is considered. We prove that the translational hull of a type B semigroup is itself a type B semi-group,and give some properties and characterizations of the translational hulls of such semigroups.Moreover,we consider the translational hulls of some special type B semigroups.These results strengthen the results of Fountain and Lawson(Semi-group Forum32:79–86,1985)on adequate semigroups.Finally,we give a new proof of a problem posted by Petrich on translational hulls of inverse semigroups in Petrich (Inverse Semigroups,Wiley,New York,1984).Keywords Translational hulls·Type B semigroups·Proper·E-reflexive1IntroductionLet a and b stand for arbitrary elements of a semigroup S.A mappingλ[resp.ρ] from S to itself is a left[resp.right]translation of S ifλ(ab)=(λa)b[resp. (ab)ρ=a(bρ)];if also a(λb)=(aρ)b,thenλandρare linked and the pair(λ,ρ)is Communicated by Marcel Jackson.This work is supported by the National Science Foundation(No.11061014),the JiangXi Educational Department Natural Science Foundation of China(No.GJJ[2010]453),and the Foundation of East China Jiaotong University.C.Li()School of Basic Science,East China Jiaotong University,Nanchang,Jiangxi330013,P.R.Chinae-mail:chunhuali66@C.Li·L.-m.WangSchool of Mathematics,South China Normal University,Guangzhou,Guangdong510631,P.R.China L.-m.Wange-mail:wanglm@C.Li,L.-m.Wang a bitranslation of S.We denote by (S)[resp.I(S)]the set of all left[resp.right] translations of S,and denote by (S)the set of all bitranslations of S.It is easy to check that (S)forms a subsemigroup of (S)×I(S).We call (S)the transla-tional hull of S(see[1,9]).The reader may consult[11]or[9]for the role played by the translational hull in the general algebraic theory of semigroups.In this paper,we are concerned with analogues in the theory of type B semigroups of Ponizovski’s theorem which states that the translational hull of an inverse semi-group is still a semigroup of the same type(see[6,11,12]).As in[2],the relations L∗and R∗are defined on a semigroup S by the rule that the elements a and b of S are related by L∗[resp.R∗]on S if and only if they are related by L[resp.R]in some oversemigroup of S(see[2]).The intersection of the equivalence relations L∗and R∗is denoted by H∗.According to Fountain [5],a semigroup S is called rpp[resp.lpp]if and only if each L∗class[resp.R∗class]contains at least one idempotent.A semigroup S is called abundant if it is both rpp and lpp.An rpp[resp.a lpp]semigroup in which the idempotents commute is right adequate[resp.left adequate].A semigroup S is called adequate if and only if it is both right and left adequate.For convenience,we denote by a+[a∗]a typical idempotent R∗-related[L∗-related]to a.And E(S)denotes the set of idempotents of S.A right adequate semigroup S is called right type B,if it satisfies:(B1)for all e,f∈E(S1),a∈S,(ef a)∗=(ea)∗(f a)∗;(B2)if for all a∈S,e∈E(S),e≤a∗,then there is an element f∈E(S1)such that e=(f a)∗,where“≤”is a natural partial order on E(S)(i.e.,(∀e,f∈E(S)) e≤f⇔e=ef=f e(see[8])).Dually,we can define a left type B semigroup.A semigroup S is called type B if and only if it is both right and left type B(see[3]).In this paper,our main aim is to show that the translational hull of a type B semi-group is itself type B.This result is an interesting result because it generalizes the corresponding results for inverse semigroups.In particular,we give a new proof of a problem posted by Petrich in1984[10]on translational hulls of E-reflexive inverse semigroups(referring to the Guo-Shum article for the original solution[7]).2PreliminariesWe follow the notions adopted in[2–4,10].First,we state some known results and notations which will be frequently used throughout the paper.Lemma2.1[2]Let S be a semigroup and a,b∈S.Then the following statements are equivalent:(1)a L∗b(a R∗b);(2)for all x,y∈S1,ax=ay[xa=ya]if and only if bx=by[xb=yb]. Corollary2.2[2]Let S be a semigroup and e2=e,a∈S.Then the following state-ments are equivalent:(1)a L∗e(a R∗e);On the translational hull of a type B semigroup(2)ae=a[ea=a]and for all x,y∈S1,ax=ay[xa=ya]implies ex=ey[xe=ye].Evidently,L∗is a right congruence while R∗is a left congruence.In an arbitrary semigroup,we have L⊆L∗and R⊆R∗.But for regular elements a,b,we get a L∗b [a R∗b]if and only if a L b[a R b].Lemma2.3[2]Let S be an abundant semigroup and a,b∈S.DefineμL={(a,b)∈S×S:(ea)L∗(eb)for all e∈E(S)},μR={(a,b)∈S×S:(ae)R∗(be)for all e∈E(S)}.Putμ=μL∩μR.ThenμL[resp.μR,μ]is the largest congruence on S contained in L∗[resp.R∗,H∗].Lemma2.4[3]Let S be an adequate semigroup and a,b∈S.Then the following statements are true:(1)(ab)+=(ab+)+and(ab)∗=(a∗b)∗;(2)S is a strong semilattice of cancellative monoids if and only if E(S)is centralin S.Since it is known that for an adequate semigroup S,each L∗-class and each R∗-class of S contains exactly one idempotent,μL andμR are idempotent-separating. As in[3],an adequate semigroup S is called fundamental ifμ=1S.Byλa[resp.ρa],we mean the inner left[resp.right]translation which is defined byλa(x)=ax [resp.xρa=xa].Let S be an adequate semigroup and a∈S,(λ,ρ)∈ (S).Then we define the mappingsλ∗,λ+,ρ∗andρ+which map S into itself by the following:λ∗a=(λa+)∗a,λ+a=(a+ρ)+a,aρ∗=a(λa∗)∗,aρ+=a(a∗ρ)+. Thus,it is clear that(λ∗,ρ∗),(λ+,ρ+)∈ (S)and furthermore,those elements are the idempotents of (S)(see[4]).Lemma2.5[4]Let S be an abundant semigroup.Then(1)λ1=λ2⇐⇒(∀e∈E(S))λ1e=λ2e;(2)ρ1=ρ2⇐⇒(∀e∈E(S))eρ1=eρ2.Lemma2.6[4]Let S be an adequate semigroup.Then the following statements are true:(1)eρ∗=λ∗e=(λe)∗∈E(S),for all e∈E(S);(2)eρ+=λ+e=(eρ)+∈E(S),for all e∈E(S);(3)(λ∗,ρ∗)L∗(λ,ρ)R∗(λ+,ρ+);(4)E( (S))={(λ,ρ)∈ (S)|λE(S)∪E(S)ρ⊆E(S)};(5) (S)is adequate.C.Li,L.-m.WangLemma2.7Let S be a type B semigroup.Define a relation on S as follows:(a,b)∈σ⇐⇒eae=ebe,for some e∈E(S).Thenσis the least cancellative congruence on S.Proof Obviously,σis an equivalence relation on S.Now,we prove thatσis left compatible.Let(a,b)∈σ.Then eae=ebe for some e∈E(S).Hence,for any c∈S,we have ceae=cebe.By Lemma2.1,c∗eae= c∗ebe.Note that c∗e≤c∗and S satisfies Condition(B2).We have c∗e=(f c)∗for some f∈E(S1),this gives(f c)∗ae=(f c)∗be.By Lemma2.1,we have f cae= f cbe.Multiplying it on the left by e and on the right by f,we obtain thatef(ca)ef=ef(cb)ef,where(ef)∈E(S).Thus(ca,cb)∈σ.Therefore,σis a left congruence on S.Similarly,we can prove thatσis a right congruence on S.Next,we prove thatσis left cancellative.To see it,let a,b,c∈S be such that cσaσ=cσbσ,that is,(ca,cb)∈σ.By the definition ofσ,we have that e(ca)e= e(cb)e for some e∈E(S).Hence,we havee(ca)e=e(cb)e=⇒(ec)ae=(ec)be=⇒(ec)∗ae=(ec)∗be=⇒[e(ec)∗]a[e(ec)∗]=[e(ec)∗]b[e(ec)∗], where e(ec)∗∈E(S).Thus,(a,b)∈σ,that is,aσ=bσ.Therefore,σis left can-cellative.Similarly,we can show thatσis right cancellative.Now,letρbe any cancellative congruence on S.If(a,b)∈σ,then eae=ebe for some e∈E(S).Hence,eρaρeρ=eρbρeρ.Again,sinceρis cancellative,we get aρ=bρ,that is,(a,b)∈ρ.Thus,σ⊆ρ.This completes the proof.In this paper,we call a type B semigroup S proper ifσ∩L∗=σ∩R∗=1S. Recall that a semigroup S is left E-unitary if(∀e∈E(S),a∈S)ea∈E(S)implies a∈E(S).Dually,we can define right E-unitary.A semigroup S is called E-unitary if it is both left and right E-unitary.Lemma2.8Let S be a proper type B semigroup.Then S is E-unitary.Proof Let e∈E(S),a∈S be such that ea∈E(S).Then(eae)a=ea=eaa∗= eeaa∗=eaea∗=(eae)a∗.That is,(eae)a=(eae)a∗.Hence,multiplying it on the right by(eae),we get that(eae)a(eae)=(eae)a∗(eae),where(eae)∈E(S).By the definition ofσ,we have(a,a∗)∈σ.Hence,a[σ∩L∗]a∗,and so a=a∗∈E(S) since S is proper.That is,S is left E-unitary.Similarly,we can show that S is right E-unitary.This completes the proof.We remark here that a proper type B semigroup is E-unitary,but the converse is not true.The following example shows that there exists an E-unitary type B semigroup which is not proper.On the translational hull of a type B semigroupExample2.1Let N be the set of all non-negative integers and put I=N×N,S= N∪I.Define a multiplication“◦”on S as follows:m◦n=m+nm◦(h,k)=(m+h,k)(h,k)◦m=(h,k+m)(h,k)◦(m,n)=(h,k+m+n)It is readily verified that“◦”is associative,and that the set of idempotents of S is {0,(0,0)}.As in[2],Fountain proved that the L∗-classes of S are N and I,and that S is a right type B semigroup.In fact,it is easily observed that the R∗-classes of S are also N and I,and that S is also a left type B semigroup.Moreover,we can easily check that S is E-unitary.However,for(1,1),(0,2)∈S,we have(0,0)◦(1,1)◦(0,0)=(0,0)◦(0,2)◦(0,0)=(0,2).That is,(1,1)[L∗∩σ](0,2)and(1,1)[R∗∩σ](0,2).But(1,1)=(0,2).This shows that S is not proper.3The translational hull of a type B semigroupIn this section,we shall prove that the translational hull of a type B semigroup is still the same type of semigroup.Furthermore,we consider some basic properties of this class of semigroups.Lemma3.1Let S be a type B semigroup and(λ1,ρ1),(λ2,ρ2)∈ (S).Then the following statements are equivalent:(1)(λ1,ρ1)=(λ2,ρ2);(2)λ1=λ2;(3)ρ1=ρ2.Proof Note that(1)⇔(2)is the dual of(1)⇔(3)and(1)⇒(2)is clear.We only need to show that(2)⇒(1).Letλ1=λ2.Thenλ1f=λ2f for all f∈E(S).Hence,for all e∈E(S),we have λ1f=λ2f=⇒e(λ1f)=e(λ2f)=⇒(eρ1)f=(eρ2)f=⇒[(eρ1)f]∗=[(eρ2)f]∗=⇒[(eρ1)∗f]∗=[(eρ2)∗f]∗=⇒(eρ1)∗f=(eρ2)∗fC.Li,L.-m.Wang since S is L∗-unipotent(i.e.,each L∗-class of S contains exactly one idempotent). Choose an idempotent(eρ1)∗of S to replace the element f of the above formula.We get that(eρ1)∗=(eρ2)∗(eρ1)∗.Similarly,(eρ2)∗=(eρ1)∗(eρ2)∗.Again,since E(S)is a semilattice,we have(eρ1)∗=(eρ2)∗.Thus,for all e∈E(S),we haveeρ1=(eρ1)(eρ1)∗=e(λ1(eρ1)∗)=e(λ2(eρ1)∗)=(eρ2)(eρ1)∗=(eρ2)(eρ2)∗=eρ2.By Lemma2.5(2),ρ1=ρ2.This together withλ1=λ2,yields that(λ1,ρ1)= (λ2,ρ2),as required.Theorem3.2Let S be a type B semigroup.Then so is (S).Proof By Lemma2.6(5), (S)is an adequate semigroup.Now,we prove that (S)is right type B.To see it,let(λ1,ρ1),(λ2,ρ2)∈E[( (S))1],(λ,ρ)∈ (S).Then,by Lemma2.6(4),λ1e,λ2e,eρ1,eρ2∈E(S)for all e∈E(S).Thuse(ρ1ρ)∗(ρ2ρ)∗=[ee(ρ1ρ)∗](ρ2ρ)∗=e(ρ1ρ)∗e(ρ2ρ)∗=(λ1λ)∗e(λ2λ)∗e=(λ1λe)∗(λ2λe)∗=(λ1(λe)+(λe))∗(λ2(λe)+(λe))∗=(λ1(λe)+λ2(λe)+(λe))∗(since S satisfies Condition(B1))=(λ1((λe)+λ2(λe)+)(λe))∗=(λ1(λ2(λe)+(λe)+)(λe))∗=(λ1λ2λe)∗=(λ1λ2λ)∗e=e(ρ1ρ2ρ)∗By Lemma2.5(2),(ρ1ρ)∗(ρ2ρ)∗=(ρ1ρ2ρ)∗.Hence,by Lemma3.1,we have ((λ1λ)∗(λ2λ)∗,(ρ1ρ)∗(ρ2ρ)∗)=((λ1λ2λ)∗,(ρ1ρ2ρ)∗).Thus,[(λ1,ρ1)(λ2,ρ2)(λ,ρ)]∗=(λ1λ2λ,ρ1ρ2ρ)∗=((λ1λ2λ)∗,(ρ1ρ2ρ)∗)=((λ1λ)∗(λ2λ)∗,(ρ1ρ)∗(ρ2ρ)∗)=((λ1λ)∗,(ρ1ρ)∗)((λ2λ)∗,(ρ2ρ)∗)=(λ1λ,ρ1ρ)∗(λ2λ,ρ2ρ)∗=[(λ1,ρ1)(λ,ρ)]∗[(λ2,ρ2)(λ,ρ)]∗,On the translational hull of a type B semigroupthis gives that (S)satisfies Condition (B1).Let (λ1,ρ1)∈E( (S)),(λ,ρ)∈ (S)be such that (λ1,ρ1)≤(λ,ρ)∗.Then (λ1,ρ1)≤(λ,ρ)∗=(λ∗,ρ∗)since (S)is L ∗-unipotent.Hence (λ1,ρ1)=(λ1,ρ1)(λ∗,ρ∗)=(λ1λ∗,ρ1ρ∗),and so λ1=λ1λ∗.By Lemma 2.5(1),λ1e =λ1λ∗e for all e ∈E(S).Note that λ1e,λ∗e ∈E(S).We haveλ1e =λ1λ∗e =λ1λ∗ee =(λ1e)(λ∗e)=(λ∗e)(λ1e).Hence,λ1e ≤λ∗e =(λe)∗.Again,since S satisfies Condition (B2),we have λ1e =[f (λe)]∗for some f ∈E(S 1).That is,λ1e =(λf λe)∗=(λf λ)∗e.By Lemma 2.5(1),λ1=(λf λ)∗.Hence,by Lemma 3.1,(λ1,ρ1)=((λf λ)∗,(ρf ρ)∗)=(λf λ,ρf ρ)∗=[(λf ,ρf )(λ,ρ)]∗,where (λf ,ρf )∈E [( (S))1].That is, (S)satisfies Condition (B2).Therefore (S)is right type B.Dually,we can prove that (S)is left type B.This completes the proof. Corollary 3.3Let S be a type B semigroup .L ∗( (S))denotes L ∗on (S),and μ (S)L [resp .μ (S)R ,μ (S)]denotes μL [resp .μR ,μ]on (S),etc .Then the follow-ing statements are true :(1)for all e ∈E(S),(λ1,ρ1),(λ2,ρ2)∈ (S),(λ1,ρ1)L ∗( (S))(λ2,ρ2)if and onlyif λ1e L ∗(S)λ2e ;(2)for all e ∈E(S),(λ1,ρ1),(λ2,ρ2)∈ (S),(λ1,ρ1)R ∗( (S))(λ2,ρ2)if andonly if eρ1R ∗(S)eρ2;(3)for all (λ1,ρ1),(λ2,ρ2)∈ (S),(λ1,ρ1)μ (S)L (λ2,ρ2)if and only if for all e ∈E(S),λ1eμS L λ2e ;(4)for all (λ1,ρ1),(λ2,ρ2)∈ (S),(λ1,ρ1)μ (S)R (λ2,ρ2)if and only if for all e ∈E(S),eρ1μS R eρ2;(5)for all (λ1,ρ1),(λ2,ρ2)∈ (S),(λ1,ρ1)μ (S)(λ2,ρ2)if and only if for all e ∈E(S),λ1eμS L λ2e and eρ1μS R eρ2.Proof (1)By Theorem 3.2, (S)is a type B semigroup.Let (λ1,ρ1),(λ2,ρ2)∈ (S)be such that (λ1,ρ1)L ∗( (S))(λ2,ρ2).Then,by Lemma 2.6(3),(λ∗1,ρ∗1)L ∗( (S))(λ∗2,ρ∗2).Hence,(λ∗1,ρ∗1)=(λ∗2,ρ∗2)since (S)is L ∗-unipotent.Thus λ∗1=λ∗2.By Lemma 2.5(1),λ∗1e =λ∗2e for all e ∈E(S).Henceλ1e L ∗(S)(λ1e)∗=λ∗1e =λ∗2e =(λ2e)∗L ∗(S)λ2e.Conversely,if for all e ∈E(S),λ1e L ∗(S)λ2e,then (λ1e)∗=(λ2e)∗since S is L ∗-unipotent.That is,λ∗1e =λ∗2e.By Lemma 2.5(1),λ∗1=λ∗2.Hence,by Lemma 3.1,(λ∗1,ρ∗1)=(λ∗2,ρ∗2).Thus,by Lemma 2.6(3),(λ1,ρ1)L ∗( (S))(λ2,ρ2).C.Li,L.-m.Wang(2)This part is the dual of(1).(3)Suppose that(λ1,ρ1),(λ2,ρ2)∈ (S)and(λ1,ρ1)μ (S)L (λ2,ρ2).Then for allf∈E(S),(λf,ρf)(λ1,ρ1)μ (S)L(λf,ρf)(λ2,ρ2),where(λf,ρf)∈E( (S)).Hence,(λf,ρf)(λ1,ρ1)L∗( (S))(λf,ρf)(λ2,ρ2).That is,(λfλ1,ρfρ1)L∗( (S))(λfλ2,ρfρ2).By(1),λfλ1e L∗(S)λfλ2e for all e∈E(S).That is,fλ1e L∗(S)fλ2e.Thus,λ1eμS Lλ2e.Conversely,if for all e∈E(S),λ1eμS Lλ2e,then,for all(λ,ρ)∈E( (S)),f∈E(S),we have(fρ)(λ1e)μS L(fρ)(λ2e).That is,f(λλ1e)μS L f(λλ2e).Hence,f(λλ1e)L∗(S)f(λλ2e).By the definition of μL,we have thatλλ1eμS Lλλ2e.Hence,λλ1e L∗(S)λλ2e.By(1),we have(λλ1,ρρ1)L∗( (S))(λλ2,ρρ2).That is,(λ,ρ)(λ1,ρ1)L∗( (S))(λ,ρ)(λ2,ρ2).Thus(λ1,ρ1)μ (S)L (λ2,ρ2),asrequired.(4)It is the dual of(3).(5)It follows from(3)and(4). Proposition3.4Let S be a type B semigroup and(λ1,ρ1),(λ2,ρ2)∈ (S).Then the following statements are equivalent:(1)(λ1,ρ1)σ (S)(λ2,ρ2);(2)for all e∈E(S),λ1eσSλ2e;(3)for all e∈E(S),eρ1σS eρ2.Proof We note that(1)⇔(2)is the dual of(1)⇔(3).We only need to show that (1)⇔(2).Let(λ1,ρ1),(λ2,ρ2)∈ (S)be such that(λ1,ρ1)σ (S)(λ2,ρ2).Then,by the def-inition ofσ,we have(λ,ρ)(λ1,ρ1)(λ,ρ)=(λ,ρ)(λ2,ρ2)(λ,ρ)for some(λ,ρ)∈E( (S)).That is,(λλ1λ,ρρ1ρ)=(λλ2λ,ρρ2ρ).On the translational hull of a type B semigroupHence,λλ1λ=λλ2λ.By Lemma2.5(1),λλ1λe=λλ2λe for all e∈E(S).That is,λ(λ1λe)+(λ1λe)=λ(λ2λe)+(λ2λe).Hence,[λ(λ1λe)+(λ1λe)]+=[λ(λ2λe)+(λ2λe)]+.Note thatλe,λ(λ1λe)+,λ(λ2λe)+∈E(S)for all e∈E(S)and S is L∗-unipotent.We haveλ(λ1λe)+(λ1λe)+=λ(λ2λe)+(λ2λe)+.That is,λ(λ1λe)+=λ(λ2λe)+.Thus, we haveλλ1λe=λλ2λe=⇒λ(λ1λe)=λ(λ2λe)=⇒λ(λ1λe)+(λ1λe)=λ(λ2λe)+(λ2λe)=⇒λ(λ1λe)+λ1(λe·e)=λ(λ1λe)+λ2(λe·e)=⇒λ(λ1λe)+λ1e·(λe)=λ(λ1λe)+λ2e·(λe)=⇒[(λe)λ(λ1λe)+]λ1e[(λe)λ(λ1λe)+]=[(λe)λ(λ1λe)+]λ2e[(λe)λ(λ1λe)+]for all e∈E(S),and[(λe)λ(λ1λe)+]∈E(S).By the definition ofσ,we have λ1eσSλ2e.Conversely,if for all e∈E(S),λ1eσSλ2e,then there exists f∈E(S)such that f(λ1e)f=f(λ2e)f.Hence,f(λ1f e)=f(λ2f e),and soλf(λ1λf e)=λf(λ2λf e). That is,λfλ1λf e=λfλ2λf e.By Lemma2.5(1),λfλ1λf=λfλ2λf.Thus,by Lemma3.1,(λfλ1λf,ρfρ1ρf)=(λfλ2λf,ρfρ2ρf).That is,(λf,ρf)(λ1,ρ1)(λf,ρf)=(λf,ρf)(λ2,ρ2)(λf,ρf),where(λf,ρf)∈E( (S)).By the definition ofσ,we have(λ1,ρ1)σ (S)(λ2,ρ2), as required.4Some special casesIn this section,we shall consider the translational hulls of some special type B semi-groups,and give a new proof of a problem posted by Petrich on translational hulls of inverse semigroups in[10].Theorem4.1Let S be a proper type B semigroup.Then so is (S).Proof By Theorem3.2, (S)is a type B semigroup.It only remains to show that (S)is proper.To see it,let(λ1,ρ1),(λ2,ρ2)∈ (S)be such that(λ1,ρ1)[L∗ (S)∩σ (S)](λ2,ρ2).Then,by Corollary3.3and Proposition3.4,we haveλ1e L∗(S)λ2e andλ1eσSλ2e for all e∈E(S).Hence,λ1e[L∗S∩σS]λ2e.Again,since S is proper,we haveλ1e=λ2eC.Li,L.-m.Wangfor all e ∈E(S).By Lemma 2.5,λ1=λ2.Therefore,by Lemma 3.1,(λ1,ρ1)=(λ2,ρ2).That is,[L ∗ (S)∩σ (S)]=1 (S).Similarly,we can prove that [R ∗ (S)∩σ (S)]=1 (S).This completes the proof. We now give an example of a proper type B semigroup.Example 4.1Let N be the set of all non-negative integers and S ={(m,n)∈N ×N |m ≥n }.Define a multiplication “•”on S by(m,n)•(p,q)=(m −n +t,q −p +t),where t =max {n,p }.Then,it is easy to check that (S,•)is a semigroup and E(S)={(m,m)∈N ×N }.In fact,we can see easily that S is adequate.Moreover,since for all (m,n)∈S,there exist idempotents (n,n),(m,m)of S such that(m,n)L ∗(n,n)and (m,n)R ∗(m,m).Hence,(m,n)∗=(n,n)and (m,n)+=(m,m).On the other hand,for any (m,m),(n,n)∈E(S 1),(p,q)∈S,we have p ≥q and[(m,m)•(n,n)•(p,q)]∗=[(t,t)•(p,q)]∗=(s,q −p +s)∗=(q −p +s,q −p +s),where t =max {m,n },s =max {t,p }=max {m,n,p }and[(m,m)•(p,q)]∗•[(n,n)•(p,q)]∗=(t 1,q −p +t 1)∗•(t 2,q −p +t 2)∗=(q −p +t 1,q −p +t 1)•(q −p +t 2,q −p +t 2)=(q −p +T ,q −p +T ),where t 1=max {m,p },t 2=max {n,p }and T =max {t 1,t 2}=s.Hence,[(m,m)•(n,n)•(p,q)]∗=[(m,m)•(p,q)]∗•[(n,n)•(p,q)]∗,and so that S satisfies Condition (B1).Let (m,m)∈E(S),(p,q)∈S such that (m,m)≤(p,q)∗.Then p ≥q,and (m,m)≤(p,q)∗=(q,q).Hence,m ≥q ,and so[(p +(m −q),p +(m −q))•(p,q)]∗=(p +m −q,m)∗=(m,m),where (p +(m −q),p +(m −q))∈E(S 1).Thus S satisfies Condition (B2).There-fore,S is right type B.Similarly,we can prove that S is left type B.Next,we prove that L ∗∩σ=1S .To see it,let (m,n),(p,q)∈S be such that (m,n)[L ∗∩σ](p,q).Then m ≥n,p ≥q,(m,n)L ∗(p,q)and (m,n)σ(p,q).Hence,n =q ,and there exists (k,k)∈E(S)such that(k,k)•(m,n)•(k,k)=(k,k)•(p,q)•(k,k).That is,(k,k)•(m,n)•(k,k)=(k,k)•(p,n)•(k,k).Hence,(t,n−m+t)•(k,k)=(s,n−p+s)•(k,k),where t=max{k,m}and s= max{k,p},and so(m−n+T1,T1)=(p−n+T2,T2),where T1=max{n−m+t,k} and T2=max{n−p+s,k}.Thus T1=T2and m−n+T1=p−n+T2,this gives m=p.Therefore,(m,n)=(p,q).That is,L∗∩σ=1S.Dually,we have R∗∩σ=1S.Summing up the above arguments,we conclude that S is a proper type B semi-group.Theorem4.2Let S be a type B semigroup.Then the following statements are true:(1)if S is primitive(i.e.,(∀0=e,f∈E(S))e≤f⇒e=f),then so is (S);(2)if S is E-unitary,then so is (S).Proof(1)By Theorem3.2, (S)is a type B semigroup.We only need to show that (S)is primitive.To see it,let(λ1,ρ1),(λ2,ρ2)∈E( (S))be such that(λ1,ρ1)≤(λ2,ρ2).Then(λ1,ρ1)=(λ1,ρ1)(λ2,ρ2)=(λ2,ρ2)(λ1,ρ1).Hence,λ1=λ1λ2=λ2λ1,and soλ1e=λ1λ2e=λ2λ1e for all e∈E(S).That is,λ1e=λ1(λ2ee)=λ2(λ1ee)=λ2λ1e.Note thatλ1e,λ2e∈E(S)and E(S)is a semilattice.We haveλ1e=(λ1e)(λ2e)=(λ2e)(λ1e).Hence,λ1e≤λ2e.Again,since S is primitive,we haveλ1e=λ2e.By Lemma2.5(1),λ1=λ2.Therefore,by Lemma3.1,(λ1,ρ1)=(λ2,ρ2),as required.(2)We need only show that (S)is E-unitary.To see it,let(λ1,ρ1)∈E( (S)),(λ,ρ)∈ (S)be such that(λ1,ρ1)(λ,ρ)∈E( (S)).That is, (λ1λ,ρ1ρ)∈E( (S)).By Lemma2.6(4),eρ1,eρ1ρ∈E(S)for all e∈E(S).Hence, eρ1ρ=eeρ1ρ=(eρ1)(eρ)∈E(S),this gives eρ∈E(S)since S is E-unitary.Thus, eρ=(eρ)(eρ)=(eρe)ρ=(eeρ)ρ=eρ2.By Lemma2.5(2),ρ=ρ2.Therefore,by Lemma3.1,(λ,ρ)2=(λ,ρ)(λ,ρ)=(λ2,ρ2)=(λ,ρ)∈E( (S)).That is, (S)is left E-unitary.Dually,we can show that (S)is right E-unitary.This completes the proof.Theorem4.3Let S be an adequate semigroup with central idempotents.Then so is (S).Proof It only remains to show that E( (S))is in the center of (S).To see it, let(λ,ρ)∈ (S),(λ1,ρ1)∈E( (S)).Then,by Lemma2.6(4),eρ1∈E(S)for all e∈E(S).Hence,eρ1ρ=(eρ1)(eρ)=(eρ)(eρ1)=eρρ1since E(S)is in the center of S.By Lemma2.5(2),we haveρ1ρ=ρρ1.Thus,by Lemma3.1,(λ1λ,ρ1ρ)= (λλ1,ρρ1).That is,(λ1,ρ1)(λ,ρ)=(λ,ρ)(λ1,ρ1),as required.Corollary4.4The translational hull of a strong semilattice of cancellative monoids is still a strong semilattice of cancellative monoids.Proof It follows from Theorem4.3and Lemma2.4(2). Theorem4.5Let S be a type B semigroup which is fundamental.Then so is (S). Proof By Theorem3.2, (S)is a type B semigroup.Now,we prove that (S)is fun-damental.To see it,let(λ1,ρ1),(λ2,ρ2)∈ (S)be such that(λ1,ρ1)μ (S)(λ2,ρ2). Then,by Corollary3.3(5),(λ1e)μS L(λ2e)and(fρ1)μS R(fρ2)for all e,f∈E(S).Hence,f(λ1e)μS L f(λ2e)and(fρ1)eμS R(fρ2)e.That is,f(λ1e)μS L f(λ2e)and f(λ1e)μS R f(λ2e).Thus,f(λ1e)μS f(λ2e),this gives f(λ1e)=f(λ2e)since S is fundamental.On the other hand,f(λ1e)=f(λ2e)=⇒[f(λ1e)]+=[f(λ2e)]+=⇒[f(λ1e)+]+=[f(λ2e)+]+=⇒f(λ1e)+=f(λ2e)+since S is L∗-unipotent.Choose an idempotent(λ1e)+of S to replace the element f of the above formula.We get that(λ1e)+=(λ1e)+(λ2e)+.Similarly,(λ2e)+= (λ2e)+(λ1e)+.Again,since S is a semilattice,we have(λ1e)+=(λ2e)+.Note that f(λ1e)=f(λ2e)for all e,f∈E(S).We have(λ1e)+(λ1e)=(λ2e)+(λ2e)for all e∈E(S).That is,λ1e=λ2e for all e∈E(S).By Lemma2.5(1),λ1=λ2.Hence,by Lemma3.1,(λ1,ρ1)=(λ2,ρ2).That is,μ (S)=1 (S).This completes the proof.In the remaining,we consider the translational hull of an E-reflexive type B semi-group.We start with the following definition.Definition1A type B semigroup S is called E-reflexive if for all e∈E(S)and x,y∈S,exy∈E(S)implies eyx∈E(S).In fact,it is easy to see that the above definition is a generalization of the E-reflexive inverse semigroup.If the semigroup S is also an inverse semigroup,then S is just an E-reflexive inverse semigroup in the sense of Petrich described in[10, Lemma III.8.2,P.157].Based on the above definition,we now give the following interesting result.Theorem4.6Let S be an E-reflexive type B semigroup.Then so is (S).Proof We will abbreviate the statement“a∈E(S)implies b∈E(S)”by a−→b.Suppose that S is an E-reflexive type B semigroup.Then,by Theorem3.2, (S) is a type B semigroup.Now,we only need to show that (S)is E-reflexive.To see it, let(λ,ρ)∈E( (S)),(λ1,ρ1),(λ2,ρ2)∈ (S)be such that(λ,ρ)(λ1,ρ1)(λ2,ρ2)∈E( (S)).That is,(λλ1λ2,ρρ1ρ2)∈E( (S)).By Lemma 2.6(4),λe,eρ,λλ1λ2e,eρρ1ρ2∈E(S)for all e∈E(S).Hence,for all e1,e,f∈E(S),we haveλλ1λ2e−→e1(λλ1λ2e)=(e1ρ)(λ1λ2e)−→e(e1ρ)(λ1λ2e)=(e1ρ)e(λ1λ2e)=(e1ρ)(eρ1)(λ2e)−→(e1ρ)(λ2e)(eρ1)(since S is E-reflexive)−→(e1ρ)(λ2e)(eρ1)f=(e1ρ)(λ2e)e(λ1f)=(e1ρ)(λ2e)(λ1f)−→(e1ρ)(λ2(λ1f)+)(λ1f)(replace e by(λ1f)+)=(e1ρ)(λ2λ1f)=e1(λλ2λ1f)−→(λλ2λf)+(λλ2λ1f)(replace e1by(λλ2λ1f)+)=λλ2λ1f.That is,λλ2λ1f∈E(S)for all f∈E(S).Similarly,we have fρρ2ρ1∈E(S)for all f∈E(S).By Lemma2.6(4),(λλ2λ1,ρρ2ρ1)∈E( (S)).That is,(λ,ρ)(λ2,ρ2)(λ1,ρ1)∈E( (S)).Therefore, (S)is an E-reflexive type B semigroup.Clearly,an arbitrary inverse semigroup is type B.As a consequence of Theo-rem4.6,we give an affirmative answer to a problem posted by Petrich(i.e.,is the translational hull of an E-reflexive inverse semigroup E-reflexive?(see[10,VII.3.7 Problems,p.323]).The answer is the following corollary.Corollary4.7Let S be an E-reflexive inverse semigroup.Then so is (S).Proof It follows from[10,Corollary V.1.4,P.209]and Theorem4.6.Acknowledgements The authors want to thank Dr.Marcel Jackson for his assistance.The authors also want to express their gratitude to the referees for their valuable suggestions which lead to an improvement of this paper and for making the paper read more easily.References1.Ault,J.E.:The translational hull of an inverse semigroup.Glasg.Math.J.14,56–64(1973)2.El-Qallali,A.,Fountain,J.B.:Idempotent-connected abundant semigroups.Proc.R.Soc.Edinb.A91,79–90(1981)3.Fountain,J.B.:Adequate semigroups.Proc.Edinb.Math.Soc.22,113–125(1979)4.Fountain,J.B.,Lawson,M.:The translational hull of an adequate semigroup.Semigroup Forum32,79–86(1985)5.Fountain,J.B.:A class of right PP monoids.Q.J.Math.,Oxford28(2),285–330(1977)6.Gould,M.:An easy proof Ponizovski’s theorem.Semigroup Forum15,181–182(1977)7.Guo,X.J.,Shum,K.P.:On translational hulls of type-A semigroups.J.Algebra269,240–249(2003)wson,M.V.:The natural partial order on an abundant semigroup.Proc.Edinb.Math.Soc.30,169–186(1987)9.Petrich,M.:The translational hull in semigroups and rings.Semigroup Forum1,283–360(1970)10.Petrich,M.:Inverse Semigroups.Wiley,New York(1984)11.Petrich,M.:Introduction to Semigroups.Merrill,Columbus(1973)12.Ponizovski,I.S.:A remark on inverse p.Mat.Nauk20,147–148(1965)。

  1. 1、下载文档前请自行甄别文档内容的完整性,平台不提供额外的编辑、内容补充、找答案等附加服务。
  2. 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
  3. 3、如文档侵犯您的权益,请联系客服反馈,我们会尽快为您处理(人工客服工作时间:9:00-18:30)。

a rXiv:h ep-th/955179v13M a y1995SOME ABSTRACT PROPERTIES OF SEMIGROUPS APPEARING IN SUPERCONFORMAL THEORIES Steven Duplij ∗†‡Physics Department,University of Kaiserslautern ,Postfach 3049,D-67653KAISERSLAUTERN ,Germany April 5,1995Abstract A new type of semigroups which appears while dealing with N =1superconformal symmetry in superstring theories is considered.The ideal series having unusual abstract properties is constructed.Various idealisers are introduced and studied.The ideal quasicharacter is de-fined.Green’s relations are found and their connection with the ideal quasicharacter is established.KL-TH-95/11hep-th/95051791IntroductionMathematical objects with new properties often appear from concrete physi-cal considerations and models.The discovery of supersymmetry[35,36]gave many new mathematical features,but its influence on the general abstract properties of the theory,in spite of the fact that among principal objects there were noninvertible ones and zero divisors[15],needs to be emphasized. The latter leaded to the conclusion that the abstract ground of supersym-metric theory should have semigroup nature[8].It was also realised that the noninvertible transformations and semigroups appearing in that way have many new nontrivial properties[7,9].In particular,it would be interest-ing to work out the general abstract structure of the N=1superconformal semigroup,which is important in the consistent construction of the super-string unified theories[16,11].In this paper we provide a consideration of the superconformal semigroups from the abstract-algebraic point of view and present their abstract properties without proofs which will appear elsewhere. 2PreliminariesThe semigroup of N=1superconformal transformations of C1,1complex superspace with the coordinates(z,θ)valued in the Grassmann algebra[2], where z∈C1,0andθ∈C0,1,is isomorphic to the semigroup S of the even C1,0→C1,0and odd C1,0→C0,1functions satisfying some multiplication law(for details see[7,9]).The even part of the laws3=s1∗s2,s i∈S,(1) in terms of the even functions g(z)can be presented asg3(z)=[g1( z)+h1( z)]·g2(z),(2) where z is some shifting and h1(z)is some even nilpotent function of second degree,i.e.h21(z)≡h1(z)·h1(z)=0.We stress that,because of the shifting z→ z and the second term in the brackets(2),S differs from the semigroups2of functions with point by point multiplication[6]as well,as from the semi-groups of functions[23,24].This leads to new unusual abstract properties of S considered below.Further we note that to study this properties it is suffi-cient to know the formal expression(2)only.This parametrisation of N=1 superconformal transformations was given in[7,9](where one can alsofind the exact formulas and the concrete background).For other considerations we refer to[1,26,27,14].Here we do not consider the physical interpretations of g(z)(see[1,5]) and stress only that g(z)controls invertibility of the superconformal trans-formations[8].Therefore,the index of g(z)which is defined byind g(z)def= n∈Z|g n(z)=0,g n−1(z)=0 (3) plays a crucial part in the following.We mention here that in(2)and(3)the multiplication is a point by point one in the Grassmann algebra[2](for clarity sometimes we use a point for it),but the star in(1)denotes the semigroup multiplication.So the semigroup S can be divided into two disjoint parts S=G∪T, G∩T=∅,whereG def={s∈S|ind g(z)=∞},(4)T def={s∈S|ind g(z)<∞}.(5) Here G is a group corresponding to the invertible transformations.From the multiplication law(2)it follows that T is a two-sided ideal.The unity element e∈S has g(z)=1,h(z)=0,and the zero element has g(z)=0,h(z)=0 (for other details see[8,9]).From(2)and the relation ind h(z)=2it follows that T is a nilsemigroup[20,12,10,33],i.e.∀t∈T∃n∈Z,t∗n=z(here the multiplication in the power expression is implied as the semigroup one (1)).So every element from T is nilpotent without bound on its index and offinite order,but every element from G is of infinite order.The superconformal transformations corresponding to G were studied earlier in[1,5,28].Therefore we concentrate our attention on the ideal T,which gives the evidence of some unusual abstract properties of such parametrised superconformal semigroup S.33Ideal seriesTo classify the elements from the ideal part T we take n-th power of the equation(2)in the Grassmann algebra and,using the relation ind h(z)=2, obtaing n(z)= g n1(˜z)+n·g n−11(˜z)·h1(˜z) ·g n2(z).(6)3We see that the natural classification can be done by means of the index of g(z)(see(3)).Let us define the following sets∆n def={s∈S|ind g(z)=n}.(7)I n def= k≤n∆k.(8)Then we notice that T is a disjoint union of the sets∆n,because T= n∆n,∆n∩∆n−1=∅.From(6)it follows that I n−1⊂I n and I n\I n−1=∆n. Therefore we obtain the following infinite chain of the sets I nz⊂I1⊂I2⊂...⊂I n⊂...⊂T.(9) To understand the meaning of I n we use(6)and obtainS∗I n⊆I n,(10)I n∗S⊆I n+1,(11)S∗I n∗S⊆I n+1.(12) From these relations we can easily observe that the sets I n are left ideals of the semigroup S,but not right ideals,because of(11).Moreover,the appearance of n+1in the right side of(11)and(12)is very unusual,and so these strange sets I n is natural to call”jumping ideals”.Therefore I n−1¡l I n and the chain(9)is a left and”jumping”ideal series.Then I n are quasiideals [32,4]since they satisfy S∗I n∩I n∗S⊆I n.Simultaneously,the sets I n are biideals,because I n∗S∗I n⊆I n[3,19].It is exciting that in our case the regularity is not necessary for coincidence quasiideals and biideals in superconformal semigroup(as distinct from[19]).Because of the inclusion I n¡U⇒I n¡S,∀U⊂S the semigroup S is afilial semigroup[17].The indices in(9)form a well ordered set for which n is an ordinal.Because of4I n−1¡l I n the chain(9)can be called a left ascending ideal series of S.From(11)and(12)we deriveS∗I n∪I n∗S⊆I n+1,(13) This condition is opposite for the chain(9)to be an ascending annihilator series of S[13,30].So we call it an ascending antiannihilator series of S.The multiplication law for the sets I n and∆n isI n∗I n+k⊆I n+1,I n+k−1∗I n⊆I n,∆n∗∆n+k⊆I n+1,∆n+k−1∗∆n⊆I n,I n∗∆n+k⊆I n+1,I n+k−1∗∆n⊆I n,(14)∆n∗I n+k⊆I n+1,∆n+k−1∗I n⊆I n,I n∗G⊆I n+1,G∗I n⊆I n,∆n∗G⊆I n+1,G∗∆n⊆∆n.where k>0.It follows that the set I n is a subsemigroup of S,because from(14)we have I n∗I n⊆I n but the set∆n is not a subsemigroup,since ∆n∗∆n⊆I n.This is a consequence of the fact that our semigroup is defined over the Grassmann algebra[2]which contains nilpotents and zero divisors, and the latter fact should be taken into account properly[15].From the last two relations of(14)and(12)we can obtainG∗∆n∗G⊆I n+1,(15) i.e.some of the elements from∆n are conjugated by the subgroup G with the elements of the next set∆n+1.By analogy with[34,22,21]we define G-normal subsets A,B⊆S as follows g−1∗A∗g⊆B,g∈G.Then from (15)we make a conclusion that any two sets∆n contain G-normal elements and one can reach any∆n using the subgroup action only.Further general abstract properties of such elements can be found in[22,29].54IdealisersThe left(right,two-sided)idealiser I l(U)(I r(U),I(U))of the subset U⊆S can be defined as the largest subsemigroup of S within which U is a left (right,two-sided)ideal,i.e.I l(U)def={s⊆S|s∗U⊆U},(16)I r(U)def={s⊆S|U∗s⊆U},(17)I(U)def={s⊆S|s∗U⊆U,U∗s⊆U}.(18) For the set I(U)the set U is really a subsemigroup,because U∗s⊆U, s∗U⊆U,U∗t⊆U,t∗U⊆U⇒U∗s∗t⊆U∗t⊆U,s∗t∗U⊆s∗U⊆U.Also,if V is a subsemigroup of U and V¡U,then∀v∈V⇒v∗U⊆U,U∗v⊆U⇒v∈I(U).Thus V⊆I(U).Let us consider the idealisers of the various introduced subsets of S.First the left idealiser for I n is S,as is follows directly from(10),i.e.I l(I n)=S.(19)From the last relation in(14)wefindI l(∆n)=G.(20)For the right idealisers of I n the situation is more ing(11) we divide S into two disjoint parts S=S I n∪S∆n,where S I n∩S∆n=∅,and they satisfy the relationsI n∗S I n⊆I n,(21)I n∗S∆n⊆∆n+1.(22)By definition(17)S I n is the right idealiser for I n,i.e.I r(I n)=S I n.(23)Obviously,that I n⊂S I n,since I n∗I n⊂I n.Therefore S I n=I n∪S II n. From(6)it follows that for the elements from S II n the second term in the brackets should disappear,therefore wefindS II n= s∈T\I n|g n−11( z)·g n2(z)=0,h1( z)·g n2(z)=0 .(24)6Then the”jumping”set S∆n from(22)is equal to S∆n=(S\In )\S II n.Another way to vanish the second term in(6)is the consideration of the special superconformal transformations(they are called Ann-transforma-tions in[9])for which the relation g n−1(z)·h(z)=0is valid(see(2) and(6)).Let us divide I n in two disjoint parts I n=I A n∪I=A n,where I A n def={s∈I n|g n−1(z)·h(z)=0}.It was shown in[9]that Ann-property is preserved from the right only,and so we obtain I A n∗S⊆I A n,which means that I A n is a right ideal in S,thenI r I A n =S.(25)For the sets∆A n=I A n\I A n−1wefind∆A n∗G⊆∆A n,thereforeI r ∆A n =G.(26)We note here that by means of the right group action we can reach a set I n with any large n,because the relation∆=A n∗G⊆∆=A n+1(see also (15)).5Ideal quasicharacterLet us defineχ(s)def={n∈N|ind g(z)=n}.(27) Using(10)and(11)we obtainmaxχ(s∗t)= χ(t),χ(s)≥χ(t)χ(s)+1,χ(s)<χ(t).(28)In particular,χ(g∗s)=χ(s),χ(s∗g)=χ(s)+1.(29)From(28)it follows that nδ=|χ(s∗t)−χ(s)−χ(t)|is bounded.This value nδshows how much the mapping s→χ(s)differs from a homomor-phism[18].The limitedness of nδallows us to conclude thatχ(s)is a qua-sicharacter[31]which can be called an ideal quasicharacter.The elements of S havingfinite ideal quasicharacter are nilpotent and belong to the ideal T,7andχ(g)=∞,g∈G.Another description of the ideal quasicharacter can be written as followsχ(s)=n⇐⇒s∈∆n.Since∆n∩∆m=∅,n=m, we conclude thatχ(s)indeed disjoins the elements of S,and the relationπdefined as sπt⇐⇒χ(s)=χ(t)is an equivalence relation in S.6Green’s relationsIn our notations the Green’s L and R relations ares L t⇐⇒∃u,v∈S,u∗s=t,v∗t=s,(30)s R t⇐⇒∃u,v∈S,s∗u=t,t∗v=s.Let usfind L and R equivalent elements in the superconformal semigroup ing(10)and(28)wefind that s L t⇒χ(s)≤χ(t)∧χ(t)≤χ(s)⇒χ(s)=χ(t).Therefore L=π,and L-equivalent elements have the same ideal quasicharacter,s L t⇒χ(s)=χ(t),(31) and they belong to the same set∆n.By analogy from(11)for the R-equivalent elements we derive s R t⇒χ(s)≤χ(t)+1∧χ(t)≤χ(s)+1. Then the ideal quasicharacters of the R-equivalent elements can differ only by1or coincide,i.e.s R t⇒|χ(s)−χ(t)|≤1.(32) Since H=L∩R,the sets∆n consist also of H-equivalent elements.Consider the L-equivalent elements.Let s=t,s=z,t=z.From(30) we derive that s=v∗(u∗s)=(v∗u)∗s=(v∗u)∗k∗s for any k∈N.If v∈T∨u∈T,then(v∗u)∗k∈T,since T is an ideal in S.Because of T is a nilsemigroup∃n∈N such that(v∗u)∗n=z.Through the arbitrariness of k we choose k=n and obtain s=(v∗u)∗n∗s=z∗s=z or s=t,which contradicts the initial assumptions.The same is valid for other Green’s relations.Therefore v∈G∧u∈G,i.e.nontrivial L and R equivalences can be constructed with regard to the invertible elements of S only.Then the principal left and right ideals generated by∀t∈S and defined by L(t)def= S∗t and R(t)def=t∗S,as a matter of fact are some analogies of the left and right cosets of G in S introduced in[25,29].87AcknowledgementsThe author is grateful to Prof.J.M.Howie for fruitful conversations and remarks and the kind hospitality at the University of St.Andrews,where the work was begun.Also the discussions with D.A.Arinkin,V.G.Drinfeld, J.Kupsch,wson,B.V.Novikov,W.R¨u hl,S.D.Sinelshchikov and J.Wess are greatly acknowledged.References[1]M.A.Baranov,I.V.Frolov,and A.S.Schwarz,Geometry of supercon-formalfield theories in two dimensions,Theor.Math.Phys.70(1987), 92–103.[2]F.A.Berezin,Introduction to Superanalysis,Reidel,Dordrecht,1987.[3]F.Catino,On bi-ideals in eventually regular semigroups,Riv.Mat.PureAppl.4(1989),89–92.[4]A.H.Clifford,Remarks on0-minimal quasi-ideals in semigroups,Semi-group Forum16(1978),183–196.[5]L.Crane and J.M.Rabin,Super Riemann surfaces:uniformization andTeichm¨u ller theory,Commun.Math.Phys.113(1988),601–623. [6]A.Cs´a sz´a r and E.Th¨u mmel,Multiplicative semigroups of continuousmappings,Acta Math.Hung.56(1990),189–204.[7]S.Duplij,On N=4super Riemann surfaces and superconformal semi-group,J.Phys.A24(1991),3167–3179.[8]———,On semigroup nature of superconformal symmetry,J.Math.Phys.32(1991),2959–2965.[9]———,Ideal structure of superconformal semigroups,University ofKaiserslautern,preprint KL-TH-95/4,CERN-SCAN-9503193,1995. [10]G.U.Garba,Nilpotents in semigroups of partial one-to-one transfor-mations,Semigroup Forum48(1994),37–49.9[11]M.B.Green,J.H.Schwarz,and E.Witten,Superstring Theory,Vol.1,2,Cambridge Univ.Press,Cambridge,1987.[12]P.A.Grillet,A construction offinite commutative nilsemigroups,Com-mun.Algebra19(1991),3145–3172.[13]I.L.Hmelnitsky,On semigroups with the idealizer condition,SemigroupForum32(1985),135–144.[14]L.Hodgkin,On metrics and super-Riemann surfaces,Lett.Math.Phys.14(1987),177–184.[15]J.A.Huckaba,Commutative Rings with Zero Divisors,Dekker,NewYork,1988.[16]M.Kaku,Introduction to Superstrings,Springer-Verlag,Berlin,1988.[17]W.B.Vasantha Kandasamy,Filial semigroups and semigroup rings,Libertas Math.12(1992),35–37.[18]D.Kazhdan,Israel J.Math.43(1982),315–323.[19]jos,Biideals in semigroups.I.A survey,Pure Appl.Math.A2(1992),215–237.[20]llement,On nilpotency in semigroups,Pacific J.Math.42(1972),693–700.[21]I.Levi,Green’s relations on G-normal semigroups,Louisville preprint,1992.[22]I.Levi and W.Williams,Normal semigroups of partial one-to-one trans-formations,2,Semigroup Forum43(1991),344–356.[23]K.D.Magill,Semigroups of functions on topological spaces,Proc.London Math.Soc.16(1966),507–518.[24]———,A survey of semigroups of continuous selfmaps,SemigroupForum11(1975),1–189.[25]D.B.McAlister,Characters of commutative semigroups,Quart.J.Math.19(1968),141–157.10[26]J.M.Rabin,Status of the algebraic approach to super Riemann surfaces,in Physics and Geometry,(L.-L.Chau and W.Nahm,eds.),Plenum Press,New York,1991,pp.653–668.[27]A.Rogers,Super Riemann surfaces,in The Interface of Mathematicsand Particle Physics,(D.G.Quillen,G.B.Segal,and S.T.Tson,eds.), Clarendon Press,New York,1990,pp.87–96.[28]A.A.Rosly,A.S.Schwarz,and A.A.Voronov,Geometry of supercon-formal manifolds,Commun.Math.Phys.119(1988),129–152. [29]B.M.Schein,Cosets in groups and semigroups,in Semigroups WithApplications,(J.M.Howie,W.D.Munn,and H.J.Weinert,eds.),World Sci.,River Edge,1992,pp.205–221.[30]L.N.Shevrin,On two longstanding problems concerning nilsemigroups,in Semigroups With Applications,(J.M.Howie,W.D.Munn,and H.J.Weinert,eds.),World Sci.,River Edge,1992,pp.222–235.[31]A.I.Shtern,Quasirepresentations and pseudorepresentations,Func.Anal.Appl.25(1991),140–143.[32]O.Steinfeld,Quasi-ideals in Rings and Semigroups,Akad´e miai Kiado,Budapest,1978.[33]R.P.Sullivan,Semigroups generated by nilpotent transformations,J.Al-gebra110(1987),324–343.[34]J.S.V.Symons,Normal transformation semigroups,J.Austr.Math.Soc.A22(1976),385–390.[35]D.V.Volkov and V.P.Akulov,On the possible universal neutrinointeraction,JETP Lett16(1972),621–624.[36]J.Wess and B.Zumino,Supergauge transformations in four dimensions,Nucl.Phys.B70(1974),39–76.11。

相关文档
最新文档