# Full text of "An asymptotic behavior of the dilatation for a family of pseudo-Anosov braids"

## See other formats

An asymptotic behavior of the dilatation for a family of pseudo-Anosov braids Eiko Kin* and Mitsuhiko Takasawa l> § : February 2, 2008 (N >. O . Abstract. The dilatation of a pseudo-Anosov braid is a conjugacy invariant. In this paper, we ^j ' study the dilatation of a special family of pseudo-Anosov braids. We prove an inductive formula Q>^ ■ to compute their dilatation, a monotonicity and an asymptotic behavior of the dilatation for this family of braids. We also give an example of a family of pseudo-Anosov braids with arbitrarily small dilatation such that the mapping torus obtained from such braid has 2 cusps and has an f-H ' arbitrarily large volume. o Keywords: mapping class group, braid, pseudo-Anosov, dilatation Mathematics Subject Classification : Primary 37E30, 57M27, Secondary 57M50 1 Introduction ^ ■ Let S = Tig^p be an orientable surface of genus g with p punctures, and let A^(E) be the mapping On , class group of S. The elements of A1(S), called mapping classes, are classified into 3 types: periodic, reducible and pseudo-Anosov [10]. For a pseudo-Anosov mapping class (p, the dilatation A(0) is ^f) • an algebraic integer strictly greater than 1. The dilatation of a pseudo-Anosov mapping class is a conjugacy invariant. Let Dn be an n-punctured closed disk. The mapping class group M{Dn) of D„ is isomorphic to a subgroup of A^($]o,n+i)- There is a natural surjective homomorphism T:Bn^M{Dn) r> I from the n-braid group Bn to the mapping class group A^(D„) [2]. We say that a braid P G B^ is C^ ' pseudo-Anosov if r(/3) is pseudo-Anosov, and if this is the case the dilatation A(/3) of /3 is defined equal to A(r(/3)). Henceforth, we shall abbreviate 'pseudo-Anosov' to 'pA'. We now introduce a family of braids. Let P(mi,m2,---,nn.+-i) be the braid as depicted in Figured! for each integer A; > 1 and each integer m-j > 1. These are all pA (Proposition 14. 1[ ). We will prove a monotonicity, an inductive formula to compute their dilatation and an asymptotic behavior of the dilatation for this family of braids. Proposition 1.1. (Monotonicity) For each integer i with 1 <i <k + 1, we have *Tho first author is partially supported by Grant-in-Aid for Young Scientists (B) (No. 17740094), The Ministry of Education, Culture, Sports, Science and Technology, Japan Figure 1: (left) /?(mi,m2,-,mfc+i), (center) /3(2,2,3), (right) /3(3^2)- Hence if rm < m'- for each i, then \{l3(rni^- ,rak+i)) > KP(m[,- ,m'^+i))- For an integral polynomial f{t) of degree d, the reciprocal of f{t), denoted by /*(t), is t'^f{l/t). Theorem 1.2. (Inductive formula) The dilatation of the pA 6raid /3(mi,---,mfc+i) ^^ the largest root of the polynomial ifc+i R (mi,---,mfc hit), t""'+'R^m„-,m,){t) + {-l where R(mi,-- ,mi){t) is given inductively as follows: R^^^){t) = t"'^+'^{t-l)-2t, and R{m,,-,m,){t) = t"''it-l)Rirm,-,m,.,){t) + {-^Y'^tR^m,,- ,m^.^),{t) for 2 < i < k. Theorem 1.3. (Asymptotic behavior) We have lim X(0(^. ... rr,. ,_-^)) = 1 and (1) lim A(/3(^,, mi,--- ,mj;_|_i— >oo ^ ,"ifc+l (^) lim A(/3(„ )) = A(i?(^ the maximal absolute value of the roots of f{t). i)(0) > 1 /o*" i > 2, where \{f{t)) denotes For a pA braid /3, let be the pA mapping class r(/3). The dilatation A(</>) can be computed as follows. A smooth graph r, called a iram irac/c and a smooth graph map (p : t ^ t are associated with (^. The edges of r are classified into real edges and infinitesimal edges, and the transition matrix M^caii'P) with respect to real edges can be defined. Then the dilatation A((/>) equals the spectral radius of Mi.eai(0)- For more details, see Section [2^ For the computation of the dilatation of the braid f3(mi , ."ife+i)' we introduce combined trees and combined tree maps in Section [3j For a given (mi, • • • ,772.^+1), one can obtain the combined tree 2(mi,--- ,mfc+i) and the combined tree map q(^rni,--- ,nn.+i) inductively. For example, for {mi,m2-,m^) = (4,2,1), the combined tree Q(jni,m2,m3)i depicted in Figure [21 is obtained by gluing the combined tree Q(mi,m2) ^^id another tree which depends m^. The combined tree map 9(mi,m2,m3)' ^^ shown in Figure O is defined by the composition of an extension of the combined tree map q(mi,m2) ^^id another tree map which depends on 7713. By the proof of Proposition 14.11 it turns out that the spectral radius of the transition matrix M{q{mi,-,mk+i)) obtained from q^mi,- ,mk+i) equals that of Mreai(0), where (j) = T{f3(^rau- ,mk+i)), -• 1 • gluing /\ -*- •- * • • • Figure 2: Qu^2,i) (right) is obtained by gluing Q(4,2) (left) and another tree (center) '(4,2) 1 5 -• — - — f • 2 1 -7 • •- • > 1 5 9 10 -• • • >- » "^(4,2.1) 'Y 5/ 10 2 1 7 m • 5). 2 1 ^J-^l g 8 10 r Figure 3: (top) ^(4,2)) (bottom) ^(4 2,1)- that is the spectral radius of M(g(-„j ... „j^^^)) equals the dilatation A(/3(^^ ... ,„^^^)). Proposition ll.il and Theorems [L2l 13] will be shown by using the properties of combined tree maps. In the final part, we will consider the two invariants of pA mapping classes, the dilatation and the volume. Choosing any representative / : S — > S of a mapping class 0, we form the mapping torus ¥(</.) = Sx [0,1]/ ~, where ~ identifies (x,0) with (/(x),l). A mapping class <j) is pA if and only if T((/)) admits a complete hyperbolic structure of finite volume [7J. Since such a structure is unique up to isometry, it makes sense to speak of the volume vol(0) of (j), the hyperbolic volume of T((/>). For a pA braid /?, we define the volume vol(/3) as equal to vol(r(/3)), the volume of the mapping torus T(r(/3)). Theorem 11.3( 1) tells us that dilatation of braids can be arbitrarily small. We consider what happen for the volume of a family of pseudo-Anosov mapping classes whose dilatation is arbitrarily small. It is not hard to see the following. Proposition 1.4. There exists a family of pA mapping classes </>„ of A4{Dn) such that lim \{(j)n) = 1 Oind lim vol(0„) = co n— >oo n^oo and such that the number of the cusps of the mapping torus T ((/>„) goes to oo as n goes to oo. Proposition 11.41 is not so surprising, because the volume of each cusp is bounded below uniformly. We show the following. Proposition 1.5. There exist a family of pA mapping classes <j)n of Ad{Dn) such that lim \{(j)n) = 1 CLnd lim vol((/)„) = oo n-^oo n— >oo and such that the number of the cusps of the mapping torus T ((/>„) is 2 for each n. Proposition 11.51 is a corollary of the following theorem. Theorem 1.6. For any real number A > 1 and any real number t> > 0, there exist an integer k >! and an integer m > 1 such that for any integer rrii > m with 1 < i < k + 1, we have A(/3{mi,-,mfe+i)) < A and vol(/?(„^^...^„^^j)) > v. Here we note that for a braid b, the mapping torus T(r(6)) is homeomorphic to the link complement S''^ \ 6 in the 3 sphere S"^, where 6 is a union of the closed braid of b and the braid axis (Figure H]). When 6 is a braid /?(mi,---,mfe+i)) the link b has 2 components, and hence the number of cusps of T(r(6)) is 2. 2 Preliminaries A homeomorphism $ : E ^ S is pseudo-Anosov {pA) if there exists a constant A = A(<1>) > 1, called the dilatation of^, and there exists a pair of transverse measured foliations JF'' and JF" such that $(^«) = i^^ and $(/•") = \J^. A Figure 4: link b. A mapping class (p ^ A1($]) is said to be pseudo-Anosov (pA) if (p contains a pA honieoniorphism. We define the dilatation of a pA mapping class (j), denoted by A(0), to be the dilatation of a pA homeomorphism of (j). Let ^ be a graph. We denote the set of vertices by V{G) and denote the set of edges by E{Q). A continuous map g : Q ^f Q' from Q into another graph Q' is said to be a graph map. When Q and Q' are trees, a graph map g : Q ^ Q' is said to be a tree map. A graph map g is called Markov if giy{Q)) C y{G') and for each point x (z G such that g{x) ^ V{G'), g is locally injective at x (that is g has no 'back track' at x). In the rest of the paper we assume that all graph maps are Markov. For a graph map g, we define the transition matrix M{g) = (rriij) such that the i^^ edge e[ or the same edge with opposite orientation (e^"^ of Q' appears mjj-times in the edge path g{ej) for the j edge ej of Q. If ^ = G', then M{g) is a square matrix, and it makes sense to consider the spectral radius, X{g) = X{M{g)), called the growth rate for g. The topological entropy of g is known to be equal to logA(5f). In Section 12.11 we recall results regarding Perron- Frobenius matrices. In Section 12.21 we quickly review a result from the train track theory which tells us that if a given mapping class (j) induces a certain graph map, called train track map, whose transition matrix is Perron-Frobenius, then (p is pA and X{(j)) equals the growth rate of the train track map. In Section [2. 31 we consider roots of a family of polynomials to study the dilatation of pA mapping classes and give some results regarding the asymptotic behavior of roots of this family. 2.1 Perron-Frobenius theorem Let M = {niij) and N = (nij) be matrices with the same size. We shall write M > N (resp. M > N) whenever ttijj > njj (resp. ruij > Uij) for each i,j. We say that M is positive (resp. non-negative) if M > (resp. M > 0), where is the zero matrix. For a square and non-negative matrix T, let A(T) be its spectral radius, that is the maximal absolute value of eigenvalues of T. We say that T is irreducible if for every pair of indices i and j, there exists an integer k = kij > such that the (i, j) entry of M'^ is strictly positive. The matrix T is primitive if there exists an integer k > such that the matrix T is positive. By definition, a primitive matrix is irreducible. A primitive matrix T is Perron-Frobenius, abbreviated to PF, if T is an integral matrix. For M > T, if T is irreducible then M is also irreducible. The following theorem is commonly referred to as the Perron-Frobenius theorem. Theorem 2.1. that Let T be a primitive matrix. Then, there exists an eigenvalue A > of T such (1) A has strictly positive left and right eigenvectors x and y respectively, and (2) A > |A'| for any eigenvalue A' 7^ A ofT. If T is a PF matrix, the largest eigenvalue A in the sense of Theorem 12.11 is strictly greater than 1, and it is called the PF eigenvalue. The corresponding positive eigenvector is called the PF eigenvector. The following will be useful. Lemma 2.2. |3 Theorem 1.6, Exercise 1.17] Let T he a primitive matrix, and let s be a positive number. Suppose that a non-zero vector y > satisfies Ty > sy. Then, (1) X> s, where A is the largest eigenvalue ofT in the sense of Theorem \2.1i and (2) s = X if and only if Ty = sy. Proof. (1) Let x be a positive left eigenvector of T. Then, xTy = Axy > sxy. Hence we have A > s. (2) ('Only if part) Suppose that s = X, and suppose that Ty > Ay and Ty 7^ Ay. Premultiplying this inequality by a positive left eigenvector x of T, we have xTy(= Axy) > Axy. Hence A > A, which is a contradiction. ('If part) Suppose that Ty = sy. Premultiplying this equality by a positive left eigenvector x of T, we obtain A = s. D For a non-negative k x k matrix T, one can associate a directed graph Gt as follows. The graph Gt has vertices numbered 1, 2, • • • ,k and an edge from the j vertex to the i vertex if and only if the (i, j) entry Tij 7^ 0. By the definition of Gt, one easily verifies the following. Lemma 2.3. Let T be a non-negative square matrix. (1) T is irreducible if and only if for each i,j, there exists an integer Uij > such that the directed graph Gt has an edge path of length Uij from the j vertex to the i vertex. (2) T is primitive if and only if there exists an integer n > such that for each i,j, the directed graph Gt has an edge path of length n from the j vertex to the i vertex. 2.2 Train track maps A smooth branched 1-manifold r embedded in Dn is a train track if each component of D„ \ t is either a non-punctured k-gon {k > 3), a once punctured A;-gon (k > 1) or an annulus such that a boundary component of the annulus coincides with the boundary of Dn and the other component has at least 1 prong. A smooth map from a train track into itself is called a train track map. Let / : Dn -^ Dn be a homeomorphism. A train track r is invariant under / if /(r) can be collapsed smoothly onto r in Dn- In this case / induces a train track map f : t ^ t. An edge of T is called infinitesimal if there exists an integer A^ > such that /'^(t) is a periodic edge under /. An edge of r is called real if it is not infinitesimal. The transition matrix of / is of the form: where Mj-eail/) (resp. Minf(/)) is the transition matrix with respect to real (resp. infinitesimal) edges. The following is a consequence of [IJ. Proposition 2.4. A mapping class (p € A4(D„) is pA if and only if there exists a homeomorphism f : Dn — > Dn of (p and there exists a train track r such that r is invariant under f , and for the induced train track map f : t ^ t, the matrix Mi-ealC/) is PF. When (f) is a pA mapping class, we have X{ct)) = A(Mreai(/)). 2.3 Roots of polynomials For an integral polynomial S{t), let X{S{t)) be the maximal absolute value of roots of S{t). For a monic integral polynomial R{t), we set Qn,±{t) = rR{t)±S{t) for each integer n > 1. The polynomial R{t) (resp. S{t)) is called dominant (resp. recessive) for a family of polynomials {Qn,±{t)}n>i- In case where S{t) = R^{t), we call t^R{t) ± R^(t) the Salem-Boyd polynomial associated to R{t). E. Hironaka shows that such polynomials have several nice properties [3,, Section 3]. The following lemma shows that roots of Qn,±{t) lying outside the unit circle are determined by those of R{t) asymptotically. Lemma 2.5. Suppose that R{t) has a root outside the unit circle. Then, the roots of Qn,±{t) outside the unit circle converge to those of R{t) counting multiplicity as n goes to oo. In particular, A(i?(t))=lim„^ooA(Q„,±(t)). The proof can be found in [3j. We recall a proof here for completeness. Proof. Consider the rational function Let ^ be a root of R{t) with multiplicity m outside the unit circle. Let Dg be any small disk centered at 6 that is strictly outside of the unit circle and that contains no roots of R{t) other than 6. Then, \R{t)\ has a lower bound on the boundary dDg by compactness. Hence there exists a number ng > depending on 6 such that |-R(t)| > \^^\ on dDg for any n > uq. By the Rouche's theorem, it follows that R{t) and R{t) it -^ (hence R{t) and Qn.±{t)) have the same m roots in Dq. Since Dq can be made arbitrarily small and there exist only finitely many roots of R{t), the proof of Lemma 12.51 is complete. D Lemma 2.6. Suppose that R{t) has no roots outside the unit circle, and suppose that Qn,±{t) has a real root Hn greater than 1 for sufficiently large n. Then, linin-,oo l^-n = 1- Proof. For any e > 0, let D^ be the disk of radius 1 + e around the origin in the complex plane. Then, for any sufficiently large n, we have |-R(i)| > |-^| for all t on dD^. Moreover, R{t) and zb^^- are holomorphic on the complement of Dg in the Riemann sphere. By Rouche's theorem, R{t) and R{t) ± -^ (hence R{t) and Qn,±{t)) have no roots outside -D^. Hence ^„. converges to 1 as n goes to oo. D 3 Combined tree maps For an n X n matrix M, let M{t) be the characteristic polynomial \tl — M\ of M, where I the n X n identity matrix. Let M^,{t) be the reciprocal polynomial of M{t). Then, In is M4t) = e -I-M tM\ that is M^,(t) equals the determinant of the matrix / — tM. This section introduces combined tree maps. Given two trees we combine these trees with another tree of star type having the valence n + 1 vertex and define a new tree, say Q„. When two tree maps on Q„ satisfy certain conditions (L1,L2,L3 and R1,R2,R3), we can define the combined tree map q^ on Qn and obtain a family of tree maps {qn : Qn -^ Qn}n>i- In Section [3T] we give a sufficient condition that guarantees M{qn) is PF. In Section [3. 21 we consider combined tree maps in a particular setting. Then, we give a formula for M{qn){t) and M{qn)^{t) and analyze the asymptotic behavior of the growth rate for qn. This analysis will be applied to train track maps in Section HI 3.1 Transition matrices and growth rate We assume that all trees are embedded in the disk D. By the trivial tree Tq, we mean the tree with only one vertex. Let Gn,+ and Gn,- be trees of star type as in Figure O having one vertex of valence n + 1. w, w„ Figure 5: trees (left) Gn,+ and (right) Gn,- having one vertex of valence n + 1. Let Gl (resp. Gr) be a tree (possibly a trivial tree) with a valence 1 vertex, say vl (resp. vji). Let wl and wr be vertices of Gn,+ as in Figure [5l and glue Gl, Gn,+ and Gr together so that for S G {L,R}, vs and ws become one vertex (Figure [6]). The resulting tree Qn,+ is called the combined tree, obtained from the triple {Gl, Gn,+ , Gr)- We define the combined tree Qn,-, obtained from the triple {Gl, Gn.-, Gr) in the same manner. Before we define combined tree maps on Q. ■n,+/-. we label the edges of Q rh+f- Let i be the number of edges of Gl, and let r be the number of edges of Gr plus 1. Note that the number of edges of Q„__|_/_ is ^ + n + r. JV J I Gi Gn,+ Gii Figure 6: combined trees Qn,+'- (left) general case, (right) case where Qr is a trivial tree. • The edges of 0^,+/- ^-^e numbered £+lto£ + n + lin the clockwise/counterclockwise direction as in Figure [71 • The edge of Ql sharing a vertex with the {i + ly^ edge is numbered i and the remaining edges of Ql are numbered 1 to £ — 1 arbitrarily. • The edge of Gr sharing a vertex with the (£ + n + 1)*^* edge is numbered i+n + 2 and the remaining edges of Gr are numbered i + n + 3 to l + n + r arbitrarily. The edge numbered i is denoted by Cj. Now we take a tree map ql '■ Qn,+/- — *■ Qn,+/- satisfying the following conditions. LI The map ql restricted to the set of vertices of -E'(Q„ _(_/_) \ [E{Ql) U {e^+i}) is the identity. L2 QLiQL) C Gl- L3 The edge path gL{ei+i) passes through e^+i only once and passes through e£. Next, we take a tree map qr : Qn^+/- -^ Qn,+/- satisfying the following conditions. Rl The map gR restricted to the set of vertices of -E(Q„,+/-) \ {E{Qr) U {e£+„+i}) is the identity. R2 gR{gR) C Gr. R3 The edge path 5R(e^+„+i) passes through e£+„+i only once and passes through e^+n+2- Finally, we define the tree map g„ : Qn,+/- — *• Qn,+/- satisfying the following conditions. nl The map gn restricted to the set of vertices of E{Q,^^_^i_) \ {E{Q^^_^i_) U {e£,ee^n+2}) is the identity. n2 gn rotates the subtree 0^.,+/- as in Figure [71 n3 The image of each e S {ee, et+n+2} is as in Figured The length of the edge path (7„(e) is 3. The composition Qn = gRgnQL ■ Qn,+/- ^ 2n,+/- is called the combined tree map, obtained from the triple igL,gn,gR)- It makes sense to consider the transition matrices M{gs) of gs\gs ■ Gs ^ Gs, S £ {L,R} and M{gn) of 5nb„,+/_ : ^n,+/- ^ Gn,+/-- The transition matrix M{qn) has the following form: Ml A \ ' * ^ M[qn)=\ B Mn C ,whereM„=| "•• | (3.1) D E Mr I 9 <^ / l+l l+n+1 l+n+2 , 1+2' '+"+2 1+2 l+n+2 Figure 7: (top) gn rotates Qn,+-, (bottom) g^ rotates Gn,-- The edges e^ and e^+„+2 and their images are drawn in bold. (each empty space in M„ represents the number 0), and the block matrices satisfy Ml > M^g^), Mn > M{gn) and Mr > M{gji). (In fact Ml = M{gL), although we will not be using this fact.) Throughout this subsection, we assume that the trees Ql and Qr are not trivial. It is straight- forward to see the following from the defining conditions of gi, gn and gR- Lemma 3.1. Let rriij be the {i,j) entry of M{qn). We have (1) me+n,i+n+2 = 1 and mi+n+i/+n+2 = 1; and (2) nii^i+i > 0, m^+1,^+1 > and mi+n+i,£+i > 1- Moreover, m^+ij 1 < j < ^ and m^+ij„ > for some I < jo < i, and (3) mi+n+2,i+l > 0. mi+n+i,j for each j with An important feature is that the growth rate of g„ is always greater than 1 if M^gi) and M^gpt) are irreducible, which will be shown in Proposition 13.31 We first show that M{qn) is irreducible in this case. Notice that M(gn) is always irreducible, and since M„ > M(qn) so M„ must be irreducible as well. Lemma 3.2. Let Qn = gRgnOL '■ Qn.+/~ -^ Qn^+I- be the combined tree map. Assume that both M^gi) and M[gR) are irreducible. Then, M{qn) is irreducible. Proof. Note that Ml Mr and M„ are irreducible. Let Gq^ be the directed graph of M(g„). We identify vertices of Gq^ with edges of Q„ _|_/_. Let Vl (resp. Vr, Vn) be the set of vertices of Gq^ coming from the set of edges of the subtree Ql (resp. Qr., 0^,+/-) of Qn,+/-- Lemma [3TlT 2) shows that there exists an edge connecting the set Vl to the set Vn, and there exists an edge connecting the set Vn to the set Vl. This is also true between Vn and Vr by Lemma lXlT l.S). Thus, one can find an edge path between any two vertices of Gq^ . D Proposition 3.3. Under the assumptions of Lemma \3.S\ M{qn) is PF. Proof. Lemma [3.1( 2) says that the directed graph Gq^ has an edge from the vertex f^+i to itself, and we denote such edge by e. Since M{qn) is irreducible, for any vertex v of Gq^ there exists an edge path E = eie2 • • • en(v) from ve-^-l to v. Thus, for any n > n{v) we have an edge path e • • • eE of length n from ti^+i to v. Since the number of vertices is finite, there exists an integer A^ > such that for any vertex w of Gq^ and any integer n > N we have an edge path of length n from V£+i 10 to w. Since there exists an edge path from any vertex x of Gq^ to f^+i, we can find a sufficiently large integer N' such that for any pair of vertices x and w there exists an edge path of length A^' from X to w. Thus, M{qn) is PF. D The following property is crucial in proving Proposition 11.11 and Theorem 1 1.3[ Proposition 3.4. Under the assumptions of Lemma \3.S\ we have \{M{qn)) > A(M(g'„+i)) > 1. Proof. To compare M{qn+i) with M{qn) we introduce a new labeling of edges of Q„+i^+/_. The trees Ql and Qr are the common subtrees for both trees Q„ +/_ and Q„+i +/_. Edges of the subtrees Ql and Qn of Q„+i +/_ are numbered in the same manner as those of Q„ 4./_, and edges of Gn+i are numbered ^ + l,£ + n + r + l,£ + 2,£ + 3,--- ,^ + n + l in the clockwise/counterclockwise direction. Here the edge sharing a vertex with the ^*^ edge is numbered i +1. Let M{qn) = irnij)i<ij<e^n+r be the matrix given in ([S?!]). Then, M(g„+i) = {m[j)i<ij<i+n+r+i with new labeling has the following form: \ M(g„^ 7 Put s = A(M(g„+i)) > 1 and let y = *(yi,--- ,yi+n+r+i) be the PF eigenvector for M{qn+i). Then, e+n+r+l y rn'ijVj = syj for i with l<i<£ + n + r + l. (3-2) i=i For i = £ + 1 and i = (, + n + r + \oi (13.21) we have / Ml A B * 1 1 * c 1 D E Mr \ 1 i+i y^ me+ijyj + y^+n+r+i = sy£+i and These two equalities together with s > 1 yield y^ me+ijVj + y^+2 > sy^+i- (3.3) The equalities (13. 2p for all i 7^^ + l,^ + n + r+ 1 together with the inequality ()3.3p imply M(g„)y > sy, where y = *(yi, • • • ,y^+„+r)- By Lemma [12^1), we have A(M(g„)) > s = A(M(g„+i)). By Lemma [121^2) together with ([M we have A(M(g„)) > s. D 11 3.2 Asymptotic behavior of growth rate In this section we concentrate on the combined tree obtained from the triple {GL,Gn,+/-T%)- We assume that QLiQi) = Ql and study the combined tree map g„ = gndL- Let TZ be the subtree of Q„ _|_ /_ such that TZ is obtained from the subtree Ql together with the {i + 1)^* edge. (Hence E{TZ) = {ei, 62, • • • , e^+i}.) For an example of IZ, see Figure [HI By the assumption gLiGb) = Ql, we have QniTZ) D TZ and hence the following tree map r : 7^ ^ 7^ is well defined: for each e G E{TZ), the edge path r(e) is given by the edge path Qnie) by eliminating edges which do not belong to E{TZ). The tree map r does not depend on the choice of n. The transition matrix Mir) is given by the upper-left (£ + 1) x (£ + 1) submatrix of M{qn). We call 7^ the dominant tree and r the dominant tree map for a family of combined tree maps {qn}n>i- l+l l+n+1 l+n+2 Y Gn,+ TR Figure 8: (left) Qn,+, (right) its subtree TZ We now define a polynomial S{t) (resp. U{t)) as follows: Consider the matrix tl — M{qn) (resp. / - tM{qn)) and replace the {i + 1)*^* row by the last row. Take the upper-left (^ -|- 1) x (^ -|- 1) submatrix of the resulting matrix, denoted by S (resp. U), and then S{t) (resp. U{t)) is defined equal to the determinant of S (resp. U). It is not hard to see that the matrices S and U do not depend on n. The following statement, which will be crucial later, tells us that Af(r)(t) is the dominant polynomial and S{t) is the recessive polynomial for a family of polynomials {M(g„)(i)}„>i. Proposition 3.5. We have (1) Miqn){t) = t"M(f){t) + S{t), and (2) M{qn),{t) = t'^Uit) + M(r),(i). Proof. The transition matrix M{qn) = (mij) is of the form M{qn) / Mir) 1 \ mi+ri+1,1 'me+n+1,1+1 1 and it is easy to see that m^+ij = m(,j^n+i,j for 1 < j < L For the proof of (1) (resp. (2)), apply the determinant expansion with respect to the last row of tl — M{qn) (resp. I — tM{qn)). D Proposition 3.6. Suppose that M[gi) is irreducible. Then, we have (1) M{qn) is PF for each n and \{M{qn)) > A(M(g„+i)), and (2) \{M{r)) = lim„^ooA(M(g„)). 12 Proof. (1) The proof is parallel to the proofs of Propositions 13.31 and [37 (2) Apply Lemma 12.51 with Proposition 13.5( 1). D 4 Proof This section is devoted to proving Proposition 11.11 and Theorems II. 2( II. 3( 11.61 Proposition 4.1. The braids P(mi,---,mk+i) ^'^^ P^- By a result of W. Menasco's O Corollary 2], if L is a non-split prime alternating link which is not a torus link, then S^\L has a complete hyperbolic structure of finite volume. Since /3(mi,---,mk+i) ^^ ^ 2 bridge link as depicted in Figure [131 his result tells us that /?{mi,---,mfe+i) is pA. Here we will show Proposition 14.11 by using Proposition 12.41 As a result, we will find the polynomial whose largest root equals the dilatation of /3(mi,---,mfc+i)- Proof of Proposition \4Tl\ To begin with, we define a tree Q(mi,--- ,mk+i) ^^^^ ^ ^^^^ ™^P ^{mi,-- ,mk+i) on the tree Q{mi,- ,mk+i) inductively. For A; = 1 let Q(mi,m2) be the combined tree obtained from the triple {Qmi + ,Gm2-T%)- Take the tree maps gmi and (7^2 with conditions nl,n2,n3 and let us define ^{mi,m2) ^s the combined tree map 9{mi,m2) = 9m29mi ■ Q{mi,m2) ^ 2(mi,m2) (Figure [1). \^ / ^^(4,2) ^ .X ?/ 6 > \ / m ^ • O , ^ 7 5 7 1 2 ^ -7 ' 8 1 Figure 9: case (mi, 7712) = (4,2). Next, suppose that these are defined up to k. Let Q(mi,-- .m^+i) be the combined tree ob- tained from the triple (Q(mi,--,mfc)i ^01^+1,+/-)^) in case k + 1 odd/even. We extend q(mi,--- ,m^.) '■ Q{mi,-,mfe) -^ Q{mi,-,mfc) to a tree map Q ■ -^(mi,--- ,mfe+i) ~^ -^{mi,- ,mk+i) satisfying L1,L2,L3 so that for the edge e of G^^ j^/_ sharing a vertex with the edge of Q(mi,--- .m^)) the length of the edge path q{e) is 3. Let us define the combined tree map (mi,-,mk+ 1) — 9mk+iQ '■ Q{mi,-,mk+i) ~^ Q{ mi,---,mfc+i)- 13 By Proposition 13. 6l f 1 ) . the transition matrix Af (g^^^ ... „j^^^)) is PF. We now deform Q = Q(m.i,---,mfc+i) iiito a train track ^mi,--- .rrn:+i) (as in Figure [TO]) : 1. Puncture a disk near each valence 1 vertex of Q, and connect a 1-gon at the vertex which contains the puncture. 2. Deform a neighborhood of a valence rrii + l vertex of the subtree Q^.^^/ of Q into an (rrii + 1)- gon for each i. 3. For each i + 1 odd/even, puncture above/below the vertex of valence 2 which connects the two subtrees Qmi,+/- and ^^ _/+. Deform a neighborhood of the vertex and connect a 1-gon which contains the puncture. 2 3 Figure 10: numbers 1,2 and 3 correspond to deformations 1,2 and 3. Then, q = q(mi,--- ,mfe+i) induces the graph map g on r = T(^mi,--- ,m^.+i) into itself. Since §■ rotates a part of the train track smoothly (Figure [TT]) . it turns out that qis a smooth map. It is easy to show the existence of a representative homeomorphism f oi (p = r(/3(„^ ... ^m^ , i)) such that r is invariant under / and such that q is the train track map induced by /. The transition matrix of q with respect to real edges of r is exactly equal to the PF matrix M(g(^^ ... ,„^ )). By Proposition 12. 4^ the braid /3(mi,--,mfc + 1) is pA. This completes the proof of Proposition 14.11 D Figure 11: §■ rotates a part of the train track smoothly. Example 4.2. Let us express the formula to compute the dilatation of the braids Pu^m) for m > 1. We know that the dilatation of the braid /3(4 m) is the largest root of the polynomial M{qu.^\){t) for the combined tree map qum) ■ By using qu^2) shown in Figure 0, the transition matrix for qu2) '^s M{q(4,2)] / ° 1 ° \ 1 1 1 1 1 1 1 1 1 , 1 / V 1 2 14 The transition matrix M(r) of the dominant tree map r (Figure \12\) for a family of combined tree maps {qu^m)}m>i is the upper-left 6x6 matrix. Hence the dominant polynomial M{r){t) for {Q(4,m)}m>i equals t^ — t^ — 2t with the largest root ~ 1.45109. In this case the recessive polynomial S{t) is Sit) t -1 t -1 t -1 t -1 -1 -1 t -1 -1 -2 -2t^ t + 1. By Proposition \3.5]( 1). the dilatation of (iurn) ^•s the largest root of M{q(^4,m)){t) = t'^M{r){t) + S{t) = t'^it'^ - t^ - 2t) + (-2*^ -t+1) and Proposition \3. 6]( 2) says that liuim-^oo MP{ i,m) ) 1.45109. 3 4/ Figure 12: dominant tree map r : TZ ^^ TZ ior {(?(4,m)}m>i- We are now ready to show Proposition II. 1[ Proof of Proposition [7771 Recall the tree Q(mi,--- ,mk+i) ^^^ the tree map q(mi,--- ,nn,+i) used in the proof of Proposition 14.11 For i even, Qi^mi.--- ,ni^,.i) is also the combined tree obtained from the triple (Q(mi,-,m,_i), ^m, ,-, Q{m,+u- ,mk+i))- Then, q^mu- ,771^+1) is also the combined tree map given by where q(mu- ,mi.i) and are suitable extensions of and m>+i,-,mfc+i) are sunaoie extensions 01 q{m-,,- ,m,.i) ana g(m,+i,-,mfc+i) respectively. By Proposition 13.41 the claim holds. For i odd, Q(m^,... ^ra^^^) is the combined tree obtained from (2(^1,- ,m,_i), ^m>,+ , 2(m,+i,- ,mfe+i))' where Q',_ _ n is the tree obtained from Q(m ^ , ... m, .,) by the horizontal reflection. Then, the proof is similar to that for the even case. D We turn to the proof of Theorems 11.21 and 11.31 Proof of Theorem M.Si By Proposition 12.41 and by the proof of Proposition 14. 11 the dilatation of /5(mi,-,mfc+i) is the largest root of M(g(^^^...^^^^^))(t). Fixing mi, •• • ,mfc > 1, let 7^(^^_...^^^) and ^(mx,--,m,k) be the dominant tree and the dominant tree map for {Q'(mi,---,mfc+i)}mfc+i>i; and we set By Proposition 13.51 M{qi^m,,-,m,^,)){t) M(9(mi,..,m,+0)*(*) i""=+'-R(mi,-,mfe)(*) + 5'(mi,-,mfc)(*) and (4.1) (4.2) 15 where 5'(„^,... ,„,)(*) is the recessive polynomial for {M(g(^^^...^^^^^))(t)}^^^j>i and f/(mi,...,m,)(t) is the dominant polynomial for {M{q(^rm,- ,mk+l))^i't)}mk+l>l■ ClaLiln 4.3. We have (1) S(mi,-,mk){t) = {-lf^'^R(mi,-,mk)^{t) and (2) t^(mi,-..,mfc)(0 = (-l)'''^^^(mi,...,mfc)(0- Proof. It is enough to show Claim SJSjl). For if (1) holds, by (j4.ip we have This together with (gj]), ([12]) impHes Claim [13^2). We prove Claim [413^ 1) by an induction on k. For k = 1, this holds [U Theorem 3.20(1)]. We assume Claim HTSl fl) up to A; — 1. Then, we have 5'(mi,...,mfc_i)(0 = (~l)'^-^(mi,...,mfc_i) (*)) M{q(rm,-,m,))it) = (-l)^M(g(„^,...,„^))^(t) and t/(„,,.., „,„,)(*) = i-l)''R^rm,- ,m,_,)it)- (4.3) For k >2, the transition matrix M(g(„^ ... ,„^ )) has the block form: 1 ni + l ?12 + 1 Wfc-l + 1 rife + 1 ^(9(mi,...,mfc+i)) = "-fc + l A^('J(mi,...,mj,)) 1 1 1 1 "■fc+1 rifc+i \1 2 2 ••• 2 2 where n^ = rrii + • • • + rrij + j for 1 < j < k + 1. Note that the last edge of the tree 'R-i-mi,— ,m) numbered rij. In this case the polynomial 5, (mi,...,mfc) (t) is the determinant of a matrix: IS / -1 -1-2 _2 ... -2 \ -1 -2 -2 ... -2 t '■■ -1 -1 t -1 -2 / Subtract the second last row from the last row of this matrix, and let A = {aij) be the resulting matrix. Applying the determinant expansion with respect to the last row of A, we have "fc+i S, (mi,...,mfc )(<) E(-i: "fc+i+i '^"fc+ijl^"fc+ij i=i 16 where Aij is the matrix obtained by A with row i and column j removed. Since am^+ij = for j / nk,nk + 1 and a„^+i,„^ = -t, a„j^+i,„j^+i = -1, we have 1^1 = t\An^+l,nJ — \An,.+l,n,,+l\- We note that |^nj.+i^nfc+i| = -^(9(mi,---,mfc))(^)- For the computation of |^rifc+i,nfc|, subtract the sec- ond last row of A„j,^-^^„^,^-^_i from the last row, and for the resulting matrix, apply the determinant expansion of the last column successively. Then, we obtain \Ank+l,nk\ = ~t -f^(mi,---,mfe_i)(*) + '-'(mi,--- ,mfc_i) (^j- Thus, S{nii,---,m^:)(t) = -M{q(^jni,---,7n^:))i't) " ^"''-^(mi,- ,mfc_i) (*) + ^^'(mi,- ,mfc_i) (*)• In the same manner we have By using (j4.3p . these two equalities imply Claim HT3l fl). This completes the proof. We now turn to proving Theorem II. 2[ We will prove an inductive formula for i?(f„^ ... ,„^)(t). It is not hard to show that i?(^^)(t) = t"'i+^(t - 1) - 2t. For k >2, one can verify ^(mi,..,m,.)(*) =*^('?(™i,-,rnfe))W -i""=^(mi,-,m,_i)W +*Vi,-,m,_i)W- M) Substitute the two equalities 5'(mi,-,mfc„i)(*) = (-1) -^{mi,-,mfe_i)^(0 into (j4.4p . then we find the inductive formula -R(mi,-,mfe)(*) = t^H*- l)-R{mi,-,mfc_i)(0 + ("l) 2t-R(mi,- ,'mfe_i)^(*)- This completes the proof of Theorem 11.21 D Proof of Theorem \1.3[ To begin with, we show Claim 4.4. We have (1) lim lim ••• lim A(/3(m ...^ )) = 1 /or z = 1 and mi— >oo r?ii+i^oo mfe+i^oo v i' i^tj-/ (2) lim lim • • • lim A(/3(m ...^ )) = A(i?(„ ...^,_^)(t)) > 1 fori > 2. m^— ►oo mi+i^oo mfe.(.i— >oo ^ '■tj-/ ^ > > , Proof. (1) By Theorem 1 1 . 2 1 and by Lemma [27 ''''k-\-l ''-"-' 17 Recall that -R(mi,---,m,)(0 = ^(^(»ni,--.m,))(0- It is not hard to see that the matrix Af (r(^^ ... ,„^)) for i > 1 is PF, and hence the largest root of R(mi,— ,mi){'t) is greater than 1. Then, by using the inductive formula of -R(mi,--,mi)(^) ™ Theorem 11.21 together with Lemma 12.51 we have lim A(i?(m^^...^^.)(i)) = >^iR(mi,--,mi.i)it)) for i > 2. Since R(m,){t) = t^^+Ht - 1) - 2t, we have lim X(R(^^)(t)) = 1 by Lemma [221 This completes the proof of (1). (2) The proof is identical to that of (1). This completes the proof of Claim [ Claim [44^ 1) says that for any A > 1 there exists an integer mi{\) for each i with \ <i <k + 1 such that A(/3mi(A),---,mfc+i(A)) < A. Set m = max{rn,j(A) | i = 1,---A; + 1}. By Proposition 11.11 A(/5(mi,--,mfc 1 1)) < A whenever rrii > m. This completes the proof of Theorem ll.3l fl). The proof of Theorem 11.3( 2) is identical to that of (1), but using Claim 14.4^ 2) instead of Claim iai^l). D We show the existence of two kinds of families of pA mapping classes with arbitrarily small dilatation and with arbitrarily large volume. Proof of Proposition \1.4\ There exists a family of pseudo-Anosov mapping classes ipn of 7W(Z)„) such that lim Xiipn) = 1- n— »oo It suffices to show that for any pA mapping class (p G M{T,g^p), there exists a family of pA mapping classes (/>„ € Al($]^p(„)) such that the dilatation of 0„ is same as (p and the volume of 0„ goes to oo as n goes to oo. Let $ G (/> be a pA homeomorphism. Since the set of periodic orbits of ^ is dense on T,g^p, one can find a periodic orbit of $, say Q = {qi,-" ,Qs}- Now puncture each point of Q, then the pA mapping class </>' G ^A{T,g^p-^-s) induced by cj) satisfies \{(j)') = \{(p)- On the other hand, vol((/)') > vol{(j)) since T(0) is a complete hyperbolic manifold obtained topologically by filling a cusp of T{<j)') with a solid torus [9^ Section 6]. The volume of any cusp is bounded below uniformly. Thus, if we puncture periodic orbits of $ G successively, we obtain a family of pA mapping class with the desired property. D Finally, we show Theorem 1 1.6i Proof of Theorem \1.6[ By Theorem ll.3( l). for each integer k > 1, the dilatation of P{mi,---,mj,.i) goes to 1 as mi, • ■ ■ ,mfc+i all go to oo. Thus, it suffices to show that the volume of /^{mi,---,™^ . goes to oo as /c goes to oo. One verifies that f3(rni,--- ,m^.,i) is a 2 bridge link as in Figure [T3l In particular it is an alternating link with twist number k + 1. Theorem 1 in [5j tells us that for each mi, • • • , m^^i > 1, V0l(/?(rni,...,mfe+i)) > ^{k - 1)^3, where v^ is the volume of a regular ideal tetrahedron. This completes the proof. D 18 2m2 2m ^^ J +1 ( ;xxx >c p<xx >o XXX XL "• 3<XX XL J 2m J +1 2mj^ 2m2 r~ r -xxx>c — ) XXX X_ *" 3<XX >C^ 2mj+l ^^k+l """^ Figure 13: /3(mi,---,m.fe , i)- (top) k odd, (bottom) even. References [1] M. Bestvina and M. Handel, Train-tracks for surface homeomorphisms, Topology 34 (1994) 109-140. [2] J. Birnian, Braids, Links, and Mapping class groups, Annals of Mathematics Studies 82, Princeton University Press, Princeton, New Jersey (1975). [3] E. Hironaka, Salem-Boyd sequences and Hopf plumbing, Osaka Journal of Mathematics 43 (2006). [4] E. Hironaka and E. Kin, A family of pseudo-Anosov braids with sm^all dilatation, Algebraic and geometric topology 6 (2006) 699-738. [5] M. Lackenby with an appendix by I. Agol and D. Thurston, The volume of hyperbolic alternat- ing link complements. Proceeding of the London Mathematical Society 88, 3 (2004) 204-224. [6] W. Menasco, Closed incompressible surface in alternating knot and link complements. Topology 23 (1982) 37-44. [7] J. -P. Otal and L. Kay, The hyperbolization theorem for fibered 3-manifolds. American Math- ematical Society (2001). [8] E. Seneta, Non-negative Matrices and Markov Chains, Springer- Verlag, New York (1981). [9] W. Thurston, The geometry and topology of 3-manifolds, Princeton University (1979). [10] W. Thurston, On the geometry and dynamics of dijfeomorphisms of surfaces, Bulletin of the American mathematical society 19, 2 (1988) 417-431. Eiko Kin and Mitsuhiko Takasawa Department of Mathematical and 19 Computing Sciences Tokyo Institute of Technology Tokyo, Japan kin@is.titech.ac.jp takasawa@is.titech.ac.jp 20