C-ALGEBRAS AND FELL BUNDLES ASSOCIATED TO A TEXTILE SYSTEM VALENTIN DEACONU Abstract. The notion of textile system was introduced by M. Nasu in order to analyze endomorphisms and automorphisms of topological Markov shifts. A textile system is given by two nite directed graphs GandH and two morphisms p;q:G!H, with some extra properties. It turns out that a textile system determines a rst quadrant two-dimensional shift of nite type, via a collection of Wang tiles, and conversely, any such shift is conjugate to a textile shift. In the case the morphisms pandqhave the path lifting property, we prove that they induce groupoid morphisms ;: (G)!(H) between the corresponding  etale groupoids ofGandH. We de ne two families A(m;n) and A(m;n) ofC-algebras associated to a textile shift, and compute them in speci c cases. These are graph algebras, associated to some one-dimensional shifts of nite type constructed from the textile shift. Under extra hypotheses, we also de ne two families of Fell bundles which encode the complexity of these two-dimensional shifts. We consider several classes of examples of textile shifts, including the full shift, the Golden Mean shift and shifts associated to rank two graphs. 1.Introduction In dynamics, the time evolution of a physical system is often modeled by the iterates of a single trans- formation. However, multiple symmetries of some systems lead to the study of the join action of several commuting transformations, where new and deep phenomena occur. The classical shift of nite type from symbolic dynamics was studied with powerful tools from linear algebra and matrix theory. The number of period npoints, the zeta function and the entropy can all be simply expressed in terms of the kktransition matrix A. The Bowen-Franks group BF(A) =Zk=(IA)Zk is invariant under ow equivalence, and it was recovered in the K-theory of the Cuntz-Krieger algebra OA generated by partial isometries s1;s2;:::;sksuch that 1 =kX i=1sis i; s jsj=kX i=1A(j;i)sis ifor 1jk: The algebraOAis simple and purely in nite if and only if Ais transitive (for every i;jthere exists msuch thatAm(i;j)6= 0), andAis not a permutation matrix. These C-algebras can also be understood as graph algebras, which were studied and generalized by several authors, see [R]. The higher dimensional analogue of a shift of nite type consists in all d-dimensional arrays of symbols from a nite alphabet subject to a nite number of local rules. Such arrays can be shifted in each of the d coordinate directions, giving dcommuting transformations. There are also dtransition matrices, which in Date : April 5, 2019. 1991 Mathematics Subject Classi cation. Primary 46L05; Secondary 46L55. Key words and phrases. textile system, shift of nite type, graph C*-algebra, Fell bundle. Research partially supported by a UNR JFR Grant. 1arXiv:1001.0037v1 [math.OA] 30 Dec 20092 VALENTIN DEACONU general do not commute. There are deep distinctions between the case d= 1 andd2: for example, it is easy to describe the space of such arrays in the rst case, but there is no general algorithm which will decide, given the set of local rules, whether or not the space of such arrays is empty in the second case. Although the general theory of multi-dimensional shifts of nite type is still in a rudimentary stage, there are particular classes where signi cant progress was made, and where graphs and matrices play a useful role. These include the class of algebraic subshifts, see [S1, S2] and the class of two-dimesional shifts associated totextile systems , or to Wang tilings . For these classes, some of the conjugacy invariants, like entropy (the growth rate of the number of patterns one can see in a square of side n), the number of periodic points and the zeta functions were computed. In the literature, there are some papers relating higher dimensional shifts of nite type and C-algebras. For example, the particular case of shifts associated to rank dgraphs was studied by A. Kumjian, D. Pask and others. In this case, the translations in the coordinate directions are local homeomorphisms, and there is a canonical  etale groupoid and a C-algebra associated to such a graph, which is Morita equivalent to a crossed product of an AF-algebra by the group Zd. Under some mild conditions, the groupoid is essentially free and theC-algebra is simple and purely in nite. For more details, see [KP]. Also, in [PRW1] and [PRW2], the authors analyze the C-algebra of rank two graphs whose in nite path spaces are Markov subgroups of (Z=nZ)N2, like the Ledrappier example, see also [KS] and [LS]. In all these examples, the entropy is zero. The connections between higher dimensional subshifts of nite type and operator algebras remains to be explored further, and we think that this is a fascinating subject. In this paper, in an attempt to apply results from operator algebra to arbitrary two-dimensional shifts of nite type supported in the rst quadrant, we construct two families of C-algebras, de ned using some one- dimensional shifts associated to a textile shift as in [MP2]. The K-theory groups of these algebras provide invariants of the two-dimensional shift. We also construct groupoid morphisms and families of Fell bundles associated to some particular textile systems. We consider several examples of textile shifts, related to rank two graphs, to the full shift, to the Golden Mean transition matrices and to cellular automata. Acknowledgements . The author wants to express his gratitude to Alex Kumjian, David Pask and Aidan Sims for helpful discussions. 2.Textile systems and two-dimensional shifts of finite type Throughout this paper, we consider nite directed graphs G= (G1;G0), whereG1is the set of edges, G0 is the set of vertices, and s;r:G1!G0are the source and range maps, which are assumed to be onto. De nition 2.1. A textile system (see [N]) is a quadruple T= (G;H;p;q ), whereG= (G1;G0),H= (H1;H0) are two nite directed graphs, and p;q:G!Hare two surjective graph morphisms such that (p(e);q(e);r(e);s(e))2H1H1G0G0uniquely determines e2G1. We have the following commutative diagram:C-ALGEBRAS AND FELL BUNDLES ASSOCIATED TO A TEXTILE SYSTEM 3 H0p G0q!H0 "r"r"r H1p G1q!H1 #s#s#s H0p G0q!H0 The dual textile system T= (G;H;s;r ) is obtained by interchanging the pairs of maps ( p;q) and (s;r). The new graphs G= (G1;H1) and H= (G0;H0) have source and range maps given by pandq, ands;rare now graph morphisms. Note that, even if the initial graphs GandHhave no sinks, the new graphs Gand Hmay have sinks (vertices vsuch thats1(v) =;). A rst quadrant textile weaved by a textile system Tis a two-dimensional array ( e(i;j))2(G1)N2, such thatr(e(i;j1)) =s(e(i;j)) and such that q(e(i1;j)) =p(e(i;j)) for alli;j2N. It is clear that (e(i;j))j2N2G1(the in nite path space of G) for alli2N. In some cases, the set of such arrays may be empty (see Example 3.1 in [A]). Remark 2.2. A textile system associates to each edge e2G1a square called Wang tile with bottom edge s(e), top edge r(e), left edge p(e), and right edge q(e): r(e) p(e)eq(e): s(e) If we letX=X(T) to be the set of all textiles weaved by T, thenXis a closed, shift invariant subset of (G1)N2, and we obtain a two-dimensional shift of nite type, de ned below. Alternatively, if we use Wang tiles, we get a tiling of the rst quadrant. We will describe in Proposition 3.1 the connection between two-dimensional shifts of nite type and textile systems. De nition 2.3. LetSbe a nite alphabet of cardinality jSj. The fulld-dimensional shift with alphabet S is the dynamical system ( SNd;), where m(x)(n) =x(n+m); x2X; n;m2Nd: A subsetXSNdwhich is closed in the product topology and which is -invariant is called a d-dimensional shift of nite type or a Markov shift if there exists a nite set (window) FNdand a set of admissible patternsPSFsuch that X=X[P] =fx2SNdj(mx)jF2Pfor everym2Ndg: Many times F=f(0;0;:::;0);(1;0;:::;0);(0;1;:::;0);:::;(0;0;:::;1)g. A shift of nite type has dtransition matrices of dimension jSjwith entries inf0;1g, which in general do not commute. De nition 2.4. LetS1andS2be alphabets, let FNdbe a nite subset, and let  : SF 1!S2be a map. A sliding block code de ned by  is the map :SNd 1!SNd 2;(x)n= (xjF+n);n2Nd:4 VALENTIN DEACONU Ford= 1 we recover the notion of cellular automaton. Two shifts of nite type X[P1];X[P2] are conjugate if there is a bijective sliding block code :X[P1]!X[P2]. In this case, the dynamical systems ( X[P1];) and (X[P2];) are topologically conjugate (see [LS]). Ford= 2, any Markov shift can be speci ed by two transition matrices. Such shifts are investigated by N.G. Markley and M.E. Paul in [MP1]. Two kktransition matrices AandBwith no identically zero rows or columns are called coherent if (AB)(i;j)>0 i (BA)(i;j)>0 and (ABt)(i;j)>0 i (BtA)(i;j)>0; whereBtis the transpose. If AandBare coherent, it is proved that X(A;B) =fx2SN2:A(x(i;j);x(i+ 1;j)) = 1 and B(x(i;j);x(i;j+ 1)) = 1 for all ( i;j)2N2g becomes a two-dimensional shift of nite type, where S=f0;1;2;:::;k1g. For more about multi-dimensional shifts of nite type, we refer to [S1, S2] and [L, LS]. We illustrate now with some examples of textile systems and their associated two-dimensional shifts. Example 2.5. LetG1=fa;bg;G0=fu;vgwiths(a) =u=r(b);s(b) =v=r(a);and letH1=fxg;H0= fwgwithp(a) =p(b) =x=q(a) =q(b): � �� ��� � �� � Figure 1. The corresponding two-dimensional shift has alphabet S=fa;bgand transition matrices A=" 1 1 1 1# ; B=" 0 1 1 0# : We will see later that this shift is a particular case of a cellular automaton, obtained from the automorphism of the Bernoulli shift ( fa;bgN;) which interchanges aandb. It also corresponds to a rank two graph, because the transition matrices commute and the unique factorization property is satis ed (see [KP] section 6). Example 2.6. LetG1=fa;b;cg;G0=fug;H1=fe;fg;H0=fvgwithp(a) =p(b) =e;p(c) =f;q(a) = f;q(b) =q(c) =e. Then the corresponding two-dimensional shift of nite type has alphabet fa;b;cgand transition matrices A=2 6640 0 1 1 1 0 1 1 03 775; B =2 6641 1 1 1 1 1 1 1 13 775: Note thatAandBare coherent in the sense of Markley and Paul, but do not commute, so this shift is not associated to a rank two graph.C-ALGEBRAS AND FELL BUNDLES ASSOCIATED TO A TEXTILE SYSTEM 5 �� ��� � �� � � � Figure 2. Example 2.7. LetG1=fa;b;cg;G0=fu;vg;s(a) =s(b) =r(c) =r(b) =u;r(a) =s(c) =v;H1= feg;H0=fwg;p(a) =p(b) =p(c) =q(a) =q(b) =q(c) =e. � � �� �� � � �� � Figure 3. This textile system is isomorphic to the dual of the previous one. The corresponding two-dimensional shift of nite type has the same alphabet, but the transition matrices are interchanged. 3.Textile systems associated to a two-dimensional shift of finite type From a two-dimensional shift of nite type Xwe will construct a double sequence of textile systems T(m;n), considering higher block presentations of Xsuch thatXand the shift determined by T(m;n) are conjugated. Recall Proposition 3.1. (see [JM] ) Let (X;)be a two-dimensional shift of nite type with alphabet S. Then, moving to a higher block presentation of Xif necessary, there exists a textile system Tsuch thatXis determined by T. Proof. ConsiderB=B(2;2) the set of 22 admissible blocks =a b c dinX, and construct a graph G withG0labeled by the rows of the blocks in B,G1=B; s( ) =c d,r( ) =a b, and a graph HwithH0=S andH1labeled by the columns of the blocks in B. De ne graph morphisms p;q:G!Hbyp( ) =a c, q( ) =b d. It is clear that T= (G;H;p;q ) is a textile system such that Xis the set of textiles weaved by T. 6 VALENTIN DEACONU Corollary 3.2. Form;n1, letB(m;n)denote the set of mnadmissible blocks in X, and forn2 de ne a graph G(m;n)withG0(m;n) =B(m;n1)andG1(m;n) =B(m;n). For 2G1(m;n), lets( ) = the lowerm(n1)block of and letr( ) = the upperm(n1)block of . Then for m2there are graph morphisms p;q:G(m;n)!G(m1;n)de ned byp( ) =the left (m1)nblock of ,q( ) = the right (m1)nblock of , where 2G1(m;n). ThenT(m;n) := (G(m;n);G(m1;n);p;q)for m;n2are textile systems, and Xis determined by T(m;n). The shift Xis also determined by the dual textile system T(m;n) := ( G(m;n);G(m;n1);s;r), where G1=B(m;n),G0(m;n) =B(m1;n)and the source and range maps are given by pandqas above. We illustrate with some two-dimensional shifts of nite type and their associated textile systems. In each case, the morphisms p;qare de ned as in 3.1. Example 3.3. (The full shift). Let S=f0;1gand letX=SN2. In the corresponding textile system T= T(2;2), the graph G=G(2;2) is the complete graph with 4 vertices. Indeed, G1=( a b c dja;b;c;d2S) andG0=f0 0;0 1;1 0;1 1g:The graph H=G(1;2) is the complete graph with 2 vertices. Indeed, H1=( 0 0;1 0;0 1;1 1) andH0=f0;1g. Example 3.4. (Ledrappier). Let S=Z=2Z, and letXSN2be the subgroup de ned by x2Xi x(i+ 1;j) +x(i;j) +x(i;j+ 1) = 0 for all ( i;j)2N2: We haveG0(2;2) =H1=SS, andG1(2;2) has 8 elements, corresponding to the 2 2 matrices ( a(i;j)) with entries in Ssuch thata(1;1) +a(2;1) +a(2;2) = 0. The Ledrappier shift is associated to a rank two graph, and if we consider the new alphabet 0 0 0;1 0 1;1 1 0;0 1 1; then the transition matrices are A=2 666641 1 0 0 0 0 1 1 1 1 0 0 0 0 1 13 77775; B =2 666641 1 0 0 0 0 1 1 0 0 1 1 1 1 0 03 77775; see [PRW1]. Example 3.5. (Golden Mean). Let S=f0;1g, with transition matrices A=B=" 1 1 1 0# : Then in the corresponding textile system T=T(2;2), the graphs G=G(2;2) andH=G(1;2) have G0=f0 0;0 1;1 0g; G1=( 0 0 0 0;0 1 0 0;0 0 0 1;1 0 0 0;1 0 0 1;0 0 1 0;0 1 1 0) ;C-ALGEBRAS AND FELL BUNDLES ASSOCIATED TO A TEXTILE SYSTEM 7 H0=f0;1g; H1=( 0 0;1 0;0 1) : Example 3.6. (Cellular automata). Let k1 and letYf0;1;:::;k1gNbe a subshift of nite type. It is known that a continuous, shift-commuting onto map ':Y!Yis given by a sliding block code. Given such a', de ne a closed, shift invariant subset X=f(ym)2YNjym+1='(ym) for allm2Ngf 0;1;:::;k1gN2: In a natural way, Xbecomes a two-dimensional Markov shift. In the corresponding textile system T= T(2;2), we have G0f0;1;:::;k1gf 0;1;:::;k1g,G1= the set of admissible 2 2 blocksa b c dwith a;b;c;d2f0;1;:::;k1g,H0=f0;1;:::;k1g, andH1= the set of admissible columnsa c. Fork= 2,Y=f0;1gNand'de ned by interchanging the letters 0 and 1, we recover the textile system from example 2.5. Recall that many rank two graphs can be obtained from two nite graphs G1andG2with the same set of vertices such that the associated vertex matrices commute, and a xed bijection :G1 1G1 2!G1 2G1 1 such that if ( ; ) = ( 0; 0), thenr( ) =r( 0) ands( ) =s( 0). Here G1 1G1 2:=f( ; )2G1 1G1 2js( ) =r( )g; ands;rare the source and range maps. This rank two graph is denoted by G1G2. The in nite path space is a rst quadrant grid with horizontal edges from G1and vertical edges from G2. Each 11 square is uniquely determined by one horizontal edge followed by one vertical edge. Proposition 3.7. Any rank two graph of the form G1G2determines a textile system. Proof. Indeed, let Hi=Gop i, the graph Giwith the source and range maps interchanged for i= 1;2. The mapinduces a unique bijection H1 1H1 2!H1 2H1 1, where H1 1H1 2=f( ; )2H1 1H1 2jr( ) =s( )g: We letGwithG1=H1 1H1 2identi ed with H1 2H1 1by the map ,G0=H1 1, and we let H=H2. De ne s( ; ) = ; r ( ; ) = 0; p( ; ) = 0, andq( ; ) = , where 0; 0are uniquely determined by the bijection( ; ) = ( 0; 0).  Remark 3.8. For a cellular automaton with 'as in 3.6 de ned by an automorphism of a rank one graph G, in [FPS] the authors associated a rank two graph whose C-algebra is a crossed product C(G)oZ, and they computed its K-theory. 4.C-algebras associated to a two-dimensional shift of finite type Recall that in Corollary 3.2 we constructed a family T(m;n) = (G(m;n);G(m1;n);p;q) of textile systems from a two-dimensional shift of nite type. This de nes a family of graph C-algebrasA(m;n) := C(G(m;n)) form;n2. The dual textile system T(m;n) = ( G(m;n);G(m;n1);s;r) determines another8 VALENTIN DEACONU family A(m;n) :=C(G(m;n)), where G(m;n) is the graph with source and range maps given by pandq, described in Corollary 3.2. Remark 4.1. We haveA(m;n)=A(m;2) for alln2 and A(m;n)=A(2;n) for allm2. Indeed, the graphG(m;n) is a higher block presentation of G(m;2) and the graph G(m;n) is a higher block presentation ofG(2;n) (see [B]). For matrix subshifts, we can be more speci c. Consider A;B two coherent kktransition matrices indexed byf0;1;:::;k1gas in [MP2], and let X(A;B) be the associated matrix shift. Theorem 4.2. For a matrix shift X(A;B)we have A(2;n)=OAnandA(n;2)=OBnforn2. The transition matrices AnandBncan be constructed inductively as in [MP2] , and they de ne two sequences (Y(An))n1and(Y(Bn))n1of one-dimensional shifts of nite type associated to X(A;B). Proof. Consider the strip Kn=f(i;j)2N2: 0jn1g and the alphabet Qn=B(1;n) =f : is a 1nblock occuring in X(A;B)g; ordered lexicographically starting at the top. De ne Y(An) =fxjKn:x2X(A;B)gto be the Markov shift with alphabet Qnand transition matrix An, obtained by restricting elements of X(A;B) to the strip Kn. The shiftY(Bn) is de ned similarly, considering strips Ln=f(i;j)2N2: 0in1g and alphabets Rn=B(n;1) =f : is an1 block occuring in X(A;B)g: Clearly,A1=AandB1=B. Forn2,Anis aknknmatrix, where knis the sum of all entries in Bn1. Suppose and 0are 1nblocks inX(A;B) andj;j02f0;1;2;:::;k1g. Then An+1 j ;j0 0! = 1 if and only if the 1 (n+1) blocksj andj0 0occur inX(A;B) andA(j;j0)An( ; 0) = 1. By Proposition 2.1 in [MP2], the matrix An+1is the principal submatrix of A Anobtained by deleting the mth row and column ofA Anif and only if B(i;j) = 0, where m=jkn+h;0h