Nested Complexes and Their Polyhedral Realizations Andrei Zelevinsky

Nested Complexes and Their Polyhedral Realizations Andrei Zelevinsky

Pure and Applied Mathematics Quarterly Volume 2, Number 3 (Special Issue: In honor of Robert MacPherson, Part 1 of 3 ) 655|671, 2006 Nested Complexes and their Polyhedral Realizations Andrei Zelevinsky To Bob MacPherson on the occasion of his 60th birthday 1. Introduction This note which can be viewed as a complement to [9], presents a self-contained overview of basic properties of nested complexes and their two dual polyhedral realizations: as complete simplicial fans, and as simple polytopes. Most of the results are not new; our aim is to bring into focus a striking similarity between nested complexes and associated fans and polytopes on one side, and cluster complexes and generalized associahedra introduced and studied in [7, 2] on the other side. First a very brief history (more details and references can be found in [10, 9]). Nested complexes appeared in the work by De Concini - Procesi [3] in the context of subspace arrangements. More general complexes associated with arbitrary ¯nite graphs were introduced and studied in [1] and more recently in [10]; these papers also present a construction of associated simple convex polytopes (dubbed graph associahedra in [1], and De Concini - Procesi associahedra in [10]). An even more general setup developed by Feichtner - Yuzvinsky in [6] associates a nested complex and a simplicial fan to an arbitrary ¯nite atomic lattice. In the present note we follow [5] and [9] in adopting an intermediate level of generality, and studying the nested complexes associated to building sets (see De¯nition 2.1 below). Feichtner and Sturmfels in [5] show that the simplicial fan associated to a building set is the normal fan of a simple convex polytope, while Postnikov in [9] gives an elegant construction of this polytope as a Minkowski sum of simplices. Received July 29, 2005. 2000 Mathematics Subject Classi¯cation.Primary 52B. Research supported by NSF grant DMS-0500534. 656 Andrei Zelevinsky In this note we develop an alternative approach, constructing the polytope di- rectly from the fan; this construction is completely analogous to that in [2]. The main role in this approach is played by the link decomposition property (Propo- sition 3.2 below) asserting that the link of any element in a nested complex can be naturally identi¯ed with a nested complex of smaller rank. This result pro- vides a uni¯ed method for establishing properties of nested complexes and their polyhedral realizations by induction on the rank of a building set. The note is structured as follows. Section 2 introduces (basically following [9]) building sets and associated nested complexes. Section 3 presents the link decomposition property (Proposition 3.2). In this generality it seems to be new; for the graph associahedra it appeared in [10]. Section 4 contains some struc- tural properties of nested complexes, all of them derived in a uni¯ed way from Proposition 3.2. In Section 5 we show that any nested complex can be realized as a smooth complete simplicial fan (Theorem 5.1 and Corollary 5.2); we refer to this fan as the nested fan. This construction has appeared in [6, 4, 5]. However, it is not so easy to extract from these papers a proof that the nested fan is well-de¯ned; this is why we prefer to give a simple self-contained proof. In Section 6, we prove (Theorem 6.1) that the nested fan is the normal fan of a simple polytope ¦, which we call the nested polytope. The argument is completely parallel to that in [2]; in particular, it leads to a speci¯c realization of ¦ which can be shown to be the same as the realization given in [9]. The concluding Section 7 discusses the special class of graphical building sets, those giving rise to graph associahedra [1, 10]. We give a simple characterization of graphical building sets (Proposition 7.3). For them an analogy with cluster complexes becomes sharper: in particular, the corresponding nested complex is a clique complex (Corollary 7.4). 2. Nested complexes We start by recalling basic de¯nitions from [9, Section 7]. De¯nition 2.1 ([9], De¯nition 7.1). Let S be a nonempty ¯nite set. A building set (or simply a building) on S is a collection B of nonempty subsets of S satisfying the conditions: (B1) If I;J 2 B and I \ J 6= ;, then I [ J 2 B. (B2) B contains all singletons fig, for i 2 S. The nested complex associated to a building B is de¯ned as follows. Let Bmax ½ B be the set of maximal (by inclusion) elements of B. We refer to the elements of Bmax as the B-components. In view of (B1) and (B2), the B-components Nested Complexes 657 are pairwise disjoint, and their union is S. We de¯ne the rank of B by setting rk(B) = jSj ¡ jBmaxj. De¯nition 2.2 (cf. [9], De¯nition 7.3). A subset N ½ B ¡ Bmax is called nested if it satis¯es the following condition: for any k ¸ 2 and any I1;:::;Ik 2 N such that none of the Ii contains another one, the union I1 [ ¢ ¢ ¢ [ Ik is not in B. The simplicial complex on B¡Bmax formed by the nested sets is called the nested complex and denoted N (B). In particular, the 1-simplices of N (B) are pairs fI;Jg ½ B such that: (2.1) either I ½ J, or J ½ I, or I [ J2 = B (in view of (B2), in the last case I and J are disjoint). Note that our de¯nition of nested sets is di®erent from the one in [9, De¯nition 7.3]: Postnikov's nested sets are obtained from ours by adjoining the set Bmax. Of course this does not a®ect the poset structure of the nested complex; an advantage of the present version is that our nested sets form a simplicial complex. 3. Link decomposition Let B be a building on S, and N (B) the associated nested complex. For every C 2 B ¡ Bmax, let 0 0 (3.1) N (B)C = fN ½ B ¡ Bmax ¡ fCg : N [ fCg 2 N (B)g denote the link of C in N (B); this is a simplicial complex on the set fI 2 B¡Bmax ¡ fCg : fC; Ig 2 N (B)g. Our main tool in studying the structure of N (B) is the recursive description of N (B)C given in Proposition 3.2 below. To state it, we need some preparation. De¯nition 3.1. Let C be a nonempty subset of S. ² The restriction of B to C is the building on C de¯ned by (3.2) BjC = fI ½ C : I 2 Bg: ² The contraction of C from B is the building on S ¡ C de¯ned by (3.3) CnB = fI ½ S ¡ C : I 2 B or C [ I 2 Bg: The fact that both BjC and CnB are indeed buildings, is immediate from De¯nition 2.1. Let S1;:::;Sk be disjoint ¯nite sets, and let Bi be a building on Si for i = 1; : : : ; k. The product B1 £ ¢ ¢ ¢ £ Bk is de¯ned as the building on S1 t ¢ ¢ ¢ t Sk formed by the elements of all the Bi. Clearly, the nested complex N (B1 £¢ ¢ ¢£Bk) is naturally isomorphic to the product N (B1) £ ¢ ¢ ¢ £ N (Bk). In particular, every building B on S decomposes as a product BjS1 £¢ ¢ ¢£BjSk , where S1;:::;Sk are the B-components; therefore, N (B) can be identi¯ed with N (BjS1 ) £ ¢ ¢ ¢ £ N (BjSk ). 658 Andrei Zelevinsky Proposition 3.2 (Link Decomposition). For every C 2 B¡Bmax, the link N (B)C is isomorphic to N (BjC £ CnB). 0 Proof. Fix C 2 B ¡ Bmax and abbreviate B = BjC £ CnB. Let B0 = fI 2 B ¡ Bmax ¡ fCg : fC; Ig 2 N (B)g be the ground set of N (B)C . In view of (2.1), B0 is the disjoint union of the following three sets: B1 = fI 2 B ¡ fCg : I ½ Cg;(3.4) B2 = fI 2 B ¡ Bmax ¡ fCg : C ½ Ig;(3.5) (3.6) B3 = fI 2 B ¡ Bmax : C \ I = ;;C [ I2 = Bg: 0 Now to every I 2 B0 = B1 [B2 [B3 we associate a subset I ½ S by setting ( I if I 2 B1 [B3; (3.7) I0 = I ¡ C if I 2 B2. Remembering the de¯nitions, we see that B1 is the ground set of the nested complex N (BjC ), and B3 is a subset of the ground set of N (CnB) formed by those I0 ½ S ¡C for which C [I0 2= B. Furthermore, the correspondence I 7! I0 = I ¡C identi¯es B2 with the remaining part of the ground set of N (CnB), formed by those I0 ½ S ¡ C for which C [ I0 2 B. Thus, the correspondence I 7! I0 is a 0 bijection between the ground sets of N (B)C and N (B ). To prove Proposition 3.2, it remains to show the following: The correspondence I 7! I0 induces an(3.8) 0 isomorphism of complexes N (B)C and N (B ). Thus, we need to show the following: for any distinct I1;:::;Ik 2 B0, the 0 0 0 set fC; I1;:::;Ikg is nested for B if and only if fI1;:::;Ikg is nested for B . 0 0 0 First suppose fI1;:::;Ikg is not nested for B . Without loss of generality, we 0 0 can assume that k ¸ 2, none of the Ii contains another one, and I1 [ ¢ ¢ ¢ [ 0 0 Ik 2 B . Choosing the smallest possible k with this property and using property 0 (B1) in De¯nition 2.1, we can further assume that the Ii are pairwise disjoint.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    17 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us