Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

In my research, the following construction came up.

Let $X$ be an $n$-dimensional simplicial complex. For an integer $m \geq 1$, let $X[m]$ denote the following simplicial complex. The vertices of $X[m]$ are pairs $(v,i)$, where $v \in X^{(0)}$ and $i \in \{1,\ldots,m\}$. A set $\{(v_0,i_0),\ldots,(v_k,i_k)\}$ of vertices of $X[m]$ forms a $k$-simplex if the $v_j$ are all distinct and the set $\{v_0,\ldots,v_k\}$ is a $k$-simplex of $X$.

I'm interested in the relationship between the homology groups of $X$ and $X[m]$.

I suspect that the answer is pretty complicated in general, so let me describe a more specific situation/question. The complexes $X$ I'm interested in are spherical buildings. For those that don't know what these are, the precise definition is not too important; rather I think the following property is the key. By the Solomon-Tits theorem, such an $X$ is $(n-1)$-connected, and thus homotopy equivalent to a wedge of $n$-spheres. In fact, even more is true. Namely, such an $X$ is a Cohen-Macaulay complex, that is, for a $k$-dimensional simplex $\sigma$ of $X$, the link of $\sigma$ is an $(n-k-1)$-dimensional complex which is $(n-k-2)$-connected.

If $X$ is as above, then $X[m]$ is still an $n$-dimensional simplicial complex. Moreover, it is an easy exercise to show that $X[m]$ is also still $(n-1)$-connected, so the interesting homology group is $H_n(X[m])$. There is a natural action of the symmetric group $S_m$ on $X[m]$, so the homology group $H_n(X[m];\mathbb{Q})$ is a representation of $S_m$.

This brings me to my more specific question. Assume that $X$ is as above. How can one describe the $S_m$-representation $H_n(X[m];\mathbb{Q})$ in terms of the homology groups of $X$ (and possibly other topological data about $X$)?

share|improve this question
1  
I think this sequence of S_m representations will be stable in the sense of Church and Farb: arxiv.org/abs/1008.1368 Send me an email if you'd like details. –  John Wiltshire-Gordon Jun 18 '12 at 17:48
    
You might want to try to prove that $X$ shellable and balanced implies $X[m]$ is shellable, since this would give you a really good description of a (co)-homology basis and also might be helpful for calculating characters. –  Patricia Hersh Jun 18 '12 at 18:20
1  
This is a really pretty problem. Could you say more about how this came up in your research? –  Vidit Nanda Jun 18 '12 at 18:36
    
@John Wiltshire-Gordon : That would be interesting, and I'll send you an email. –  Martin Jun 18 '12 at 18:40
1  
Russ, this doesn't sound right -- you may go up in a chain in P with order complex X while needing to decrease the decoration. Also, Cartesian product increases dimension. –  Patricia Hersh Jun 19 '12 at 15:57
show 6 more comments

3 Answers

up vote 11 down vote accepted

If I understand correctly, $w \in S_m$ acts by sending $(v,i)$ to $(v,w(i))$. If so, you are in the nice situation where the stabilizer of a face fixes every point in that face. So, for $w \in S_m$, the set of faces of $X[m]$ fixed by $w$ is a subcomplex of $X[m]$. By what is often called (if my memory is correct) the Hopf trace formula, the value at $w$ of the character of your representation is $(-1)^n$ times the reduced Euler characteristic of this fixed point subcomplex, where $n=dim(X)$ as in your question.

Now $w$ fixes a face $\{(v_0,i_0),\ldots,(v_k,i_k)\}$ if and only if $w$ fixes each $i_j$. Therefore, the fixed point subcomplex for $w$ is isomorphic with $X[fix(w)]$, where $fix(w)$ is the number of fixed points of $w$ on $[m]$.

So, if you know the reduced Euler characteristic of $X[k]$ (which is the same as knowing the rank of $\widetilde{H}_n(X[k])$) for each $k \leq m$ then you know the character values for your representation. You can use the formula of Bj\"orner-Wachs_Welker to get this Euler characteristic. I hope this is of use to you.

If what I have said so far is correct, your complexes provide nice examples of the type requested in problem 7.65(b) of Richard Stanley's ``Enumerative Combinatorics, Volume 2". That is, you have a sequence $\theta_1,\theta_2,\ldots$, where $\theta_m$ is a character of $S_m$, and for each $w \in S_m$, $|\theta_m(w)|=\theta_k(1)$ for some $k \leq m$.

share|improve this answer
7  
Congratulations, John! This is post number $10^5$ on MO! mathoverflow.net/questions/100000 –  François G. Dorais Jun 19 '12 at 19:04
add comment

This probably should just be a comment, but I do not have enough reputation to do that. I think your $X[m]$ are particular cases of the "inflated simplicial complexes" of

Björner, Anders; Wachs, Michelle L.; Welker, Volkmar Poset fiber theorems. Trans. Amer. Math. Soc. 357 (2005), no. 5, 1877–1899.

share|improve this answer
    
This is actually pretty helpful. In particular, Corollary 6.3 of this paper gives what appears to be a complete description of the homology groups of $X[m]$ in terms of the homology groups of links in $X$ (with almost no assumptions on $X$!). It doesn't give the characters of the representations, but it still is nicer than I was even hoping for! –  Martin Jun 18 '12 at 19:46
    
It was hard to figure out which of the two answers to accept (both have really helpful information in them). I decided to accept the other one largely because it seems to be getting fewer upvotes, so I wanted to give the poster some additional credit :). Thank you very much for your help. –  Martin Jun 19 '12 at 16:38
add comment

If I understand your question correctly $C_i(X[m])=\oplus_P C_i(X)\otimes U^{i,m}_P\otimes V_P$ where $U^{i,m}_P$ has dimension $d^{i,m}_P$ which is the number of semistandard fillings of the shape $P$ with $0, 1, 2, \ldots, i, \ldots i$ and $V_P$ is an irreducible representation of $S_m$.

For some representations this is easy to compute. For instance if $P_1=m-d-1$ then $d^{d-1,m}_P= 0$, but $d^{d,m}_P > 0$, so $H_{d,P}(X[m])=C_d(X)\otimes U^{d,m}_P\otimes V_P$.

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.