MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I am trying to study (finite) spherical buildings from a very combinatorial point of view : Every rank 3 spherical building is a finite simplicial complex of dimension 3, so one can define its density as the ratio #triangles/#vertices, expressed as a function of the number $n$ of vertices. Maybe it is the wrong name for a very classical notion but I did not find it anywhere. One of the first distinctions in buildings is the concept of thick buildings. In a way, this density measures the thickness of a given building, but I never saw this notion.

I am wondering what is the density of the spherical buildings, and which one are the densest ?

Starting from the classification of spherical buildings, it is enough to compute the density of the buildings of type $A_3$ and $C_3$. I know how to do it for $A_3$, since the building in this case is the flag complex of a projective space on $\mathbb{F}_q^4$, and it seems to give $\frac{[4]_q!}{\binom{4}{2}_q}=O(q^2)=O(\sqrt{n})$ where $n$ is the number of vertices, with the usual q-analog notations.

But $C_3$ looks scary to me, as there are multiple possible geometries, and I don't know of a systematic way to extract the (simple) combinatorial information out of the groups. Is this something well known, or is there a simple way to do this ?

share|cite|improve this question
You write $\frac{[4]_q!}{\binom{4}{2}_q}\in O(n^{3/2})$, where I guess the $n$ should be a $q$. But when I compute this quotient, I get $q^2+2q+1 \in O(q^2)$... Huh? – Max Horn Jun 25 '13 at 11:27
You are right of course, I got messed up between several edits. It should be correct now. – Arnaud Jun 25 '13 at 13:01
up vote 5 down vote accepted

While I have never seen this notion of "density" of a finite building before, let's see what we can do...

A thick finite $C_3$ building corresponds to a thick finite rank 3 polar space. These are well-known, they are $W(5,q)$, $Q(6,q)$, $Q^-(7,q)$, $H(5,q^2)$ and $H(6,q^2)$, corresponding to the groups $Sp(6,q)$, $O(7,q)$, $O^-(8,q)$, $U(6,q)$ and $U(7,q)$.

To compute the densities, I took a look at a table indicating the number of points contained in these spaces as well as the number of points collinear to a single fixed point (in my case, I looked at the book Diagram Geometry by Buekenhout and Cohen, p. 485; there are other, older sources, but I already had that book sitting next to me... Google just showed me another source here).

From this and the knowledge how many points are contained in a line and a plane (namely $q+1$ and $q^2+q+1$), it is a matter of simple combinatorics to count the total number of points, lines and planes, as well as chambers (= maximal flags = triples of a point, line and plane containing each other). Each point, line and plane corresponds to a vertex in the simplicial complex, while the chambers correspond to triangles. Thus the "density" as you defined it can be computed via the formula $$ \frac{\#chambers}{\#points + \#lines + \#planes}.$$

Plugging in the numbers of points and perp sizes for all of the $C_3$ buildings listed above, the "density" turns out to be in $O(q^3)$ in all cases (modulo mistakes on my side resp. in the small GAP program I wrote to compute all of this). So it is a bit larger than for projective spaces, where I confirmed it to be in $O(q^2)$.

share|cite|improve this answer
Thank you very much, I do not know much about polar spaces and missed a good reference. Would you happen to have an asymptotic bound on the density in terms of the number of vertices, i.e. n=#points+#lines+#planes ($q$ is not a very informative measure) ? I will play with Mathematica on my side.. – Arnaud Jun 25 '13 at 13:15
The number of vertices varies between $O(q^7)$ and $O(q^{30})$ (it can also be $O(q^{10})$ and $O(q^{27})$). So the density in terms of $n$ varies between $O(n^{3/7})$ and $O(n^{3/30})=O(n^{1/10})$. – Max Horn Jun 25 '13 at 13:42
Sweet. So it looks like according to this measure, the projective space is the densest one. Many thanks for this great answer. – Arnaud Jun 25 '13 at 13:47

Your Answer


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.