Why are these algebraic manipulations needed to conclude this fixed point theorem?

31 Views Asked by At

In Evasiveness of Graph Properties and Topological Fixed Point Theorems on page 46 it states that (where $F$ is the associated chain map for any function $f$):

$$Tr_F(H_0(\Delta, F_p)) = 1$$

Because $F$ acts "trivially" on $H_0$.

Later (on the same page) it mentions, as, in this case, $f$ is an automorphism that the matrix $F$ has zeros on the diagnol, and hence, the trace is always zero. So:

$$Tr_F(H_n(\Delta, F_p)) = 0$$ for $n \geq 0$.

The paper then proceeds to give this long complicated series of summation/algebraic manipulations to show that $0$ does not in fact equal $1$. Why is it not already a contradiction to note that the trace is always $1$ (because $F$ acts trivially) and simutanteously $0$ because $f$ being an automorphism implies a zero trace for $H_0$?

1

There are 1 best solutions below

1
On BEST ANSWER

You're confusing two different actions of $f$. The reason that $Tr_F(H_n(\Delta, \mathbb{F}_p)) = 0$ for $n>0$ is not that the matrix has zeroes on the diagonal; it's simply that $H_n(\Delta, \mathbb{F}_p)=0$ so you are taking the trace of a $0\times 0$ matrix. This argument only applies for $n>0$, since $H_0(\Delta,\mathbb{F}_p)$ is not trivial.

The matrix which has zeroes on the diagonal is the action of $f$ on $K_n(\Delta,\mathbb{F}_p)$: that is, on the chains, not the homology. This is because we are assuming $f$ fixes no simplices (not because $f$ is an automorphism), and a basis for the chains is given by the simplices. This does not apply to the action of $f$ on homology, and in particular does not tell you that $Tr_F(H_0(\Delta, \mathbb{F}_p))$ should be $0$.