3
$\begingroup$

In the paper Homology fibrations and group completion theorem, McDuff-Segal (www.maths.ed.ac.uk/~aar/papers/mcdsegal.pdf), page 281:

Let $M$ be a topological monoid such that $\pi_0M$ is generated by $s_1,s_2, \cdots,s_k$. Let $s=s_1s_2\cdots s_k$. Then $$ H_*(M)[\pi_0M^{-1}]=H_*(M)[s^{-1}]. $$ However, $\pi_0M$ may not be able to be generated by $s$. Let $m_1\in M$ be a point lying in $s$ ($s$ corresponds to a path-connected component of $M$). Let $M_\infty$ be the mapping telescope formed by the following sequence of maps $$ M\to ^{\cdot m_1}M\to ^{\cdot m_1}M\to ^{\cdot m_1}M\to ^{\cdot m_1}\cdots. $$ It is claimed in Homology fibrations and group completion theorem, McDuff-Segal (www.maths.ed.ac.uk/~aar/papers/mcdsegal.pdf), page 281, line 22- line 23 that "the preceding argument applies" for the proof of group completion theorem for $M$.

In "the preceding argument" of Homology fibrations and group completion theorem, McDuff-Segal (www.maths.ed.ac.uk/~aar/papers/mcdsegal.pdf), $\pi_0M$ is assumed to be $(\mathbb{Z}_{\geq 0},+)$, i.e., $$ \pi_0(M)= [M_1]^{\mathbb{Z}_{\geq 0}}=\{1, s=[M_1], [M_1]^2, [M_1]^3,\cdots\} $$ where $1, [M_1],[M_1]^2,\cdots,[M_1]^j,\cdots$ are the path-connected components of $M$. The following claim is essential for the proof in "the preceding argument".

Claim.[line 14-line 15, page 281]. For any $m\in M$, the left action of $m$ on $M_\infty$ given by
$$m(x\mapsto xm_1\mapsto xm_1^2\mapsto\cdots)= (mx\mapsto mxm_1\mapsto mxm_1^2\mapsto \cdots)$$ is a homology equivalence.

Question: For the case that $\pi_0M$ is generated by $s_1,s_2, \cdots,s_k$, since the proof of the above claim does not hold any longer, how can "the preceding argument applies"?

Apologize: I apologize sincerely that I originally posted the question on math.stackexchange and then deleted the post and shifted it here. Sorry for disturbing!

$\endgroup$

1 Answer 1

5
$\begingroup$

The submonoid (isomorphic to $\mathbb{Z}_{\geq 0}$) generated by the product $s = s_1 \cdots s_k$, while not equal to $\pi_0(M)$, is cofinal in it, and so has the same localization. More directly: the mapping telescope for $s$ and the iterated mapping telescope for each of the $s_i$ have isomorphic homologies.

This is perhaps easiest to see if $k=2$. For an element $x \in \pi_0 M$, write $M[x^{-1}]$ for the telescope for multiplication by $x$. There is a comparison map $\alpha: M[s_1^{-1}] \to M[s^{-1}]$ induced by the comparison of direct systems defining each. I don't know how to do diagrams in mathoverflow, so I can't draw what would be very helpful here. But the comparison of the direct systems comes from the commutativity of $s_1$ and $s_2$.

From $\alpha$, we may build a second comparison of direct limits to get a map $\beta: M[s_1^{-1}][s_2^{-1}] \to M[s^{-1}]$. The domain is $M[s_1^{-1}][s_2^{-1}] = M[s_1^{-1}, s_2^{-1}] = M[\pi_0^{-1}]$. There's also a natural map in the opposite direction, since $s \in \pi_0$. I claim that these are inverse homology equivalences, as can be seen by noting that in

$$H_*(M[s^{-1}]) = H_*(M)[s^{-1}]$$

an inverse to the element $s_1$ is given by $s_2 s^{-1}$, and an inverse for $s_2$ is $s_1 s^{-1}$. Thus $H_*(M)[s^{-1}] = H_*(M)[s_1^{-1}, s_2^{-1}]$.

For $k>2$, iterate.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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