Let $\mathcal{C}$ be a small category and $\Sigma$ a collection of morphisms in $\mathcal{C}$. Denote by $F_\Sigma:\mathcal{C} \to \mathcal{C}[\Sigma^{-1}]$ the usual quotient functor from $\mathcal{C}$ to its [localization][1] about $\Sigma$.

> Are there conditions on $\Sigma$ which guarantee that $F_\Sigma$ induces a homotopy equivalence $B\mathcal{C} \sim B\mathcal{C}[\Sigma^{-1}]$ of classifying spaces?

For example: if $\mathcal{C}$ consists of two objects with a single arrow from one to the other, then localization about that single arrow preserves homotopy type of classifying spaces: everything is contractible before and after localization. On the other hand, see [this paper][2] for a counter-example to the conjecture that the group completion of a monoid has the same classifying space as that monoid. Clearly, we can't just shamelessly start inverting arrows all over the place without destroying homotopy type.

One always has Quillen's Theorem A: if the under categories $F_\Sigma \downarrow c$ are all contractible, then $BF_\Sigma$ is a homotopy-equivalence of classifying spaces. So, one possible answer would highlight those conditions on $\Sigma$ which magically give contractible over/under categories. Is there a known result that does the trick?


  [1]: http://ncatlab.org/nlab/show/localization
  [2]: http://arxiv.org/abs/math/0202260