Are there any nontrivial spaces $Y$ so that for all weak homotopy equivalences $A\to B$, the induced map $[B, Y]\to [A,Y]$ is bijective?

This would be a property of the homotopy type of $Y$, and if $Y$ is homotopy equivalent to a space with has some kind of local structure under which very close maps (probably of compact spaces like $S^k$) are necessarily homotopic, then it probably won't have this property.

My idea is to use the following construction: let $L^+ = \{ {1\over n} \mid n \geq 1\}$ and let $L = \{ 0\} \cup L^+$. Then this hypothetical local property of $Y$ would ensure that the restriction $L \times X \to L^+\times X$ would induce an injection on homotopy sets. But $L\times X$ is weakly equivalent to $\coprod_{0}^\infty X$, and in the latter space we can have maps which are $f$ on $X\times {n}$ for $n > 0$ and $g$ on $X\times 0$, where $f\not \simeq g$.

  • 2
    $\begingroup$ If the space $Y$ is also a $T_1$ space, then its path components are weakly contractible, by choosing $A=S^n$, $n>0$, and $B$ a suitable space with finitely many points. Each of these path components is then contractible, by choosing $A$ to be the path-component and $B$ to be a point. For any two path components there is a sequence in one converging to a point in the other, by choosing $B=${0,1, 1/2, 1/3,...} and $A$ discrete. Thus $Y$ is either connected or empty. Can one take this further and see that then $Y$ is either or contractible or empty? $\endgroup$ – Tom Goodwillie Nov 23 '10 at 14:22
  • 1
    $\begingroup$ Great! So in particular, singular cohomology is not represented by any $T_1$-space in the homotopy category of spaces. $\endgroup$ – Lennart Meier Feb 5 '14 at 20:20
  • $\begingroup$ This arxiv.org/abs/1709.08734 came up on the arXiv today. I hope Jonathan Ariel Barmak will post soon and collect his checkmark. $\endgroup$ – Jeff Strom Sep 27 '17 at 13:15

The answer seems to be "no": only for contractible spaces Y (and Y=$\emptyset$) the functor [-,Y] inverts weak equivalences. As mentioned above I wrote an argument in https://arxiv.org/abs/1709.08734. It uses Jeff Strom and Tom Goodwillie's idea of considering a space whose path-components are the singletons. In this case its topology is similar to the cofinite topology.


Here is an interesting test case: let $B$ be the Stone-Cech compactification of a set $S$, let $A$ be the underlying set of $B$ with the discrete topology, and let $f$ be the identity map. Then $B$ is totally disconnected so every map from a simplex to $B$ is constant, and it follows that $f$ is a weak equivalence, so we must have $[B,Y]=[A,Y]=\text{Map}(A,\pi_0(Y))$. Note that $B$ is always compact and that if $S$ is large enough we can choose a surjective map $A\to Y$; it follows that there is a compact subset of $Y$ that meets every path component. I think it should be possible to extract a lot more from this line of argument, but I do not see it at the moment.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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