We know that covering spaces of many low dimensional CW complexes such as graph (CW structures with only 0-cells and 1-cells) and all compact surfaces has CW structure. [The second fact is due to Uniformation Theorem from theory of Riemann surfaces. Basically, the only simply connected 2-manifold are the sphere, the Euclidean plane and hyperbolic plane.]

So I wonder, as put forth in the title, if this is true in general. That is, if $X$ is connected, locally path-connected and it has a CW complex structure then does its universal covering space also have a CW complex structure?

  • $\begingroup$ Yes. Open cells are (embedded and) contractible, so they lift to unions of open cells. (One copy for each deck transformation, because the Universal covering Is normal.) $\endgroup$ – user39082 Feb 14 '15 at 20:24

This fact is statement (N) in Whitehead's paper "Combinatorial homotopy I". The harder part is to prove that the covering complex has the CW-topology. That is also carefully proved in the book by W.S. Massey, "Algebraic topology: an Introduction".

The result on the CW-topology may also be proved as a corollary of general results proved in P.I. Booth and R. Brown, ``Spaces of partial maps, fibred mapping spaces and the compact-open topology'', Gen. Top. Appl. 8 (1978) 181-195. This gives reasonable circumstances under which if $p: E \to B$ is a map, then the pullback functor $p^*$ from spaces over $B$ to spaces over $E$ has a right adjoint, and so preserves colimits. This immediately gives the result, and even implies more general types of result, for certain generalised inductively defined structures.

  • $\begingroup$ "$(N)$ Any covering complex, $\tilde{K}$, of $K$ is a CW-complex." This presupposes that the covering is a complex. So the question becomes: is every covering of a CW complex a complex (and hence a CW-complex by $(N)$)? $\endgroup$ – Michael Albanese Mar 16 '16 at 3:36
  • $\begingroup$ It is quite easy to give the cell structure on $\tilde{K}$; the more difficult part is to prove that the topology on the cover is the correct one. By the CW-structure, $K^{n+1}$ is obtained from $K^n$ by attaching $(n+1)$-cells. One wants to prove the same is true for $\tilde{K}$. For this it is helpful to know that the pullback functor $p^*$ where $p:\tilde{K}^{n+1} \to K^{n+1}$ preserves pushouts. $\endgroup$ – Ronnie Brown Mar 19 '16 at 23:03

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.