# Start a new discussion

## Not signed in

Want to take part in these discussions? Sign in if you have an account, or apply for one below

## Discussion Tag Cloud

Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.

• CommentRowNumber1.
• CommentAuthorUrs
• CommentTimeMay 11th 2011
• (edited Oct 24th 2012)

added in CW-complex in the Examples section something about noncompact smooth manifolds.

Eventually it would be good to state here precisely Milnor’s theorem etc. Googling around I seem to see a lot of misleading imprecision in the usual statements along these lines (on Wikipedia and MO) concerning the distinctions between countably generated and general CW-complexes and concerning homotopy equivalence vs weak homotopy equivalence.

• CommentRowNumber2.
• CommentAuthorzskoda
• CommentTimeMay 12th 2011
• (edited May 12th 2011)

Besides there are two notions of CW-complex. One is a topological space with fixed CW-complex structure on it, and another is a topological space for which exist a CW-structure. The categories are very different, the first category, sometimes called CELL has morphisms respecting cell structure (cellular maps) and the latter does not. Cellular approximation theorem says that the two become equivalent at the level of homotopy category. I am in a hurry now, so can not edit at the moment.

A “relative” CW-complex (X,A) is similar, except X 0 is the disjoint union of A with a discrete space.

I think I learned that $X_0$ is only $A$ in that case. Is this standard ? (Or maybe $A$ needs to be $X_{-1}$ ?)

• CommentRowNumber3.
• CommentAuthorjim_stasheff
• CommentTimeMay 12th 2011
Besides there are two notions of CW-complex. One is a topological space with fixed CW-complex structure on it, and another is a topological space for which exist a CW-structure.

In the old days, CW-complex meant a topological space with fixed CW-complex structure on it.
Otherwise we would refer to a space of the homotopy type of a CW complex.
• CommentRowNumber4.
• CommentAuthorTim_Porter
• CommentTimeMay 12th 2011

I think this is still the better terminology. (Am I old fashioned?)

• CommentRowNumber5.
• CommentAuthorzskoda
• CommentTimeMay 12th 2011

I agree with Tim and Jim, however I would say more precisely a topological type not the homotopy type of CW-complex. Whitehead and Postinkov I think use such terminology in their courses. I will put some remarks into nLab entry. Maybe not now.

• CommentRowNumber6.
• CommentAuthorMike Shulman
• CommentTimeMay 13th 2011

Of course there is also a difference between a space of the homotopy type of a CW complex and a space of the homeomorphism type of a CW complex. I would tend to think that this difference is even bigger than the difference between a space of the homeomorphism type of a CW complex and a space with a specified CW complex structure. (We also have another name for “space of the homotopy type of the CW complex” now – m-cofibrant space.)

• CommentRowNumber7.
• CommentAuthorUrs
• CommentTimeMay 13th 2011

And one should maybe better say “space of the strict homotopy type of a CW-complex”. To distinguish from the weak homotopy type for which the statement becomes empty.

• CommentRowNumber8.
• CommentAuthorzskoda
• CommentTimeMay 14th 2011
• (edited May 14th 2011)

No, homeomorphism type, i.e. topological type, not homotopy type. With homotopy type we get OUT of the class of spaces homeomorphic to the underlzing topological space of a CW complex. In my opinion the clean way is to talk about two differenet entities:

• CW-complexes

• and their UNDERLYING topological spaces.

Period. Self-explanatory, said in a bit more modern and unambiguous way, but expressing the classical distinction.

Then one can later talk about more rough questions of homotopy theory, but this is not the original matter or the class of topological spaces defined.

Urs: I think people say strong homotopy type, not “strict” homotopy type.

• CommentRowNumber9.
• CommentAuthorUrs
• CommentTimeOct 24th 2012

added a section Properties - Singular homology with some basic statement about the singular homology of CW-complexes.

• CommentRowNumber10.
• CommentAuthorTodd_Trimble
• CommentTimeOct 24th 2012
• (edited Oct 24th 2012)

There seems to be some duplication of material here; see also cellular homology.

Why are there two notations, $H_k(X_n | X_{n-1})$ and $H_k(X_n, X_{n-1})$? I can’t recall ever seeing the bar notation.

Fixed a couple of typos.

Edit: I just looked higher up on the page, where the bar notation is introduced. But I honestly don’t think there is any danger at all in thinking $H_k(X_n, X_{n-1})$ might stand for homology of $X_n$ with coefficients in $X_{n-1}$ (it doesn’t even make sense)! Do other people use the bar notation? Anyway, I don’t think we should have both notations on the same page; it’s confusing.

• CommentRowNumber11.
• CommentAuthorUrs
• CommentTimeOct 24th 2012

Okay, I’ll change the notation. And, yes, I am currently working on cellular homology.

• CommentRowNumber12.
• CommentAuthorUrs
• CommentTimeOct 25th 2012
• (edited Oct 25th 2012)

More little properties at CW complex.

Also, I have exanded and slightly rearranged the Idea-section there.

• CommentRowNumber13.
• CommentAuthorTodd_Trimble
• CommentTimeSep 20th 2015

Added material to CW complex in a section titled “Up to homotopy equivalence”.

• CommentRowNumber14.
• CommentAuthorUrs
• CommentTimeApr 25th 2016

Added a section Terminology on the meaning of the letters “CW”.

• CommentRowNumber15.
• CommentAuthorUrs
• CommentTimeJun 22nd 2016
• (edited Jun 22nd 2016)

Do the following quotient spaces admit cell complex structure?

For $q \leq n \in \mathbb{N}$, let $O(n-q)$ act on the sphere $S^{n-q}$ via the canonical linear action on $\mathbb{R}^{n-q}$ passed to the one-point compactification $S^{n-q} \simeq (\mathbb{R}^{n-q})^\ast$. Then does

$( O(n)_+) \wedge_{O(n-q)} S^{n-q}$

admit cell complex structure? I suppose it does, but what’s a solid argument?

In fact it would be sufficient for me if it were the retract of a cell complex.

• CommentRowNumber16.
• CommentAuthorUrs
• CommentTimeJun 22nd 2016
• (edited Jun 22nd 2016)

Here an attempt to prove that $O(n)_+ \wedge_{O(n-q)} S^{n-q}$ does have cell complex structure.

We may think of this space equivalently as the result of first forming the manifold with boundary

$O(n) \times D^{n-q}$

then forming the group quotient

$( O(n) \times D^{n-q} ) / O(n-q)$

and then collapsing the boundary to the point.

By Illmann 83, corollary 7.2, the smooth manifold $O(n) \times D^{n-q}$ does admit $O(n-q)$-CW structure, hence the quotient $( O(n) \times D^{n-q} ) / O(n-q)$ does inherit CW-structure.

But moreover, by the sentence just above theorem 7.1 in Illmann 83, this $O(n-q)$-CW structure may be chosen such that the boundary is a $O(n-q)$-CW subcomplex.

This means that as we now collapse the boundary, the result is still a CW-complex.

Right?

• CommentRowNumber17.
• CommentAuthorDavidRoberts
• CommentTimeJun 22nd 2016
• (edited Jun 22nd 2016)

I think your argument does hold water, assuming that $O(n)_+ \wedge_{O(n-q)} S^{n-q}$ is indeed $\left((O(n) \times D^{n-q} ) / O(n-q)\right)/boundary$ – I guess that $O(n)_+$ is just $O(n)$ with a disjoint basepoint added?

• CommentRowNumber18.
• CommentAuthorUrs
• CommentTimeJun 23rd 2016
• (edited Jun 23rd 2016)

Thanks for the sanity check.

I guess that $O(n)_+$ is just $O(n)$ with a disjoint basepoint added?

Yes.

So for $X$ an unpointed space and $Y$ a pointed space then

\begin{aligned} (X_+) \wedge Y & \simeq \frac{(X_+) \times Y}{ (X_+) \times \{y_0\} \sqcup \{*\} \times Y } \\ & \simeq \frac{ X \times Y \sqcup \{\ast\} \times Y }{ X \times \{y_0\} \sqcup \{\ast\} \times \{y_0\} \sqcup \{\ast\} \times Y } \\ & \simeq \frac{X \times Y}{ X \times \{y_0\}} \end{aligned} \,.

Now $S^{n-q} \simeq D^{n-q}/S^{n-q-1}$ with basepoint the image of $S^{n-q-1}$. So

\begin{aligned} (O(n)_+) \wedge S^{n-1} & \simeq \frac{ O(n) \times D^{n-q} }{ O(n) \times S^{n - q - 1} } \\ & \simeq \frac{ O(n) \times D^{n-q} }{ \partial( O(n) \times D^{n-q}) } \end{aligned} \,.

Finally, the two quotients clearly commute:

$\frac{ O(n) \times D^{n-q} } { \partial( O(n) \times D^{n-q} )} / O(n-q) \simeq \frac{ (O(n) \times D^{n-q}) / O(n-q) } { (\partial( O(n) \times D^{n-q} ))/O(n-q)}$
• CommentRowNumber19.
• CommentAuthorDavidRoberts
• CommentTimeJun 23rd 2016

And boundary commutes with quotient, since the $O(n-q)$-action preserves both the boundary and interior of $O(n) \times D^{n-q}$ :-)

• CommentRowNumber20.
• CommentAuthorUrs
• CommentTimeJun 23rd 2016

Sure. Okay, great. Thanks.

• CommentRowNumber21.
• CommentAuthorDavidRoberts
• CommentTimeJun 23rd 2016

Where is this material going again? I’ve tried to keep track, and I guess it’s to do with the model structure for orthogonal spectra?

• CommentRowNumber22.
• CommentAuthorUrs
• CommentTimeJun 23rd 2016

Yes, I am typing this into model structure on orthogonal spectra which then becomes part of Introduction to Stable homotopy theory – 1-2.

I’ll drop you a note a little later today when the above argument has found its place.

• CommentRowNumber23.
• CommentAuthorUrs
• CommentTimeJun 23rd 2016

So the argument above is now this lemma at “model structure on orthogonal spectra”.

It is used in the proof of this lemma to show that the generating acylic cofibrations of the stable model structure are indeed acyclic, and it is used in th proof of this theorem stating that the stable model structure is monoidal.

• CommentRowNumber24.
• CommentAuthorDavidRoberts
• CommentTimeJun 23rd 2016

Ah, cool, thanks! Nice to see where it is used, too.

• CommentRowNumber25.
• CommentAuthorUrs
• CommentTimeApr 21st 2017

I was contacted by a young person saying that the definition at CW-complex was hard to read. I went and expanded the Definition section, breaking it up into smaller steps, including examples, and reordering a little.

• CommentRowNumber26.
• CommentAuthorDmitri Pavlov
• CommentTimeApr 10th 2019
• CommentRowNumber27.
• CommentAuthorUrs
• CommentTimeApr 11th 2019
• (edited Apr 11th 2019)

Thanks for the alert. I have fixed it in the entry, replacing “the” by “among the”. But the entries being pointed to for details (such as CW-approximation) are clear about this.