Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I'm having trouble following the proof of Proposition I.5.2 in Goerss-Jardine (Simplicial Homotopy Theory). After establishing the adjunction $\hat\Delta(X\times K,Y) \simeq \hat\Delta(K,[X,Y])$, they claim that diagrams of the form

enter image description here

correspond via the exponential adjunction to diagrams of the form

enter image description here

where $i\colon K\hookrightarrow L$ is an inclusion, and $p\colon X\to Y$ is a fibration, and the pullback is induced byenter image description here

I imagine this is at the level of trivial manipulations with the adjunction, but I can't get it to come out right (mainly not sure how to go from the pullback to the pushout).

share|cite|improve this question
A pullback is a limit; right adjoints preserve limits (the right adjoint of a limit is the limit of the right adjoints). If your functor is an appropriate contravariant adjoint, then it will transform limits into colimits. – Arturo Magidin Dec 18 '11 at 5:02
@ArturoMagidin: But the pushout [K,X] -> [K,Y] <- [L,Y] is not the result of an application of [-,Y] or [K,-] to some diagram of simplicial sets... – Yuri Sulyma Dec 18 '11 at 5:47
up vote 5 down vote accepted

Apart from performing a somewhat lengthy diagram chase, the only thing to do here is to notice that giving a map $(a,b) : A \to [K,X] \mathrel{\times_{[K,Y]}} [L,Y]$ amounts to giving a commutative diagram as on the left by the universal property of the pull-back $\DeclareMathOperator{\Hom}{Hom}$

pull-back and its adjoint

while the diagram on the right is equivalent to it by the adjunction between the cartesian product and the internal $\Hom$ and the definitions of $p_\ast$ and $i^\ast$.

Let me give names to the maps in the first commutative diagram of the proof you ask about:

First commutative square of Goerss-Jardine

The map $f: \Lambda_{k}^n \to [L,X]$ corresponds to a map $\tilde{f}: \Lambda_{k}^n \times L \to X$ while giving the bottom map $(g,h) : \Delta^n \to [K,X] \mathrel{\times_{[K,Y]}} [L,Y]$ amounts to either of the two commutative diagrams

Squares corresponding to maps from Delta to the pull-back

Using the universal property of the pull-back defining $[K,X] \mathrel{\times_{[K,Y]}} [L,Y]$ we see that asserting the commutativity of the square we started with is equivalent to giving the map $f: \Lambda_{k}^n \to [L,X]$, the commutative square $(1)$ and requiring the two squares

The two squares corresponding to commutativity of the first square of GJ

to be commutative.

Passing to the adjoint side using the map $\tilde{f} : \Lambda_{k}^n \times L \to X$ and the square $\widetilde{(1)}$, the commutativity of the squares $(2)$ and $(3)$ is equivalent to the two commutative squares

The two squares from before, but now on the adjoint side

At this point I think I can leave it to you to contemplate the commutative diagrams $\widetilde{(1)}$, $\widetilde{(2)}$ and $\widetilde{(3)}$ and the push-out square defining $(\Lambda_{k}^n \times L) \mathrel{\cup_{(\Lambda_{k}^n \times K)}} (\Delta^n \times K)$ and to think about what it means to define a map from $(\Lambda_{k}^n \times L) \mathrel{\cup_{(\Lambda_{k}^n \times K)}} (\Delta^n \times K)$ in order to see that giving the squares $\widetilde{(1)}$, $\widetilde{(2)}$ and $\widetilde{(3)}$ and requiring their commutativity is equivalent to giving the commutative square

Second commutative square of Goerss-Jardine

as claimed by Goerss and Jardine.

share|cite|improve this answer

Halfway through typing my answer I saw that t.b. already gave it. Since I seem to be unable to comment on anything, I post this comment as an answer. The original problem is certainly not trivial because there is more than one pair of adjoint functors: you have $(-)\times K \dashv [K,-]$ and $(-)\times L\dashv [L,-]$.

The missing concept is that of conjugate natural transformations as explained in Mac Lane's CWM (Section IV-7): suppose $F,F':\mathcal{X}\rightarrow\mathcal{A}$ and $G,G':\mathcal{A}\rightarrow\mathcal{X}$ are functors with $F\dashv G$ and $F'\dashv G'$. Two natural maps $\alpha: F\rightarrow F'$ and $\beta: G'\rightarrow G$ (note the opposite directions) are said to be conjugate if the diagrams $$ \matrix{ & \mathcal{A}(F'x,a) & \cong & \mathcal{X}(x,G'a) & \cr \mathcal{A}(\alpha_x,a) &\Big\downarrow& &\Big\downarrow& \mathcal{X}(x,\beta_a)\cr & \mathcal{A}(Fx,a) & \cong & \mathcal{X}(x,Ga) & } $$ commute for all objects $x\in\mathcal{X}$, $a\in\mathcal{A}$, where the horizontal maps are the isomorphisms given by adjointness. In fact, any given $\alpha: F\rightarrow F'$ determines a unique conjugate $\beta=\alpha^*: G'\rightarrow G$ (take $x=G'a$ and start with $id_{G'a}$ in the upper right corner).

In your situation the inclusion map $i:K\hookrightarrow L$ gives a natural map $(-)\times i:(-)\times K\rightarrow (-)\times L$ and the corresponding conjugate is $i^*:[L,-]\rightarrow [K,-]$ given by precomposing with $i$. This is conveniently already named $i^*$ in your diagram, although with objects dropped from the notation.

I always wondered why this notion of conjugate transformations is never made explicit in books on algebraic topology although it is used all the time.

share|cite|improve this answer
Yes, I should probaly have mentioned that. However, I fail to see how this gives the desired correspondence between the diagrams without doing the verifications I outlined. – t.b. Dec 23 '11 at 10:32
I've seen your comment, thank you for your response to my query. I post it here since it seems to fit a comment better instead of an edit to the post): "note: @t.b: there is nothing missing from your answer. My above remark was meant as a comment to the original question, because the OP wondered if he was missing some basic concept." – t.b. Dec 23 '11 at 20:43

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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