Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

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

Let $f:X\rightarrow Y$ be a projective morphism of algebraic varieties and $\mathcal{F}$ be a coherent sheaf on $X$. Then some people say that there is a canonical evaluation map $$ f^{*}(f_{*}\mathcal{F})\rightarrow\mathcal{F}, $$ which I don't quite understand the definition. I do understand both pushforward $f_{*}$ and pullback $f^{*}$ of sheaves, but how does one define "evaluation" map above?

share|cite|improve this question
up vote 3 down vote accepted

There is an adjunction between $f^*$ and $f_*$ that says that morphisms $f^* f_* \mathscr{F} \to \mathscr{F}$ are in bijective correspondence with morphisms $f_* \mathscr{F} \to f_* \mathscr{F}$. The "evaluation" morphism is the counit of this adjunction and corresponds to $\textrm{id} : f_* \mathscr{F} \to f_* \mathscr{F}$.

Less abstractly, recall that $f^* f_* \mathscr{F}$ is defined to be the sheaf $a \mathscr{G}$ associated with the presheaf $\mathscr{G}$ defined by $$U \mapsto \mathscr{O}_X (U) \otimes_{f^{-1} \mathscr{O}_Y (U)} \varinjlim_{V \supseteq f U} f_* \mathscr{F} (V) = \varinjlim_{V \supseteq f U} \mathscr{O}_X (U) \otimes_{\mathscr{O}_Y (V)} \mathscr{F} (f^{-1} V)$$ There is an evident presheaf morphism $\mathscr{G} \to \mathscr{F}$ induced by the restriction maps: after all, if $V \supseteq f U$, then $f^{-1} V \supseteq f^{-1} f U \supseteq U$, and each $\mathscr{F}(U)$ is a $\mathscr{O}_X(U)$-module, so there is a map $\mathscr{G} (U) \to \mathscr{F} (U)$ induced by the two universal properties. The universal property of associated sheaves then assures us that $\mathscr{G} \to \mathscr{F}$ factors through the universal presheaf-to-sheaf morphism $\mathscr{G} \to a \mathscr{G} = f^* f_* \mathscr{F}$, and the "evaluation" morphism is precisely the sheaf morphism $f^* f_* \mathscr{F} \to \mathscr{F}$ so obtained.

share|cite|improve this answer
Dear @Zhen, The presheaf you have described is the presheaf version of $f^{-1}f_*\mathscr{F}$, not $f^*f_*\mathscr{F}$, which is $f^{-1}f_*\mathscr{F}\otimes_{f^{-1}\mathscr{O}_Y}\mathscr{O}_X$. You've produce a morphism $f^{-1}f_*\mathscr{F}\rightarrow\mathscr{F}$ from an $f^{-1}\mathscr{O}_Y$-module to an $\mathscr{O}_X$-module, so the morphism $f^*f_*\mathscr{F}\rightarrow\mathscr{F}$ comes by the universal property of base change along the comorphism $f^{-1}\mathscr{O}_Y\rightarrow\mathscr{O}_X$. – Keenan Kidwell Sep 16 '12 at 4:33
Blah, I keep forgetting about the extra complications of $\mathscr{O}_X$-modules... – Zhen Lin Sep 16 '12 at 5:09
Thank you for the detailed answer, Zhen. Then name "evaluation mop" comes form the adjunction bijection evaluated at $id$. I have seen this bijection before. The actual map is quite complicated and I never really understand the map from the bottom of my heart. – M. K. Sep 16 '12 at 7:07

Let $U \subset X$ be an open subset. A section $s \in (f^{\ast}f_{\ast}\mathcal{F})(U)$ is a function $s: U \rightarrow \bigsqcup_{p \in U} (f_{\ast}\mathcal{F})_{f(p)} \otimes_{\mathcal{O}_{Y, f(p)}} \mathcal{O}_{X, p}$ with certain properties (see Hartshorne's book, Proposition-Definition II.1.2). We have a natural projection homomorphism $\varphi: (f_{\ast}\mathcal{F})_{f(p)} = \varinjlim_{V \ni f(p)} \mathcal{F}(f^{-1}(V)) \rightarrow \varinjlim_{V' \ni p} \mathcal{F}(V') = \mathcal{F}_{p}$. If $s(p) = \sum_{i} t_{i} \otimes s_{i} \in (f_{\ast}\mathcal{F})_{f(p)} \otimes_{\mathcal{O}_{Y, f(p)}} \mathcal{O}_{X, p}$, then the evaluation morphism of sheaves takes $s$ to $\tilde{s}: U \rightarrow \bigsqcup_{p \in U} \mathcal{F}_{p}$, given by $\tilde{s}(p) = \sum_{i} s_{i}\varphi(t_{i})$.

In my opinion, this is correct, but I would like to read comments. One curious and natural question is: what are the kernel and image of such morphism?

share|cite|improve this answer

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.