I am attempting to prove Theorem 1.3A in III.1 of Hartshorne's algebraic geometry, which says that $\delta$-functors $T=(T^i)_{i\geq0}$ with each $T^i$ effaceable/erasable for $i\geq 1$, are universal.

My setup is as follows: I have let $T$ be an effaceable $\delta$-functor as above, and considered another $\delta$-functor $\bar{T}$ with a natural transformation $f^0:T^0\implies \bar{T}^0$ between them in the zeroth degree. I want to construct the remaining natural transformations $f^i:T^i\implies \bar{T}^i$. Not having many tools to work with, I plan to construct these by components, that is, for each $A$ in the source category I wish to define morphisms $f^i_A:T^i(A) \to \bar{T}^i(A)$.

I believe I have constructed the first such natural transformation (in an inductively extendable fashion). Letting $A \stackrel{\phi}{\hookrightarrow} M$ be a $T^1$-erasure of $A$ (i.e. an embedding such that $T^1(\phi) = 0$), we can write a short exact sequence $$ A \stackrel{\phi}{\hookrightarrow} M \stackrel{\rho}{\twoheadrightarrow} N.$$ From the definition of a $\delta$-functor, and the assumed existence of $f^0$, we then have the following "ladder" with exact rows: enter image description here where the morphism $f^1_A$ is uniquely induced by the universal property of the cokernel $T^1(A)$ of $T^0(M) \to T^0(N)$.

My issue lies in showing that such an induced morphism is in fact well defined for a given $A$. The construction appears to heavily depend on the chosen erasure $\phi$, which would be useless for defining the desired natural transformation $f^1$.

Any help is much appreciated.


I refer to this > https://mathoverflow.net/a/260485/105001 for answering your questions.I just give the details and its idea come from others.

if $u:A\longrightarrow M$ and $v:A\longrightarrow N$ are both monomorphisms and such that $T^1(u)=0=T^1(v)$,we can get $w:A\longrightarrow M\bigoplus N$.(I dont know how to draw commutative diagram with diagonal arrows in stackexchange)

let $i_M:M\longrightarrow M\bigoplus N,p_M:M\bigoplus N\longrightarrow M$ be canonical morphisms and $i_N,p_N$ be canonical morphisms for $N$.

Then $w=i_M \circ u + i_N\circ v$. It is obvious $T^1(w)=0$ for $T^1$ is additive functor and $w$ is monomorphism.

then we can easily prove that $f^1:T^1(A)\longrightarrow \bar{T^1}(A)$ from $u:A\longrightarrow M$ is equal to $f^1:T^1(A)\longrightarrow \bar{T^1}(A)$ from $w:A\longrightarrow M\bigoplus N$.

similarly we can easily prove that $f^1:T^1(A)\longrightarrow \bar{T^1}(A)$ from $v:A\longrightarrow N$ is equal to $f^1:T^1(A)\longrightarrow \bar{T^1}(A)$ from $w:A\longrightarrow M\bigoplus N$.

So $f^1$ is well-defined.

  • $\begingroup$ While this link may answer the question, it is better to include the essential parts of the answer here and provide the link for reference. Link-only answers can become invalid if the linked page changes. - From Review $\endgroup$ – Theoretical Economist Sep 17 '18 at 14:18
  • $\begingroup$ @TheoreticalEconomist I have changed my answer but more detailed information need to draw a 3-dimension commutative diagram which I don’t know how to get in stack exchange. $\endgroup$ – ngegjnjkgn Sep 17 '18 at 15:22

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.