MathOverflow is a question and answer site for professional mathematicians. 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

There's a well-known correspondence (traditionally called Hartshorne-Serre) between codimension 2 smooth subvarieties $S\subset X$ of a smooth algebraic variety $X$ and certain rank two vector bundles on $X$. Is there anything similar for higher codimensional subvarieties of $X$, i.e. codimension 3 and more?

share|cite|improve this question
The title made me think of Hartshorne and Serre sending letters to one another on 6-codimensional paper! – MTS Jul 16 '13 at 1:25
To my knowledge, the problem of writing letters in codimension higher than one is still open... – IMeasy Jul 16 '13 at 8:08
@MTS: conversely, one day I saw a fat book in the library entitled "Grothendieck--Serre correspondence". I didn't look inside, but instead spent quite a while trying to guess what the objects involved in this correspondence could be... – user5117 Jul 16 '13 at 14:05
@Artie: The objects involved were missives :) – Jason Starr Jul 16 '13 at 14:46
up vote 5 down vote accepted

To get some idea of the difficulty in generalizing Serre's correspondence to higher codimension, bear in mind that what makes a resolution $0 \rightarrow \mathcal{F} \rightarrow \mathcal{E} \rightarrow \mathcal{I}_{Z|X} \rightarrow 0$ plausible when $\mathcal{F},\mathcal{E}$ are vector bundles on $X$ and $Z \subset X$ is smooth of codimension 2 is that $Z$ is locally Cohen-Macaulay. If $Z$ is assumed to be locally Cohen-Macaulay of possibly higher codimension, then $\mathcal{F}$ is at best reflexive, and even if we pass to a resolution of $\mathcal{I}_{Z|X}$ by vector bundles, recovering $Z$ from a series of vector bundle maps is a highly nontrivial matter.

EDIT: As indicated below by Libli's informative answer, what is more relevant is that $Z$ is at least locally Gorenstein (which in codimension 2 is equivalent to being a local complete intersection).

share|cite|improve this answer

In codimension 3, you can hope for your varieties to be cut out (scheme theoretically) by the Pfaffians of an alternating map of vector bundles on $X$.

If you assume that $S$ is subcanonical, which you have to assume for Serre's correspondence (by the way, traditionnally, this is not called Hartshorne-Serre, but only Serre's correspondence) that $X = \mathbb{P}^N$ and some small extra-assumptions on $S$, then it is a theorem of Walter that $S$ is cut out by such Pfaffians (see : Walter, Pfaffians subschemes, Journal of Algebraic Geometry, 1996).

If $X$ is not the projective space, you can still hope for some structure theorems, but they are more difficult to explain (see :

In codimension $4$ there is this : (which seems to me even much more complicated) and after (codimension $5$ or bigger) I guess nothing is known.

share|cite|improve this answer

As far as I have ever heard (and I have asked around), there is no general analogue of Serre's construction in higher codimension.

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.