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

What's the easiest (by which I mean uses the least fancy machinery) proof of the direct summand conjecture in dimension 2?

Recall that the direct summand conjecture says that:

Conjecture (Hochster): If $R$ is a regular ring and $S$ is a module finite integral extension, then $R \to S$ splits as a map of $R$-modules.

It is trivial in characteristic zero (via the trace map) and not that hard in characteristic $p > 0$ using Frobenius-type methods. In mixed characteristic it is known up to dimension 3.

share|cite|improve this question
Are you asking for the easiest proof? – Manny Reyes Oct 5 '10 at 17:17
Yes. I am asking for the easiest proof. I fixed the typo above. – Karl Schwede Oct 5 '10 at 20:58
up vote 5 down vote accepted

You may assume that $R,S$ are complete and $S$ is a domain. Now take the integral closure $T$ of $S$, which is $S$-finite. Since we are in dimension $2$, $T$ is maximal Cohen-Macaulay module over $R$, so $T$ is $R$-free. Thus the composition map $R\to T$ splits (as it takes $1$ to $1$) whence the map $R\to S$ splits.

The moral of this is that existence of small Cohen-Macaulay modules implies a lot of things, and you can get that for free in dimension $2$ via integral closure.

share|cite|improve this answer
May be Mel's CBMS note has a reference? Embarrassingly, I do not own a copy (-:. – Hailong Dao Oct 5 '10 at 17:58
I looked in Mel's CBMS book before asking, but I didn't see this there. That's a very nice proof, thanks! Do you know of any other approaches? – Karl Schwede Oct 5 '10 at 20:45
Karl, the other way I know is via the monomial conjecture, the reduction to that case is elementary: you need that the extension is cyclically pure, and it's enough to use ideals gen. by powers of elements in a s.o.p. In dimension 2, monomial conjecture amounts to showing $x^ty^t \notin (x^{t+1},y^{t+1})$ for all $t>0$, $x,y$ s.o.p. Probably that has an elementary proof. – Hailong Dao Oct 6 '10 at 2:39
Long, thanks. That's very helpful. – Karl Schwede Oct 6 '10 at 4:56

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.