Krull intersection theorem for modules: Difference between revisions

From Commalg
No edit summary
Line 1: Line 1:
{{Noetherian ring result}}
{{applicationof|Artin-Rees lemma}}
{{applicationof|Artin-Rees lemma}}
{{applicationof|Nakayama's lemma}}
{{applicationof|Nakayama's lemma}}
Line 18: Line 19:
* [[Nakayama's lemma]]
* [[Nakayama's lemma]]


==Proof==
===The intersection equals its product with <math>I</math>===
We first show that the intersection equals its product with <math>I</math>. This is the step where we se the Artin-Rees lemma.
Let:
<math>N := \bigcap_1^\infty I^jM</math>
Now consider the filtration:
<math>M \supset IM \supset I^2M \supset \ldots</math>
this is an <math>I</math>-adic filtration and the underlying ring is [[Noetherian ring|Noetherian]], hence by the [[Artin-Rees lemma]], the following filtration is also <math>I</math>-adic:
<math>N \supset IM \cap N \supset I^2M \cap N\supset \ldots</math>
Since each <math>I^jM</math> contains <math>N</math>, the filtration below is the same as the filtration:
<math>N \supset N \supset N \supset \ldots</math>
This being <math>I</math>-adic forces that <math>IN = N</math>.
===Finding the element <math>r</math>===
Since <math>IN = N</math>, we can find an element <math>r \in I</math> such that <math>(1 - r)N = 0</math>. This is an application of the [[Cayley-Hamilton theorem]]: we first find the Cayley-Hamilton polynomial, then observe that <math>1</math> is a root of the polynomial, and then take the negative of the sum of all coefficients of higher degree terms.
===Proving the intersection is trivial for an integral domain===
If <math>R</math> is an integral domain, and we set <math>M = R</math> in the above, we find an element <math>r \in I</math> such that <math>(1 - r)(\bigcap_j I^j) = 0</math>. Thus <math>1 - r</math> is a zero divisor on the intersection. Since <math>R</math> is an [[integral domain]], one of these must hold:
* <math>1 - r = 0</math>. This forces <math>1 \in I</math>, contradicting the assumption that <math>I</math> is a [[proper ideal]]
* <math>\bigcap_j I^j = 0</math>
This completes the proof.
===Proving that the intersection is trivial for a local ring===
In this case, <math>I</math> is contained inside the unique maximal ideal <math>\mathfrak{m}</math>, which is the Jacobson radical. As above, let <math>N = \bigcap_j I^j</math>. Then, we have <math>IN = N</math>. Since <math>I</math> is contained in the [[Jacobson radical]], this forces <math>N = 0</math>.
==References==
==References==



Revision as of 18:21, 3 March 2008

This article defines a result where the base ring (or one or more of the rings involved) is Noetherian
View more results involving Noetherianness or Read a survey article on applying Noetherianness

This fact is an application of the following pivotal fact/result/idea: Artin-Rees lemma
View other applications of Artin-Rees lemma OR Read a survey article on applying Artin-Rees lemma

This fact is an application of the following pivotal fact/result/idea: Nakayama's lemma
View other applications of Nakayama's lemma OR Read a survey article on applying Nakayama's lemma

Statement

Let R be a Noetherian ring and I be an ideal inside R.

  • If M is a finitely generated R-module, then there exists rI such that:

(1r)(1IjM)=0

1Ij=0

Results used

Proof

The intersection equals its product with I

We first show that the intersection equals its product with I. This is the step where we se the Artin-Rees lemma.

Let:

N:=1IjM

Now consider the filtration:

MIMI2M

this is an I-adic filtration and the underlying ring is Noetherian, hence by the Artin-Rees lemma, the following filtration is also I-adic:

NIMNI2MN

Since each IjM contains N, the filtration below is the same as the filtration:

NNN

This being I-adic forces that IN=N.

Finding the element r

Since IN=N, we can find an element rI such that (1r)N=0. This is an application of the Cayley-Hamilton theorem: we first find the Cayley-Hamilton polynomial, then observe that 1 is a root of the polynomial, and then take the negative of the sum of all coefficients of higher degree terms.

Proving the intersection is trivial for an integral domain

If R is an integral domain, and we set M=R in the above, we find an element rI such that (1r)(jIj)=0. Thus 1r is a zero divisor on the intersection. Since R is an integral domain, one of these must hold:

  • 1r=0. This forces 1I, contradicting the assumption that I is a proper ideal
  • jIj=0

This completes the proof.

Proving that the intersection is trivial for a local ring

In this case, I is contained inside the unique maximal ideal m, which is the Jacobson radical. As above, let N=jIj. Then, we have IN=N. Since I is contained in the Jacobson radical, this forces N=0.

References

Textbook references