ProofComplete

Proof of Krull's Principal Ideal Theorem

We present the complete proof of Krull's Hauptidealsatz, one of the cornerstones of commutative algebra.


Proof

Theorem (Krull's Principal Ideal Theorem): Let RR be a Noetherian ring, x∈Rx \in R a non-unit, and p\mathfrak{p} a prime ideal minimal over (x)(x). Then ht⁑(p)≀1\operatorname{ht}(\mathfrak{p}) \leq 1.

Step 1: Reduction to the local case.

Replacing RR by RpR_\mathfrak{p}, we may assume (R,p)(R, \mathfrak{p}) is a Noetherian local ring with p\mathfrak{p} the unique minimal prime over (x)(x). We need to show ht⁑(p)≀1\operatorname{ht}(\mathfrak{p}) \leq 1, i.e., there is no chain p0⊊p1⊊p\mathfrak{p}_0 \subsetneq \mathfrak{p}_1 \subsetneq \mathfrak{p} of prime ideals.

Step 2: Suppose for contradiction that ht⁑(p)β‰₯2\operatorname{ht}(\mathfrak{p}) \geq 2.

Then there exists a chain p0⊊p1⊊p\mathfrak{p}_0 \subsetneq \mathfrak{p}_1 \subsetneq \mathfrak{p}. We will derive a contradiction.

Since p\mathfrak{p} is minimal over (x)(x), the ring R/(x)R/(x) has a unique minimal prime, namely p/(x)\mathfrak{p}/(x), so dim⁑R/(x)=0\dim R/(x) = 0. In particular, p/(x)\mathfrak{p}/(x) is both the maximal and the unique minimal prime of R/(x)R/(x).

Step 3: Symbolic powers.

Consider the nn-th symbolic power of p1\mathfrak{p}_1: p1(n)=p1nRp1∩R={r∈R:sr∈p1nΒ forΒ someΒ sβˆ‰p1}\mathfrak{p}_1^{(n)} = \mathfrak{p}_1^n R_{\mathfrak{p}_1} \cap R = \{r \in R : sr \in \mathfrak{p}_1^n \text{ for some } s \notin \mathfrak{p}_1\}

Define ideals an=(p1(n),x)/(x)\mathfrak{a}_n = (\mathfrak{p}_1^{(n)}, x) / (x) in the ring RΛ‰=R/(x)\bar{R} = R/(x). Since RΛ‰\bar{R} is a zero-dimensional Noetherian local ring (Artinian), the descending chain a1βŠ‡a2βŠ‡β‹―\mathfrak{a}_1 \supseteq \mathfrak{a}_2 \supseteq \cdots stabilizes: there exists NN such that an=aN\mathfrak{a}_n = \mathfrak{a}_N for all nβ‰₯Nn \geq N.

Step 4: Unraveling the stabilization.

The equality an=aN\mathfrak{a}_n = \mathfrak{a}_N for nβ‰₯Nn \geq N means: for every r∈p1(N)r \in \mathfrak{p}_1^{(N)}, we can write r=a+xbr = a + xb for some a∈p1(n)a \in \mathfrak{p}_1^{(n)} and b∈Rb \in R. Thus p1(N)βŠ†p1(n)+(x)β‹…RforΒ allΒ nβ‰₯N\mathfrak{p}_1^{(N)} \subseteq \mathfrak{p}_1^{(n)} + (x) \cdot R \quad \text{for all } n \geq N

Step 5: Working in Rp1R_{\mathfrak{p}_1}.

Localize at p1\mathfrak{p}_1. In Rp1R_{\mathfrak{p}_1}, the ideal (x)(x) is not contained in p1Rp1\mathfrak{p}_1 R_{\mathfrak{p}_1} (otherwise x∈p1⊊px \in \mathfrak{p}_1 \subsetneq \mathfrak{p}, contradicting minimality of p\mathfrak{p} over (x)(x); actually we need to be more careful β€” since p\mathfrak{p} is minimal over (x)(x) and p1⊊p\mathfrak{p}_1 \subsetneq \mathfrak{p}, the element xx is not in p1\mathfrak{p}_1). Wait β€” we need xβˆ‰p1x \notin \mathfrak{p}_1.

Indeed, since p\mathfrak{p} is minimal over (x)(x) and p1⊊p\mathfrak{p}_1 \subsetneq \mathfrak{p}, if x∈p1x \in \mathfrak{p}_1 then p1βŠ‡(x)\mathfrak{p}_1 \supseteq (x), contradicting the minimality of p\mathfrak{p} over (x)(x). So xβˆ‰p1x \notin \mathfrak{p}_1.

In Rp1R_{\mathfrak{p}_1}, the element xx becomes a unit. Therefore: p1(N)Rp1βŠ†p1(n)Rp1+(x)Rp1=p1nRp1+Rp1=Rp1\mathfrak{p}_1^{(N)} R_{\mathfrak{p}_1} \subseteq \mathfrak{p}_1^{(n)} R_{\mathfrak{p}_1} + (x) R_{\mathfrak{p}_1} = \mathfrak{p}_1^n R_{\mathfrak{p}_1} + R_{\mathfrak{p}_1} = R_{\mathfrak{p}_1}

This gives p1NRp1βŠ†p1nRp1\mathfrak{p}_1^N R_{\mathfrak{p}_1} \subseteq \mathfrak{p}_1^n R_{\mathfrak{p}_1} for all nn, hence p1NRp1βŠ†β‹‚np1nRp1\mathfrak{p}_1^N R_{\mathfrak{p}_1} \subseteq \bigcap_n \mathfrak{p}_1^n R_{\mathfrak{p}_1}.

Step 6: Applying the Krull Intersection Theorem.

By the Krull intersection theorem (valid in Noetherian local rings), β‹‚np1nRp1=0\bigcap_n \mathfrak{p}_1^n R_{\mathfrak{p}_1} = 0. Therefore p1NRp1=0\mathfrak{p}_1^N R_{\mathfrak{p}_1} = 0, which means p1Rp1\mathfrak{p}_1 R_{\mathfrak{p}_1} is nilpotent. Since Rp1R_{\mathfrak{p}_1} is a Noetherian local ring with nilpotent maximal ideal, it has dimension 00.

But p0Rp1\mathfrak{p}_0 R_{\mathfrak{p}_1} is a prime ideal strictly contained in p1Rp1\mathfrak{p}_1 R_{\mathfrak{p}_1}, giving dim⁑Rp1β‰₯1\dim R_{\mathfrak{p}_1} \geq 1 β€” a contradiction.

Therefore ht⁑(p)≀1\operatorname{ht}(\mathfrak{p}) \leq 1. β–‘\square

β– 

RemarkHistorical significance

Krull proved this theorem in 1928. It was the first major achievement of the abstract approach to dimension theory. Together with its converse (that every height-one prime is minimal over a principal ideal), it shows that height-one primes are precisely the "codimension-one" phenomena, providing the algebraic foundation for the theory of divisors in algebraic geometry.