4

Let $R$ be a commutative Noetherian ring, and let $a\in R$ be a non-unit and a non-zerodivisor. Let $P$ be a prime ideal of $R$ such that $a\in P$. Prove that $\mathrm{ht}(P/Ra)=\mathrm{ht}(P)-1$. (Sharp, Steps in Commutative Algebra, Exercise 15.16)

This is what I have tried so far:

Proof. We know $\mathrm{ht}(P)\leq \mathrm{ht}(P/I)+n$, where $I$ is an ideal generated by $n$ elements. Hence $\mathrm{ht}(P)\leq \mathrm{ht}(P/Ra)+1$ therefore $\mathrm{ht}(P)-1\leq \mathrm{ht}(P/Ra)$.

But I cannot do the other inequality. Can you help me please?

user26857
  • 52,094
Pradip
  • 840

1 Answers1

3

Lemma 1: Let $R$ be a Noetherian ring and $(a) \subset I\in \mathrm{Spec(R)}$ such that $\mathrm{ht}(I) = 0$. Then $a$ is a divisor of $0$.

Proof: Consider the localized ring $R_{I}$. Clearly, $R_{I}$ is a Noetherian ring and $\mathrm{Spec}(R_{I}) = \{I':= I_I\}$ (since $I$ has height $0$ and there is a bijection between $\mathrm{Spec}(R_I)$ and $\{A\in \mathrm{Spec}(R);\ A\cap (R\setminus I) = \emptyset\}$) we can conclude that $R_I$ is an Artinian ring. Theorefore $I'$ is nilpotent. Thus, there exists $s\in R\setminus I$ and $n\in\mathbb{R}$, such that $s a^{n} = 0$. Implying that $a$ is a $0$ divisor.


Suppose by reductio ad absurdum that $\mathrm{ht}(P/(a)) = \mathrm{ht}(P) = n$. Then there there exists a saturated chain of prime ideals in $R/(a)$ such that $$Q_0 \subsetneq Q_1 \subsetneq \ldots\subsetneq Q_{n-1}\subsetneq P/(a). $$

It is well known that there is a bijection between the set $\mathrm{Spec}(R/(a))$ and $\{I\in \mathrm{Spec}(R); (a)\subset I\}$. Then $\forall i\in\{0,1,\ldots,n-1\}$, there exists a unique $P_i\in \mathrm{Spec}(R)$, such that $(a)\subset P_i$ and $P_i/(a) = Q_i$. Thus, we can define the following chain of prime ideals in $R$ $$P_0 \subsetneq P_1 \subsetneq \ldots\subsetneq P_{n-1}\subset P. $$ Since $\mathrm{ht}(P)=n$ the above chain is saturated. So $\mathrm{ht}(P_0) = 0$ and $(a)\subset P_0$. Using Lemma 1, $a$ is a zero divisor, which is a contradiction. Therefore $\mathrm{ht}(P/(a))<\mathrm{ht}(P)$.

Corollary 15.15, of the same book, states that the following inequality holds $$\mathrm{ht}(P/(a)) \leq \mathrm{ht}(P)\leq \mathrm{ht}(P/(a)) +1, $$ so, $$\left|\mathrm{ht}(P) - \mathrm{ht}(P/(a)) \right| \leq 1. $$

We already have proved that $ 0<\left|\mathrm{ht}(P) - \mathrm{ht}(P/(a)) \right|$. Therefore $$\mathrm{ht}(P) = \mathrm{ht}(P/(a)) +1.$$

$$\hspace{10cm}\square $$

  • I'm following the same reference as the author of the post "Steps in Commutative Algebra - R.Y.Sharp". I was having the same doubt as the OP, so I started the bounty. The inequality $\mathrm{ht}(P)−1≤\mathrm{ht}(P/Ra)$ comes from a result of the book (Corollary 15.15). The purpose of bounty was just to understand how to prove the reverse inequality. – Matheus Manzatto Jul 19 '20 at 22:23
  • 1
    Ok. Then the short answer is this: https://math.stackexchange.com/questions/518 – user26857 Jul 20 '20 at 08:46