Why is Z a Hilbert ring but Z_(p) not?

In Z the prime (0) is the intersection of the infinitely many maximal ideals (p), because a nonzero integer is divisible by only finitely many primes. In Z_(p) there is exactly one maximal ideal, pZ_(p), and it does not intersect down to (0). Localisation destroys maximal ideals, and the Hilbert condition is precisely a statement that there are enough of them.

Jacobson Rings, Finitely Generated Algebras and the Nullstellensatz

Q: Is (5.3) true without the domain hypothesis?

No. Z is J-semisimple and A = Z[x]/(x^2) is a finitely generated Z-algebra with rad A = (x) ≠ 0. The correct statement for general commutative rings is (5.4): Hilbertness ascends, and it yields rad A = Nil A rather than rad A = 0.

Q: Where exactly is J-semisimplicity used in the proof of (5.3)?

In exactly one line: to find a maximal ideal of R missing the nonzero element r_n s_0. Everything else is localisation, Nakayama and a contraction argument. That is also why the hypothesis cannot be weakened — Z_(p) ⊆ Q fails the conclusion of (5.5) for want of such a maximal ideal.

Q: What is the arithmetic analogue of a point?

A maximal ideal of a finitely generated Z-algebra, whose residue field is a finite field. The proof is the case distinction on characteristic given in the worked example, and it is the basis of reduction modulo p throughout arithmetic geometry.

Executive Summary

Snapper's Theorem says that adjoining variables to a commutative ring collapses the Jacobson radical onto the nilradical. The natural next question is which rings already have that property intrinsically, and whether it survives passage to finitely generated algebras. Rings with $rad = Nil$ in every quotient are the Hilbert rings, called Jacobson rings in most modern sources.

The technical core is a single ascent theorem $(5.3)$ : if $R \subseteq A$ are domains, $A$ is a finitely generated $R$ -algebra, and $rad R = 0$ , then $rad A = 0$ . From it follow the ascent of the Hilbert property $(5.4)$ , Zariski's Lemma and the weak and strong Nullstellensatz $(5.5)$ — the whole ideal–variety dictionary of classical algebraic geometry, obtained from radical theory rather than from Noether normalisation.

$rad = Nil$ Hilbert ring, in every quotient

AscendsTo f.g. algebras

$ℤ$ , $k$ , $k [x]$ Base Hilbert rings

1951Goldman, Krull

Overview

Two chains of ideals are attached to every commutative ring: the intersection of the primes, $Nil R$ , and the intersection of the maximals, $rad R$ . Since maximal ideals are prime, $Nil R \subseteq rad R$ always, and the gap measures how badly maximal ideals fail to detect the prime spectrum.

Nil R = ⋂_{𝔭 prime} 𝔭 \subseteq ⋂_{𝔪 maximal} 𝔪 = rad R

(5.H)

The Hilbert condition forces equality here, and in every quotient ring simultaneously.

For a local ring such as $ℤ_{(p)}$ or $k [[x]]$ the gap is maximal: there is one maximal ideal, so $rad$ is large while $Nil$ is zero. For $ℤ$ , for $k [x_{1}, \dots, x_{n}]$ and for every affine algebra the gap is zero. The dividing line is the Hilbert condition, and the substance of this page is that it is stable under the operation pass to a finitely generated algebra.

Learning Objectives

Define Hilbert ring and prove $rad = Nil$ for Hilbert rings and their quotients.
State $(5.3)$ with all its hypotheses and identify where each is used.
Follow the localisation-plus-Nakayama step that produces $𝔪 A \neq A$ .
Deduce $(5.4)$ and conclude that affine algebras over a field are Hilbert rings.
State $(5.5)$ , including the J-semisimplicity hypothesis, and derive Zariski's Lemma.
Derive the weak Nullstellensatz over an algebraically closed field, and identify the strong form as $rad A = Nil A$ .

Definitions

DefinitionHilbert ring

A commutative ring $R$ is a Hilbert ring if every prime ideal of $R$ is an intersection of maximal ideals. A Hilbert ring that is a domain is a Hilbert domain. The now dominant terminology is Jacobson ring; the two names denote the same class.

Finitely generated as an $R$ -algebra: $A = R [a_{1}, \dots, a_{m}]$ for finitely many $a_{i} \in A$ . Strictly weaker than being finitely generated as an $R$ -module: $ℚ = ℤ_{(p)} [1 / p]$ is a one-generator algebra but an infinitely generated module.
J-semisimple: $rad R = 0$ ; on this page always applied to commutative rings, where it means the maximal ideals intersect in zero.
$V (𝔞)$ and $I (X)$: The common zero set in $k^{n}$ of an ideal $𝔞 \subseteq k [x_{1}, \dots, x_{n}]$ , and the ideal of all polynomials vanishing on a subset $X \subseteq k^{n}$ .
$\sqrt{𝔞}$: The radical of an ideal: all $f$ with $f^{m} \in 𝔞$ for some $m \geq 1$ ; equal to the intersection of the primes containing $𝔞$ .

All rings on this page are commutative with identity, and every extension $R \subseteq A$ is unital.

Core Concepts

Elementary consequences of the Hilbert condition

PropositionBasic properties

Let $R$ be a Hilbert ring. Then (i) every quotient $R / 𝔞$ is a Hilbert ring; (ii) $rad R = Nil R$ ; (iii) if $R$ is in addition a domain, then $rad R = 0$ .

Proof

(i) The primes of $R / 𝔞$ correspond to primes $𝔭 \supseteq 𝔞$ . Writing $𝔭 = ⋂_{α} 𝔪_{α}$ with $𝔪_{α}$ maximal, each $𝔪_{α} \supseteq 𝔭 \supseteq 𝔞$ , so the whole expression descends to $R / 𝔞$ .

(ii) $Nil R$ is the intersection of all primes, each of which is an intersection of maximal ideals, so $Nil R$ is an intersection of maximal ideals and therefore contains $rad R$ . The reverse inclusion holds in any commutative ring.

(iii) In a domain $Nil R = 0$ , so (ii) gives $rad R = 0$ .

What has to be proved, and why it is not formal

By Snapper's Theorem the polynomial ring $R [t]$ over a J-semisimple domain is J-semisimple, so the transcendental step of an algebra extension is free. The difficulty is the algebraic step $A = R [a]$ with $a$ algebraic over $Frac R$ , where $A$ is a quotient of $R [t]$ and quotients can create radical from nothing — $rad ℤ = 0$ but $rad (ℤ / 4) \neq 0$ .

A = R [a_{1}, \dots, a_{m}]

reduce to

A = R [a]

a

transcendental: Snapper

a

algebraic: localise and apply Nakayama

The algebraic case is handled by inverting a single well-chosen element of $R$ . J-semisimplicity of $R$ is what guarantees a maximal ideal avoiding that element, and this is the only place the hypothesis enters — but it is indispensable.

Key Results

Theorem(5.3)Ascent of J-semisimplicity

Let $R \subseteq A$ be commutative domains such that $A$ is finitely generated as an $R$ -algebra, and suppose $R$ is J-semisimple. Then $A$ is J-semisimple.

Proof

Reduction. If $A = R [a_{1}, \dots, a_{m}]$ , insert the chain $R \subseteq R [a_{1}] \subseteq R [a_{1}, a_{2}] \subseteq \dots \subseteq A$ . Each step is a one-generator extension of domains, so it suffices to treat $A = R [a]$ .

Transcendental case. If $a$ is transcendental over $K = Frac R$ , then $A ≅ R [t]$ and $R$ is reduced, so $rad A = (Nil R) [t] = 0$ by Snapper's Theorem $(5.1)$ .

Algebraic case. Suppose $a$ is algebraic over $K$ and, for contradiction, that some $b \neq 0$ lies in $rad A$ . Both $a$ and $b$ are algebraic over $K$ , so after clearing denominators choose polynomials over $R$ of least possible degrees $n, m \geq 1$ satisfied by $a$ and $b$ respectively:

\sum_{j = 0}^{n} r_{j} a^{j} = 0 (r_{n} \neq 0), \sum_{i = 0}^{m} s_{i} b^{i} = 0 (s_{m} \neq 0) .

(5.3a)

From the second relation, $s_{0} = - b (\sum_{i \geq 1} s_{i} b^{i - 1}) \in b A \subseteq rad A$ . Moreover $s_{0} \neq 0$ : otherwise, since $A$ is a domain and $b \neq 0$ , we could cancel $b$ and obtain a relation for $b$ of degree $m - 1$ , contradicting minimality. As $A$ is a domain, $r_{n} s_{0} \neq 0$ , and since $rad R = 0$ there is a maximal ideal $𝔪 \subset R$ with $r_{n} s_{0} \notin 𝔪$ .

Localise at $S = R ∖ 𝔪$ . In $R_{𝔪} = S^{- 1} R$ the element $r_{n}$ is a unit, so dividing the first relation of $(5.3 a)$ by $r_{n}$ exhibits $a$ as integral over $R_{𝔪}$ ; hence $S^{- 1} A = R_{𝔪} [a]$ is a finitely generated $R_{𝔪}$ -module. It is nonzero, so Nakayama's Lemma $(4.22)$ gives $(rad R_{𝔪}) \cdot S^{- 1} A \neq S^{- 1} A$ , that is $𝔪 R_{𝔪} S^{- 1} A \neq S^{- 1} A$ . Consequently $𝔪 A \neq A$ , since $𝔪 A = A$ would survive localisation.

Choose a maximal ideal $𝔪' \subset A$ containing $𝔪 A$ . Then $𝔪' \cap R$ is a proper ideal of $R$ containing the maximal ideal $𝔪$ , so $𝔪' \cap R = 𝔪$ . But $s_{0} \in rad A \subseteq 𝔪'$ and $s_{0} \in R$ , so $s_{0} \in 𝔪$ — contradicting $r_{n} s_{0} \notin 𝔪$ . Hence $rad A = 0$ .

Corollary(5.4)Ascent of the Hilbert property

Let $R \subseteq A$ be commutative rings with $A$ finitely generated as an $R$ -algebra, and suppose $R$ is a Hilbert ring. Then $A$ is a Hilbert ring; in particular $rad A = Nil A$ . Taking $R = k$ a field: **every affine $k$ -algebra is a Hilbert ring.**

Proof

Let $𝔭 \subset A$ be prime. Then $A / 𝔭$ is a domain, finitely generated as an algebra over the domain $R / (𝔭 \cap R)$ , which is a Hilbert domain by part (i) of the Proposition above and hence J-semisimple by part (iii). Theorem $(5.3)$ gives $rad (A / 𝔭) = 0$ , which says exactly that $𝔭$ is the intersection of the maximal ideals of $A$ containing it. So $A$ is Hilbert, and $rad A = Nil A$ by the Proposition.

Theorem(5.5)Descent from a field

Let $R \subseteq A$ be commutative domains with $A$ finitely generated as an $R$ -algebra, and suppose $R$ is J-semisimple. If $A$ is a field, then $R$ is a field and $A / R$ is a finite algebraic field extension.

Proof

**Monogenic case $A = R [a]$ .** Since $A$ is a field it is not a polynomial ring over $R$ , so $a$ is algebraic over $Frac R$ ; take $\sum_{j = 0}^{n} r_{j} a^{j} = 0$ with $r_{n} \neq 0$ , $n \geq 1$ . Because $rad R = 0$ there is a maximal ideal $𝔪 \subset R$ with $r_{n} \notin 𝔪$ , and exactly as in the proof of $(5.3)$ localisation plus Nakayama yields $𝔪 A \neq A$ . But $A$ is a field, so its only proper ideal is $0$ ; hence $𝔪 A = 0$ and therefore $𝔪 = 0$ . A ring whose zero ideal is maximal is a field, so $R$ is a field, and $A = R [a]$ with $a$ algebraic over $R$ is a finite extension.

General case. Write $A = R [a_{1}, \dots, a_{m}]$ and set $R' = R [a_{1}]$ . By $(5.3)$ , $rad R' = 0$ , and $A$ is generated over $R'$ by $m - 1$ elements, so induction on $m$ shows $R'$ is a field and $A / R'$ is finite. Now apply the monogenic case to $R \subseteq R' = R [a_{1}]$ : $R$ is a field and $a_{1}$ is algebraic over $R$ . Then $R' / R$ is finite and $A / R'$ is finite, so $A / R$ is finite algebraic.

CorollaryZariski's Lemma

If $k$ is a field and $A$ is a field that is finitely generated as a $k$ -algebra, then $[A : k] < \infty$ . Equivalently, for every maximal ideal $𝔪$ of $k [x_{1}, \dots, x_{n}]$ the residue field $k [x_{1}, \dots, x_{n}] / 𝔪$ is a finite extension of $k$ .

CorollaryWeak Nullstellensatz

Let $k$ be algebraically closed and $𝔪$ a maximal ideal of $k [x_{1}, \dots, x_{n}]$ . Then $𝔪 = (x_{1} - b_{1}, \dots, x_{n} - b_{n})$ for a unique point $(b_{1}, \dots, b_{n}) \in k^{n}$ . Consequently every proper ideal of $k [x_{1}, \dots, x_{n}]$ has a common zero in $k^{n}$ .

Proof

By Zariski's Lemma the field $F = k [x_{1}, \dots, x_{n}] / 𝔪$ is finite over $k$ , hence equals $k$ as $k$ is algebraically closed. Let $b_{i} \in k$ be the image of $x_{i}$ . Then $x_{i} - b_{i} \in 𝔪$ , so $𝔪$ contains the maximal ideal $(x_{1} - b_{1}, \dots, x_{n} - b_{n})$ and equals it. A proper ideal lies in some maximal ideal, whose associated point is a common zero.

RemarkStrong Nullstellensatz

For $k$ algebraically closed and $𝔞 \subseteq k [x_{1}, \dots, x_{n}]$ an ideal, $I (V (𝔞)) = \sqrt{𝔞}$ . This is the geometric translation of $rad A = Nil A$ for the affine algebra $A = k [x_{1}, \dots, x_{n}] / 𝔞$ , which is $(5.4)$ : the functions vanishing at all $k$ -points are exactly those in every maximal ideal, and the Hilbert property identifies that intersection with the nilradical.

CounterexampleJ-semisimplicity cannot be dropped from (5.5)

Let $R = ℤ_{(p)}$ and $A = ℚ$ . Then $A = R [1 / p]$ is generated by one element as an $R$ -algebra, both are domains, and $A$ is a field — yet $R$ is not a field and $A / R$ is not algebraic. The hypothesis that fails is precisely $rad R = 0$ : here $rad ℤ_{(p)} = p ℤ_{(p)} \neq 0$ , and there is no maximal ideal of $R$ avoiding $p$ .

Proof Techniques and Method

The reusable moves behind (5.3)–(5.5).

Move 1

Invert one leading coefficient

An algebraic relation becomes an integral relation once its leading coefficient is a unit. Localising at a maximal ideal that avoids that coefficient converts finitely generated algebra into finitely generated module, which is the hypothesis Nakayama needs.

Move 2

Use rad R = 0 as a supply of maximal ideals

J-semisimplicity is used exactly once, and only in the form: for every nonzero $c \in R$ there is a maximal ideal missing $c$ . That reading makes clear why the hypothesis is unavoidable.

Move 3

Reduce to one generator, then to a prime quotient

Statements about finitely generated algebras are proved one generator at a time; statements about Hilbert rings are proved one prime at a time by passing to $A / 𝔭$ , where the domain hypothesis of $(5.3)$ becomes available.

The pattern localise, apply Nakayama, contract a maximal ideal is the same one that proves lying-over for integral extensions. What is special here is that integrality is not assumed — it is manufactured at one prime by localisation.

Reduce to one generatorFilter

A = R [a_{1}, \dots, a_{m}]

by intermediate domains and treat each step separately.

Split on transcendenceTranscendental generator: quote Snapper. Algebraic generator: proceed.

Pick a maximal ideal by J-semisimplicityChoose

𝔪

missing the product of the two leading coefficients.

Localise and apply NakayamaObtain

𝔪 A \neq A

and lift to a maximal ideal

𝔪'

A

over

𝔪

Contradict membership in the radicalThe constant term produced from the relation for

b

lies in

rad A \subseteq 𝔪'

, hence in

𝔪

— which was chosen to avoid it.

Worked Example

Maximal ideals of $ℤ [x_{1}, \dots, x_{n}]$ have finite residue fields

This is the arithmetic Nullstellensatz, and it comes out of $(5.4)$ and $(5.5)$ with no extra work.

First, $ℤ$ is a Hilbert ring: its primes are $(0)$ and the $(p)$ ; each $(p)$ is maximal, and $(0) = ⋂_{p} (p)$ because a nonzero integer has only finitely many prime divisors. Hence by $(5.4)$ every finitely generated $ℤ$ -algebra — every finitely generated commutative ring — is a Hilbert ring.

Now let $𝔪$ be a maximal ideal of $B = ℤ [x_{1}, \dots, x_{n}]$ and put $F = B / 𝔪$ , a field generated as a $ℤ$ -algebra by the images of the $x_{i}$ . Two cases:

** $char F = 0$ .** Then $ℤ ↪ F$ , and $F$ is a field finitely generated as a $ℤ$ -algebra with $rad ℤ = 0$ , so $(5.5)$ would force $ℤ$ to be a field. It is not, so this case is impossible.
** $char F = p > 0$ .** Then the image of $ℤ$ in $F$ is $𝔽_{p}$ , and $F$ is a field finitely generated as an $𝔽_{p}$ -algebra, so Zariski's Lemma makes $F / 𝔽_{p}$ finite.

ℤ [x_{1}, \dots, x_{n}] / 𝔪 ≅ 𝔽_{p^{d}} for some prime p and d \geq 1 .

(E.1)

Every maximal ideal of a finitely generated commutative ring of characteristic $0$ has finite residue field.

Concrete instance

Take $n = 1$ and $𝔪 = (5, x^{2} + x + 1) \subset ℤ [x]$ . Then $B / 𝔪 ≅ 𝔽_{5} [x] / (x^{2} + x + 1)$ . Since $x^{2} + x + 1$ has discriminant $- 3$ and $- 3 \equiv 2 (\mod 5)$ is a non-residue modulo $5$ (the squares mod $5$ are $1$ and $4$ ), the polynomial is irreducible and $B / 𝔪 ≅ 𝔽_{25}$ , in agreement with $(E . 1)$ .

Comparison and Classification

Hilbert or not
Ring	Hilbert?	$rad$ vs $Nil$	Reason
Any field $k$	yes	$0 = 0$	the only prime is $(0)$ , which is maximal
$ℤ$	yes	$0 = 0$	$(0) = ⋂_{p} (p)$
$k [x_{1}, \dots, x_{n}]$	yes	$0 = 0$	f.g. over the Hilbert ring $k$ , by (5.4)
Any affine $k$ -algebra	yes	equal	(5.4)
Any finitely generated commutative ring	yes	equal	f.g. over $ℤ$ , by (5.4)
$ℤ_{(p)}$	no	$p ℤ_{(p)} \neq 0$	$(0)$ is not an intersection of maximal ideals
$k [[x]]$	no	$(x) \neq 0$	one maximal ideal, and it is not $(0)$
$ℚ$ as a $ℤ_{(p)}$ -algebra	yes (it is a field)	$0 = 0$	but its Hilbertness is not inherited from a Hilbert base

Which hypothesis each result needs
	$R \subseteq A$ domains	$A$ f.g. as $R$ -algebra	$rad R = 0$	$R$ Hilbert
(5.3) $rad A = 0$	yes	yes	yes	no
(5.4) $A$ Hilbert	no	yes	no	yes
(5.5) $A$ field $\Rightarrow$ $R$ field	yes	yes	yes	no
Zariski's Lemma	yes	yes	free ( $R = k$ )	free
Weak Nullstellensatz	yes	yes	free	free

Which hypothesis each result needs

The domain hypothesis in $(5.3)$ is not decorative. Take $R = ℤ$ , which is J-semisimple, and $A = ℤ [x] / (x^{2})$ , a finitely generated $ℤ$ -algebra: $rad A = Nil A = (x) \neq 0$ . The right general statement for non-domains is $(5.4)$ , which concludes Hilbertness rather than J-semisimplicity.

Relationship Map

Commutative rings

Nil R \subseteq rad R

Hilbert (Jacobson) ringsevery prime is an intersection of maximals;

rad = Nil

in every quotient

Finitely generated commutative ringsf.g. over

ℤ

, hence Hilbert by

(5.4)

Affine $k$ -algebrasquotients of

k [x_{1}, \dots, x_{n}]

Affine domainsJ-semisimple by

(5.3)

$(5.3)$ ascent of J-semisimplicity — the engine
- gives
  - $(5.4)$ ascent of the Hilbert property
  - $(5.5)$ descent from a field
  - Zariski's Lemma ( $R = k$ )
- uses
  - Snapper $(5.1)$ for the transcendental step
  - Nakayama $(4.22)$ after localising
  - $rad R = 0$ to find a maximal ideal
- feeds
  - weak Nullstellensatz over $\bar{k}$
  - strong Nullstellensatz: $I (V (𝔞)) = \sqrt{𝔞}$
  - finiteness of residue fields of f.g. rings

Applications and Industry Use

Applications here means where this structure is used — inside mathematics and in the engineering and computing disciplines that consume it.

Algebraic geometry

The ideal–variety dictionary

Over an algebraically closed field, points of $k^{n}$ correspond to maximal ideals and the strong Nullstellensatz makes $𝔞 \mapsto V (𝔞)$ a bijection between radical ideals and closed sets. Every affine coordinate ring computation rests on this.

Symbolic computation

Polynomial system solving

The weak Nullstellensatz is the correctness statement behind Gröbner-basis consistency testing: a system has no solution over $\bar{k}$ exactly when $1$ lies in the ideal, which a Gröbner basis detects.

Engineering

Kinematics, vision, robotics

Inverse kinematics, camera calibration and constraint solving all reduce to deciding solvability of polynomial systems over $ℂ$ and extracting the solution variety — the Nullstellensatz is what licenses the algebraic answer as a geometric one.

Arithmetic

Residue fields of finitely generated rings

That every maximal ideal of a finitely generated commutative ring has finite residue field underpins reduction-modulo- $p$ arguments and the finite-field methods used in cryptography and coding theory.

The honest description of $(5.3)$ – $(5.5)$ themselves is that they are internal infrastructure: they are the reason the geometric dictionary exists, and they are almost never invoked by name outside algebra.

Computational Notes

Computational notes cover algorithms, cost and library behaviour rather than manufacturing process.

Consistency. For $𝔞 = (f_{1}, \dots, f_{s}) \subseteq k [x_{1}, \dots, x_{n}]$ with $k$ algebraically closed, $V (𝔞) = \emptyset$ iff a reduced Gröbner basis of $𝔞$ is ${1}$ — a direct algorithmic reading of the weak Nullstellensatz.
Radical membership. $f \in \sqrt{𝔞}$ iff $1 \in 𝔞 + (1 - y f)$ in $k [x_{1}, \dots, x_{n}, y]$ ; this Rabinowitsch trick reduces the strong Nullstellensatz to one consistency test.
Effective bounds. Certificates $1 = \sum g_{i} f_{i}$ exist with $\deg g_{i}$ bounded roughly by $d^{n}$ where $d = \max \deg f_{i}$ ; results of Brownawell and Kollár made such bounds explicit, and they are close to optimal.
Cost. Gröbner basis computation is doubly exponential in $n$ in the worst case and singly exponential for zero-dimensional systems; the theory here is decidable but not cheap.
Libraries. Macaulay2, Singular, Magma and Sage all implement radical, saturation and dimension for affine algebras; the Hilbert property is used implicitly whenever a package equates *no solutions over $\bar{k}$ * with the unit ideal.

Failure Modes and Common Mistakes

Do not drop $rad R = 0$ from $(5.5)$ : the extension $ℤ_{(p)} \subseteq ℚ$ is a counterexample to every conclusion of the theorem.
Do not apply $(5.3)$ to non-domains; use $(5.4)$ , whose conclusion is Hilbertness.
Do not expect the Hilbert property to descend: $ℚ$ is a Hilbert ring but is a finitely generated algebra over the non-Hilbert ring $ℤ_{(p)}$ .
Do not use the weak Nullstellensatz over a field that is not algebraically closed: $x^{2} + 1$ generates a proper ideal of $ℝ [x]$ with no real zero.
Do not confuse $\sqrt{𝔞}$ with $rad$ of a ring; the strong Nullstellensatz is a statement relating the two, not an identification of notation.

Historical Notes and Lessons Learned

1893HilbertHilbert proves the Nullstellensatz for polynomial rings over an algebraically closed field, establishing the correspondence between radical ideals and algebraic sets.
1947ZariskiZariski isolates the finiteness lemma — a field finitely generated as an algebra over a field is a finite extension — which becomes the standard route to the Nullstellensatz.
1951Goldman and KrullIndependently, Goldman and Krull identify the class of rings in which every prime is an intersection of maximal ideals and show that the Nullstellensatz is really a statement about this class.
1967EagonEagon gives the short radical-theoretic proof that J-semisimplicity ascends along finitely generated extensions of domains; this is the treatment Lam follows.
1974KaplanskyKaplansky's Commutative Rings popularises the name Hilbert ring and the systematic development from the Hilbert condition.
1987–88Effective NullstellensatzBrownawell and Kollár establish sharp degree bounds for Nullstellensatz certificates, turning a qualitative statement into a complexity-theoretic tool.

The methodological lesson is that the Nullstellensatz is not really about polynomials: it is about a class of rings closed under finitely generated extension. Once the class is identified, the geometry over $ℂ$ and the arithmetic over $ℤ$ are the same theorem.

Quick Reference

Hilbert ringevery prime is an intersection of maximal ideals

Consequence

rad = Nil

R

and in every quotient

$(5.3)$

R \subseteq A

domains,

A

f.g.

R

-algebra,

rad R = 0

\Rightarrow

rad A = 0

$(5.4)$

R

Hilbert,

A

f.g.

R

-algebra

\Rightarrow

A

Hilbert

$(5.5)$ same hypotheses as

(5.3)

;

A

a field

\Rightarrow

R

a field,

[A : R] < \infty

Zariski

A

a field, f.g. as a

k

-algebra

\Rightarrow

[A : k] < \infty

Weak NSS

k = \bar{k}

: maximal ideals of

k [x_{1}, \dots, x_{n}]

are

(x_{i} - b_{i})

Strong NSS

I (V (𝔞)) = \sqrt{𝔞}

Is my commutative ring $A$ a Hilbert ring?

It is finitely generated over $ℤ$ or over a fieldYes, by

(5.4)

. In particular

rad A = Nil A

and every residue field at a maximal ideal is finite over the base.

It is local of Krull dimension $\geq 1$ No. The unique maximal ideal is nonzero and contains no non-maximal prime as an intersection of maximals, so

rad A \neq Nil A

It is a localisation of a Hilbert ringNot necessarily —

ℤ_{(p)}

is a localisation of

ℤ

. Localisation at a multiplicative set can destroy the property; quotients and finitely generated extensions cannot.

It is a field, or a finite ringYes, trivially: in a field the only prime is maximal, and in a finite ring every prime is maximal.

Frequently Asked Questions

Why is

ℤ

a Hilbert ring but

ℤ_{(p)}

not?

In $ℤ$ the prime $(0)$ is the intersection of the infinitely many maximal ideals $(p)$ , because a nonzero integer is divisible by only finitely many primes. In $ℤ_{(p)}$ there is exactly one maximal ideal, $p ℤ_{(p)}$ , and it does not intersect down to $(0)$ . Localisation destroys maximal ideals, and the Hilbert condition is precisely a statement that there are enough of them.

(5.3)

true without the domain hypothesis?

No. $ℤ$ is J-semisimple and $A = ℤ [x] / (x^{2})$ is a finitely generated $ℤ$ -algebra with $rad A = (x) \neq 0$ . The correct statement for general commutative rings is $(5.4)$ : Hilbertness ascends, and it yields $rad A = Nil A$ rather than $rad A = 0$ .

Where exactly is J-semisimplicity used in the proof of

(5.3)

In exactly one line: to find a maximal ideal of $R$ missing the nonzero element $r_{n} s_{0}$ . Everything else is localisation, Nakayama and a contraction argument. That is also why the hypothesis cannot be weakened — $ℤ_{(p)} \subseteq ℚ$ fails the conclusion of $(5.5)$ for want of such a maximal ideal.

How does this compare with the Noether normalisation proof of the Nullstellensatz?

Noether normalisation writes an affine algebra as a module-finite extension of a polynomial ring and deduces Zariski's Lemma from integrality. The route here replaces global normalisation by local integrality at one well-chosen maximal ideal, and it proves more: the ascent statement holds over any J-semisimple base domain, not only over a field.

Does the weak Nullstellensatz need the field to be algebraically closed?

The point-form does. Zariski's Lemma holds over any field and says the residue field at a maximal ideal is a finite extension; only when $k$ is algebraically closed is that extension trivial, giving a point of $k^{n}$ . Over $ℝ$ , the maximal ideal $(x^{2} + 1)$ has residue field $ℂ$ and no real zero.

What is the arithmetic analogue of a point?

A maximal ideal of a finitely generated $ℤ$ -algebra, whose residue field is a finite field. The proof is the case distinction on characteristic given in the worked example, and it is the basis of reduction modulo $p$ throughout arithmetic geometry.

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §5, results (5.3)–(5.5) (pp. 72–74).
J. A. Eagon, “Finitely generated domains over Jacobson semi-simple rings”, American Mathematical Monthly 74 (1967).
O. Goldman, “Hilbert rings and the Hilbert Nullstellensatz”, Mathematische Zeitschrift 54 (1951), 136–140.
W. Krull, “Jacobsonsche Ringe, Hilbertscher Nullstellensatz, Dimensionstheorie”, Mathematische Zeitschrift 54 (1951), 354–387.
I. Kaplansky, Commutative Rings, revised edition, University of Chicago Press, 1974, Chapter 1, §3.
D. Eisenbud, Commutative Algebra with a View Toward Algebraic Geometry, Graduate Texts in Mathematics 150, Springer-Verlag, 1995, Chapter 4.

AI Suggested Questions

Give the Noether normalisation proof of Zariski's Lemma and compare it line by line with the localisation proof used here.
Prove that a ring is a Hilbert ring if and only if every radical ideal is an intersection of maximal ideals.
Show that $R$ Hilbert implies $R [x]$ Hilbert directly, without invoking $(5.4)$ .
Is a subring of a Hilbert ring a Hilbert ring? Give a proof or a counterexample.
How do the results of this page change if the base is a noncommutative ring finitely generated as an algebra?
Explain how the effective Nullstellensatz degree bounds interact with Gröbner basis complexity for zero-dimensional systems.
Formulate and prove the Nullstellensatz for finitely generated algebras over an arbitrary Hilbert ring rather than over a field.

Jacobson Rings and the Nullstellensatz

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Elementary consequences of the Hilbert condition

What has to be proved, and why it is not formal

Key Results

Proof Techniques and Method

Invert one leading coefficient

Use rad R = 0 as a supply of maximal ideals

Reduce to one generator, then to a prime quotient

Worked Example

Maximal ideals of $ℤ [x_{1}, \dots, x_{n}]$ have finite residue fields

Concrete instance

Comparison and Classification

Relationship Map

Applications and Industry Use

The ideal–variety dictionary

Polynomial system solving

Kinematics, vision, robotics

Residue fields of finitely generated rings

Computational Notes

Failure Modes and Common Mistakes

Historical Notes and Lessons Learned

Quick Reference

Frequently Asked Questions

References

AI Suggested Questions

Continue learning

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

Elementary consequences of the Hilbert condition

What has to be proved, and why it is not formal

06Key Results

07Proof Techniques and Method

Invert one leading coefficient

Use rad R = 0 as a supply of maximal ideals

Reduce to one generator, then to a prime quotient

08Worked Example

Maximal ideals of ℤ[x1,…,xn] have finite residue fields

Concrete instance

09Comparison and Classification

10Relationship Map

11Applications and Industry Use

The ideal–variety dictionary

Polynomial system solving

Kinematics, vision, robotics

Residue fields of finitely generated rings

12Computational Notes

13Failure Modes and Common Mistakes

14Historical Notes and Lessons Learned

15Quick Reference

16Frequently Asked Questions

17Related KEVOS Topics

18References

19AI Suggested Questions

Continue learning

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Key Results

Proof Techniques and Method

Worked Example

Maximal ideals of $ℤ [x_{1}, \dots, x_{n}]$ have finite residue fields

Comparison and Classification

Relationship Map

Applications and Industry Use

Computational Notes

Failure Modes and Common Mistakes

Historical Notes and Lessons Learned

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions