Local analytic Nullstellensatz and the ideal–germ correspondence

Intuition Beginner

In one variable, a holomorphic function near a point is pinned down up to a non-vanishing factor by where it vanishes and how hard. The zero set is a handful of points, and the algebra of the functions and the geometry of the points line up cleanly. You want the same dictionary in many variables: a way to pass from a list of functions to the shape they cut out, and back from a shape to the functions that vanish on it.

There are two natural directions. Start with a collection of holomorphic functions near the origin and look at the set where they all vanish: a curve, a surface, a web of vanishing. Or start with such a set and collect every function that vanishes on it. These two moves go opposite ways, and you want to know how close they come to being inverse to each other.

The catch is repetition. The function $z$ and the function $z$ squared vanish on the same set, the single point or line where $z$ is zero. So passing to the zero set forgets the difference between a function and its powers. The local analytic Nullstellensatz says that forgetting powers is the only thing the dictionary loses, and it says exactly how to put the lost information back.

Visual Beginner

Picture two facing columns. The left column lists algebraic data near the origin: collections of holomorphic functions, packaged so that if a function is in the collection then so is any multiple of it. The right column lists geometric data: the shapes those functions carve out, the zero sets passing through the origin.

A line runs from each collection on the left to the shape it cuts out on the right, and a line runs back from each shape to all the functions vanishing on it. Following an arrow over and then back does not always return you to where you started. It returns you to a slightly fattened version: every function whose some power was in your original collection. The picture below shows the two columns, the over-and-back loop, and the fattening that the loop performs.

Worked example Beginner

Work in one variable near the origin, where everything can be seen by hand. A collection of holomorphic functions that is closed under multiples is, in this setting, always the set of multiples of a single power of the coordinate: all multiples of $z$, or all multiples of $z$ squared, and so on, plus the collection containing only the zero function.

Take the collection of all multiples of $z$ squared. Every such function vanishes at the origin and only there, so the shape it cuts out is the single point at the origin. Now collect every holomorphic function vanishing at that point: that is every function with no constant term, which is exactly the multiples of $z$. So the over-and-back loop sent "multiples of $z$ squared" to "multiples of $z$".

The collection grew. It now contains $z$ itself, which was not a multiple of $z$ squared. What got added is precisely the functions one of whose powers was already present: $z$ squared is a power of $z$. The two collections that survive the loop unchanged are the multiples of $z$ and the zero collection, matching the two shapes available, the whole disc and the single origin.

What this tells us: passing to a zero set and back fattens a collection to include every function some power of which it already held, and nothing more.

Check your understanding Beginner

Formal definition Intermediate+

Let ${}_n\mathcal{O}=\mathbb{C}\{z_1,\dots ,z_n\}$ denote the ring of germs at the origin of holomorphic functions on ${\mathbb{C}}^n$, the convergent power series in $n$ variables. By the Weierstrass theory 06.10.14 it is a Noetherian local ring with maximal ideal $\mathfrak{m}=\{f:f(0)=0\}$, residue field $\mathbb{C}$, and it is a unique factorisation domain. Germs are represented by holomorphic functions on small neighbourhoods of $0$; two representatives define the same germ when they agree on some neighbourhood.

An analytic-set germ at the origin is the germ $V$ of a set of the form $\{z:g_1(z)=\cdots =g_m(z)=0\}$ for finitely many $g_i$ holomorphic near $0$, where two such sets define the same germ when they coincide on some neighbourhood of $0$. The collection of analytic-set germs is partially ordered by inclusion of germs.

Two operators link ideals and germs. For an ideal $I\subseteq {}_n\mathcal{O}$, define the zero-set germ $$ \mathcal{V}(I) ;=; \text{germ at } 0 \text{ of } {z : f(z) = 0 \text{ for all } f \in I}. $$ By Noetherianity $I=(f_1,\dots ,f_m)$ is finitely generated, so $\mathcal{V}(I)$ is the common zero germ of finitely many germs and is an analytic-set germ. For an analytic-set germ $V$, define the ideal of $V$ $$ \mathcal{I}(V) ;=; { f \in {}_n\mathcal{O} : f \text{ vanishes on a representative of } V \text{ near } 0 }. $$ This is an ideal of ${}_n\mathcal{O}$, and it is radical: if $f^k\in \mathcal{I}(V)$ then $f^k$ vanishes on $V$, hence $f$ does, so $f\in \mathcal{I}(V)$. Recall the radical of an ideal $I$, $$ \sqrt{I} ;=; { f \in {}_n\mathcal{O} : f^k \in I \text{ for some } k \ge 1 }, $$ the smallest radical ideal containing $I$.

Both operators reverse inclusions: $I\subseteq J$ gives $\mathcal{V}(I)\supseteq \mathcal{V}(J)$, and $V\subseteq W$ gives $\mathcal{I}(V)\supseteq \mathcal{I}(W)$. Composing, $\mathcal{I}(\mathcal{V}(I))\supseteq I$ always, and since the left side is radical, $\mathcal{I}(\mathcal{V}(I))\supseteq \sqrt{I}$. The depth of the subject is the reverse inclusion.

Counterexamples to common slips

• The germ $V$ must be cut out by holomorphic equations. The germ of a real line $\{\mathrm{Im}{z}_1=0\}$ in $\mathbb{C}$ is not an analytic-set germ, and $\mathcal{I}$ of it is not an ideal of ${}_n\mathcal{O}$; the correspondence is internal to the holomorphic category.
• $\mathcal{V}(I)$ depends only on $\sqrt{I}$: the ideals $(z_1^2)$ and $(z_1)$ have the same zero germ $\{z_1=0\}$. Reading geometry off $I$ rather than $\sqrt{I}$ overcounts.
• An ideal need not be its own radical even when prime-looking: $(z_1^2+z_2^3)$ is radical in ${}_2\mathcal{O}$ (its generator is square-free), but $(z_1^2)$ is not. Square-freeness of a single generator in a UFD is what controls whether the principal ideal is radical.

Key theorem with proof Intermediate+

Theorem (Rückert Nullstellensatz, Rückert 1933). For every ideal $I\subseteq {}_n\mathcal{O}$, $$ \mathcal{I}(\mathcal{V}(I)) ;=; \sqrt{I}. $$ Consequently $\mathcal{V}$ and $\mathcal{I}$ restrict to mutually inverse, inclusion-reversing bijections between the radical ideals of ${}_n\mathcal{O}$ and the analytic-set germs at the origin.

Proof. The inclusion $\mathcal{I}(\mathcal{V}(I))\supseteq \sqrt{I}$ holds for the elementary reason recorded above: if $f^k\in I$ then $f^k$, hence $f$, vanishes on $\mathcal{V}(I)$, so $f\in \mathcal{I}(\mathcal{V}(I))$, and that ideal is radical. The work is the reverse inclusion $\mathcal{I}(\mathcal{V}(I))\subseteq \sqrt{I}$: a germ vanishing on the common zero set of $I$ has a power lying in $I$ ^{[Gunning Vol. II]}.

Since ${}_n\mathcal{O}$ is Noetherian, $\sqrt{I}={\mathfrak{p}}_1\cap \cdots \cap {\mathfrak{p}}_r$ is a finite intersection of prime ideals (the minimal primes over $I$), and $\mathcal{V}(I)=\mathcal{V}({\mathfrak{p}}_1)\cup \cdots \cup \mathcal{V}({\mathfrak{p}}_r)$. A germ $f$ vanishing on $\mathcal{V}(I)$ vanishes on each $\mathcal{V}({\mathfrak{p}}_j)$, and if each such $f$ lies in ${\mathfrak{p}}_j$ then $f\in {\bigcap }_j{\mathfrak{p}}_j=\sqrt{I}$. So it suffices to prove $\mathcal{I}(\mathcal{V}(\mathfrak{p}))=\mathfrak{p}$ for a prime ideal $\mathfrak{p}$.

Fix a prime $\mathfrak{p}⊊{}_n\mathcal{O}$ and set $d=\dim {}_n\mathcal{O}/\mathfrak{p}$. The local parametrisation theorem, itself a consequence of iterated Weierstrass preparation 06.10.14, supplies a linear coordinate choice and an integer $d$ so that the inclusion ${}_d\mathcal{O}\hookrightarrow {}_n\mathcal{O}/\mathfrak{p}$ is a finite injective map of local rings: the quotient is a finitely generated ${}_d\mathcal{O}$-module, and the field of fractions $\mathcal{K}=\mathrm{Frac}({}_n\mathcal{O}/\mathfrak{p})$ is a finite field extension of $\mathcal{M}=\mathrm{Frac}({}_d\mathcal{O})$, of some degree $q$. Geometrically this realises $\mathcal{V}(\mathfrak{p})$ as a $q$-sheeted branched cover of a $d$-dimensional polydisc in the first $d$ coordinates.

By the primitive element theorem $\mathcal{K}=\mathcal{M}(\theta )$ for a single $\theta$, integral over ${}_d\mathcal{O}$, with monic minimal polynomial $P(z^{\prime };T)\in {}_d\mathcal{O}[T]$ of degree $q$ whose discriminant $\delta (z^{\prime })\in {}_d\mathcal{O}$ is a non-zero germ. Off the branch locus $\{\delta =0\}$ the cover is unramified with $q$ distinct sheets, and a germ $f\in {}_n\mathcal{O}$ restricts on $\mathcal{V}(\mathfrak{p})$ to a function determined by its values on these sheets.

Now take $f\in \mathcal{I}(\mathcal{V}(\mathfrak{p}))$, so $f$ vanishes on $\mathcal{V}(\mathfrak{p})$. Its image $\bar{f}\in {}_n\mathcal{O}/\mathfrak{p}$ vanishes on every sheet of the cover over the complement of $\{\delta =0\}$, where the points of $\mathcal{V}(\mathfrak{p})$ are exactly the $q$ values of the chart. The element $\bar{f}\in \mathcal{K}$ thus vanishes identically as an algebraic function over $\mathcal{M}$, which forces $\bar{f}=0$ in $\mathcal{K}$, hence $\bar{f}=0$ in the domain ${}_n\mathcal{O}/\mathfrak{p}$. Therefore $f\in \mathfrak{p}$. This is the analytic content reduced to a statement of the Hilbert Nullstellensatz over the field $\mathcal{M}$: a function on a finite extension vanishing on all geometric points lies in the defining prime, exactly the algebraic theorem at 04.02.07 transported through the finite extension preparation builds. Hence $\mathcal{I}(\mathcal{V}(\mathfrak{p}))\subseteq \mathfrak{p}$, with equality, completing the prime case and the theorem.

The bijection follows formally. On radical ideals $\mathcal{I}(\mathcal{V}(\cdot ))=\mathrm{id}$ by the equality just proved with $\sqrt{I}=I$; on analytic-set germs $\mathcal{V}(\mathcal{I}(\cdot ))=\mathrm{id}$ because every analytic-set germ $V$ equals $\mathcal{V}(\mathcal{I}(V))$ (a germ is recovered from its full vanishing ideal). Both maps reverse inclusion, so they are mutually inverse anti-isomorphisms of partially ordered sets. $\square$

Bridge. This correspondence builds toward the structure theory of analytic sets in 06.10.18, where the prime decomposition $\sqrt{I}={\mathfrak{p}}_1\cap \cdots \cap {\mathfrak{p}}_r$ used in the proof becomes the decomposition of a germ into irreducible components, and it appears again in 06.10.19 where ${}_n\mathcal{O}/\mathcal{I}(V)$ is named the local ring of the germ $V$. The foundational reason the analytic and algebraic theories agree is the local parametrisation theorem: Weierstrass preparation 06.10.14 turns the prime ${}_n\mathcal{O}/\mathfrak{p}$ into a finite extension of a coordinate ring ${}_d\mathcal{O}$, and over that extension the analytic vanishing statement is exactly the algebraic Hilbert Nullstellensatz of 04.02.07. This is exactly the move that lets a convergent-series problem be settled by polynomial algebra, and it generalises the one-variable picture where the only radical ideals are $(0)$ and $\mathfrak{m}$. The central insight is that the radical, not the ideal, is the geometric invariant, so the bridge is the passage from $I$ to $\sqrt{I}$, and putting these together the lattice of germs is recovered intact as the lattice of radical ideals read backwards.

Exercises Intermediate+

Advanced results Master

The prime case carries the full content; the structure below records what the bijection organises and how the analytic theorem sits against its algebraic and formal neighbours.

Decomposition and the geometry of the radical. Because ${}_n\mathcal{O}$ is Noetherian, every radical ideal is a finite irredundant intersection of primes $\sqrt{I}={\mathfrak{p}}_1\cap \cdots \cap {\mathfrak{p}}_r$, and the bijection turns this into $\mathcal{V}(I)=V_1\cup \cdots \cup V_r$ with $V_j=\mathcal{V}({\mathfrak{p}}_j)$ the irreducible components, uniquely determined and none contained in the union of the others ^{[Gunning–Rossi]}. The dimension $\dim V_j=\dim {}_n\mathcal{O}/{\mathfrak{p}}_j$ is the transcendence degree of $\mathrm{Frac}({}_n\mathcal{O}/{\mathfrak{p}}_j)$ over $\mathbb{C}$, equal to the integer $d$ of the local parametrisation. The Nullstellensatz is thus the gateway through which the prime spectrum of the local ring becomes the component structure of the germ.

The maximal ideal and the Nullstellensatz at a point. The germ of the single point $\{0\}$ is $\mathcal{V}(\mathfrak{m})$, and $\mathcal{I}(\{0\})=\mathfrak{m}$, the maximal ideal. This is the analytic shadow of the weak Nullstellensatz: the maximal ideals of ${}_n\mathcal{O}$ that arise as $\mathcal{I}$ of a germ are exactly the geometric ones, and at the origin there is only one. Unlike the polynomial ring, where maximal ideals correspond to all points of affine space over the algebraically closed field, the local ring sees a single point, so its only height-$n$ prime appearing as a vanishing ideal is $\mathfrak{m}$.

Contrast with the algebraic Hilbert Nullstellensatz. Over $\mathbb{C}[x_1,\dots ,x_n]$ the Hilbert Nullstellensatz at 04.02.07 states $\mathcal{I}(\mathcal{V}(I))=\sqrt{I}$ with $\mathcal{V}(I)\subseteq {\mathbb{C}}^n$ a global affine variety, and its engine is Noether normalisation: $\mathbb{C}[x]/\mathfrak{p}$ is finite over a polynomial subring $\mathbb{C}[y_1,\dots ,y_d]$. The analytic statement is identical in form, but the engine is Weierstrass preparation 06.10.14 in place of Noether normalisation, the polynomial subring is replaced by the convergent coordinate ring ${}_d\mathcal{O}$, and the global affine space is replaced by a germ. The convergent-series machinery is doing the work the algebraically closed base field does in the polynomial case: it manufactures, from a prime ideal, an honest finite branched cover whose finitely many sheets are the geometric points the algebraic theorem then judges. This is the precise sense in which the analytic Nullstellensatz reduces to the polynomial Nullstellensatz over ${}_n\mathcal{O}$ regarded as a finite extension, rather than being a separate theorem.

Formal and convergent agree, and where they part. The same statement and proof hold verbatim for the formal power series ring $\mathbb{C}[[z_1,\dots ,z_n]]$, since formal Weierstrass preparation supplies the same finite extensions; the radical correspondence is insensitive to convergence. Convergence reappears one level up, in the realisation of $\mathcal{V}(\mathfrak{p})$ as a genuine analytic subset of a neighbourhood with actual points to vanish on — the formal ring has no points, so the geometric side of the bijection has no formal analogue. This is the reason the convergent ${}_n\mathcal{O}$, not its formal completion, is the carrier of the local analytic geometry, and the reason the unit ships without a Mathlib formalisation: the formal Nullstellensatz is reachable from existing algebra, but the geometric germ side is not yet present.

Synthesis. The Rückert Nullstellensatz is the foundational reason the local geometry of analytic sets is governed by the commutative algebra of ${}_n\mathcal{O}$ read through the radical. The local parametrisation theorem, built from Weierstrass preparation 06.10.14, realises each prime quotient as a finite branched cover, and on that cover the analytic vanishing statement is exactly the Hilbert Nullstellensatz of 04.02.07; putting these together, the analytic theorem is not independent of the algebraic one but its transport along a preparation-built finite extension. The central insight is that the radical, and not the ideal, is the geometric invariant, so the lattice of analytic-set germs is dual to the lattice of radical ideals, prime ideals are dual to irreducible germs, and the dimension of a germ is dual to the Krull dimension of its local ring. This is exactly the duality that the component decomposition, the local ring of 06.10.19, and the coherence theory of 06.10.20 all read off; the bridge in every case is the passage from an ideal to its radical and back through a finite cover, and this pattern recurs wherever a space is recovered from its ring of functions, generalising the affine dictionary of 04.02.07 to the analytic local setting.

Full proof set Master

Proposition 1 (the elementary inclusion). For every ideal $I\subseteq {}_n\mathcal{O}$, $\mathcal{I}(\mathcal{V}(I))\supseteq \sqrt{I}$.

Proof. Let $f\in \sqrt{I}$, so $f^k\in I$ for some $k\ge 1$. Then $f^k$ vanishes on $\mathcal{V}(I)$ by definition of the zero germ. At each point $p$ of a representative of $\mathcal{V}(I)$, $f(p{)}^k=0$ in $\mathbb{C}$, so $f(p)=0$; hence $f$ vanishes on the representative and $f\in \mathcal{I}(\mathcal{V}(I))$. Thus $\sqrt{I}\subseteq \mathcal{I}(\mathcal{V}(I))$. $\square$

Proposition 2 (reduction to primes). If $\mathcal{I}(\mathcal{V}(\mathfrak{p}))=\mathfrak{p}$ for every prime $\mathfrak{p}\subseteq {}_n\mathcal{O}$, then $\mathcal{I}(\mathcal{V}(I))=\sqrt{I}$ for every ideal $I$.

Proof. By Noetherianity, $\sqrt{I}={\mathfrak{p}}_1\cap \cdots \cap {\mathfrak{p}}_r$ with the ${\mathfrak{p}}_j$ the minimal primes over $I$. Then $\mathcal{V}(I)=\mathcal{V}(\sqrt{I})={\bigcup }_j\mathcal{V}({\mathfrak{p}}_j)$, since a point annihilates $\sqrt{I}$ exactly when it annihilates some minimal prime is replaced by: a point lies in $\mathcal{V}(\sqrt{I})$ iff every element of $\sqrt{I}={\bigcap }_j{\mathfrak{p}}_j$ vanishes there, and the standard prime-avoidance translation gives the union of the $\mathcal{V}({\mathfrak{p}}_j)$. A germ $f\in \mathcal{I}(\mathcal{V}(I))$ vanishes on each $\mathcal{V}({\mathfrak{p}}_j)$, so $f\in \mathcal{I}(\mathcal{V}({\mathfrak{p}}_j))={\mathfrak{p}}_j$ for every $j$ by hypothesis, whence $f\in {\bigcap }_j{\mathfrak{p}}_j=\sqrt{I}$. With Proposition 1 this gives equality. $\square$

Proposition 3 (the prime case). For a prime $\mathfrak{p}⊊{}_n\mathcal{O}$, $\mathcal{I}(\mathcal{V}(\mathfrak{p}))=\mathfrak{p}$.

Proof. The inclusion $\supseteq$ is Proposition 1 with $\sqrt{\mathfrak{p}}=\mathfrak{p}$. For $\subseteq$, set $d=\dim {}_n\mathcal{O}/\mathfrak{p}$. By the local parametrisation theorem ^[Whitney] — iterated Weierstrass preparation 06.10.14 applied to a generating set of $\mathfrak{p}$ — choose linear coordinates $(z^{\prime },z^{\prime \prime })=(z_1,\dots ,z_d,z_{d+1},\dots ,z_n)$ so that the projection germ exhibits ${}_d\mathcal{O}\hookrightarrow A:={}_n\mathcal{O}/\mathfrak{p}$ as a finite injective extension of local rings. Then $\mathcal{K}=\mathrm{Frac}(A)$ is a finite field extension of $\mathcal{M}=\mathrm{Frac}({}_d\mathcal{O})$ of degree $q=[\mathcal{K}:\mathcal{M}]$, and by the primitive element theorem $\mathcal{K}=\mathcal{M}(\theta )$ with $\theta \in A$ having monic minimal polynomial $P(z^{\prime };T)\in {}_d\mathcal{O}[T]$ of degree $q$ and non-zero discriminant $\delta =\delta (z^{\prime })\in {}_d\mathcal{O}$.

On a polydisc $U^{\prime }\times U^{\prime \prime }$ chosen so all data converge, the analytic set $V=\mathcal{V}(\mathfrak{p})$ projects onto $U^{\prime }$ as a $q$-sheeted cover, unramified over $U^{\prime }\setminus \{\delta =0\}$; over such a regular value $z^{\prime }$ the $q$ points of $V$ are the $q$ distinct roots $T={\theta }_1(z^{\prime }),\dots ,{\theta }_q(z^{\prime })$ of $P(z^{\prime };\cdot )$, and these are exactly the geometric points of the cover. Let $f\in \mathcal{I}(V)$, with image $\bar{f}\in A\subseteq \mathcal{K}$. Write $\bar{f}=R(z^{\prime };\theta )$ for a representative $R\in {}_d\mathcal{O}[T]$ of degree $<q$ (reduce modulo $P$). For each regular $z^{\prime }$, the value of $f$ at the sheet ${\theta }_i(z^{\prime })$ is $R(z^{\prime };{\theta }_i(z^{\prime }))$, and $f$ vanishes on $V$, so $R(z^{\prime };{\theta }_i(z^{\prime }))=0$ for all $i=1,\dots ,q$. A polynomial of degree $<q$ in $T$ with $q$ distinct roots ${\theta }_1(z^{\prime }),\dots ,{\theta }_q(z^{\prime })$ is the zero polynomial, so $R(z^{\prime };\cdot )\equiv 0$ for every regular $z^{\prime }$, hence $R\equiv 0$ by continuity across $\{\delta =0\}$. Therefore $\bar{f}=0$ in $\mathcal{K}$, so $\bar{f}=0$ in the domain $A$, that is $f\in \mathfrak{p}$. $\square$

Proposition 4 (the order-reversing bijection). The maps $\mathcal{V}$ and $\mathcal{I}$ restrict to mutually inverse, inclusion-reversing bijections between radical ideals of ${}_n\mathcal{O}$ and analytic-set germs at $0$.

Proof. By Propositions 1–3, $\mathcal{I}(\mathcal{V}(I))=\sqrt{I}$, which equals $I$ when $I$ is radical; so $\mathcal{I}\circ \mathcal{V}=\mathrm{id}$ on radical ideals. For the other composite, let $V$ be an analytic-set germ. Then $\mathcal{V}(\mathcal{I}(V))\supseteq V$ because every germ in $\mathcal{I}(V)$ vanishes on $V$. For the reverse, $V=\mathcal{V}(J)$ for some finitely generated ideal $J$ (the germ is cut out by finitely many germs), and $J\subseteq \mathcal{I}(\mathcal{V}(J))=\mathcal{I}(V)$, so applying the order-reversing $\mathcal{V}$ gives $\mathcal{V}(\mathcal{I}(V))\subseteq \mathcal{V}(J)=V$. Hence $\mathcal{V}(\mathcal{I}(V))=V$, so $\mathcal{V}\circ \mathcal{I}=\mathrm{id}$ on germs. Both operators reverse inclusion, so the inverse bijections are anti-isomorphisms of partially ordered sets. $\square$

Proposition 5 (prime $\iff$ irreducible). Under the bijection, the prime ideals correspond exactly to the irreducible analytic-set germs.

Proof. A radical ideal $\mathfrak{a}$ is prime iff it is not an intersection of two strictly larger radical ideals: if $\mathfrak{a}={\mathfrak{a}}_1\cap {\mathfrak{a}}_2$ with both strictly larger, picking $f_i\in {\mathfrak{a}}_i\setminus \mathfrak{a}$ gives $f_1{f}_2\in \mathfrak{a}$ with neither factor in $\mathfrak{a}$, so $\mathfrak{a}$ is not prime; conversely a non-prime radical ideal $\mathfrak{a}$ has $fg\in \mathfrak{a}$ with $f,g\notin \mathfrak{a}$, and $\mathfrak{a}=(\mathfrak{a}+(f))\cap (\mathfrak{a}+(g))$ after passing to radicals exhibits the splitting. The bijection sends intersections of ideals to unions of germs and reverses strict inclusions, so $\mathfrak{a}$ prime corresponds to $V=\mathcal{V}(\mathfrak{a})$ admitting no decomposition into two strictly smaller analytic-set germs, that is to $V$ irreducible. $\square$

Connections Master

• Weierstrass preparation and division 06.10.14. The entire proof rests on this prerequisite. Preparation makes ${}_n\mathcal{O}$ Noetherian and factorial, supplying the finite intersection of primes and the radical operation; iterated preparation is the local parametrisation theorem that realises each prime quotient as a finite branched cover of ${}_d\mathcal{O}$. The finite extension on which the Hilbert step runs is exactly the one preparation builds.

• Analytic sets and their decomposition 06.10.18. This unit's bijection is the algebraic engine of that one. The decomposition $\sqrt{I}={\mathfrak{p}}_1\cap \cdots \cap {\mathfrak{p}}_r$ becomes the unique decomposition of an analytic-set germ into irreducible components, with the components' dimensions read as transcendence degrees, so the structure theory of analytic sets is the geometric face of the radical-ideal lattice established here.

• The local ring of an analytic set 06.10.19. The quotient ${}_n\mathcal{O}/\mathcal{I}(V)$ is the local ring of the germ $V$, and the Nullstellensatz is what makes $\mathcal{I}(V)$ the correct radical ideal to quotient by: it guarantees ${}_n\mathcal{O}/\mathcal{I}(V)$ is reduced and that its prime spectrum reproduces the component structure of $V$.

• Coherent analytic sheaves 06.10.20. The stalkwise identity $\mathcal{I}(\mathcal{V}(I))=\sqrt{I}$ proved here is the local input to coherence of the ideal sheaf of an analytic set: each stalk $\mathcal{I}(V)$ is a finitely generated radical ideal of ${}_n\mathcal{O}$, and the Oka coherence theorem promotes this stalkwise finiteness to coherence of the sheaf of vanishing ideals.

• Nullstellensatz and dimension theory 04.02.07. The algebraic original. The analytic theorem is its transport to the convergent-germ setting: same radical-ideal correspondence, with Weierstrass preparation replacing Noether normalisation and the convergent coordinate ring ${}_d\mathcal{O}$ replacing the polynomial subring. The Hilbert Nullstellensatz over the function field is the step that closes the analytic proof.

Historical & philosophical context Master

The analytic Nullstellensatz was proved by Walther Rückert in his 1933 paper Zum Eliminationsproblem der Potenzreihenideale ^{[Rückert 1933]} (Mathematische Annalen 107, 259–281), which established the radical–germ correspondence for ideals of convergent power series and placed the local theory of analytic sets on an algebraic foundation. Rückert's method was the elimination theory of power series ideals, carried out through the Weierstrass preparation theorem that Karl Weierstrass had developed in his Berlin lectures of the 1860s; the reduction of an analytic problem to a finite extension of a coordinate ring is already the substance of Rückert's argument, and it is the analytic counterpart of the elimination theory underlying Hilbert's 1893 algebraic Nullstellensatz.

The result was systematised in the mid-twentieth century as the local foundations of several-variable complex analysis took their modern form. Hellmuth Behnke and Peter Thullen catalogued the local theory, and the Oka–Cartan school recast the correspondence sheaf-theoretically, with the radical-ideal statement becoming the stalk computation behind coherence of ideal sheaves; the standard expositions are Gunning–Rossi 1965 ^{[Gunning–Rossi]} and Gunning's later three-volume Introduction to Holomorphic Functions of Several Variables ^{[Gunning Vol. II]}, with Grauert–Remmert giving the preparation-based proof ^{[Grauert–Remmert]} and Whitney developing the geometry of the branched covers that realise the finite extensions ^[Whitney]. The same correspondence holds for formal power series, the analytic case being distinguished only by the existence of genuine points on which functions vanish — the feature that gives the geometric side of the bijection its content.

Bibliography Master

@article{Ruckert1933,
  author  = {R{\"u}ckert, Walther},
  title   = {Zum Eliminationsproblem der Potenzreihenideale},
  journal = {Math. Ann.},
  volume  = {107},
  year    = {1933},
  pages   = {259--281}
}

@book{GunningSCVII,
  author    = {Gunning, Robert C.},
  title     = {Introduction to Holomorphic Functions of Several Variables,
               Volume II: Local Theory},
  publisher = {Wadsworth \& Brooks/Cole},
  year      = {1990}
}

@book{GunningRossi,
  author    = {Gunning, Robert C. and Rossi, Hugo},
  title     = {Analytic Functions of Several Complex Variables},
  publisher = {Prentice-Hall},
  year      = {1965}
}

@book{GrauertRemmertCAS,
  author    = {Grauert, Hans and Remmert, Reinhold},
  title     = {Coherent Analytic Sheaves},
  series    = {Grundlehren der mathematischen Wissenschaften},
  volume    = {265},
  publisher = {Springer},
  year      = {1984}
}

@book{WhitneyCAV,
  author    = {Whitney, Hassler},
  title     = {Complex Analytic Varieties},
  publisher = {Addison-Wesley},
  year      = {1972}
}

@book{HilbertNullstellensatz1893,
  author    = {Hilbert, David},
  title     = {{\"U}ber die vollen Invariantensysteme},
  journal   = {Math. Ann.},
  volume    = {42},
  year      = {1893},
  pages     = {313--373},
  note      = {Source of the algebraic Nullstellensatz}
}