Syzygies — Hilbert's syzygy theorem, projective dimension, and Auslander-Buchsbaum

Intuition Beginner

A module is hard to describe directly. The first move is to list its generators: a free module that maps onto it. But generators satisfy relations — equations the generators obey. Those relations form a new module, and to describe that one you list its generators, which obey relations among relations, and so on. This cascade is a free resolution.

Hilbert's syzygy theorem is the remarkable statement that over the polynomial ring in $n$ variables, this cascade always stops: after at most $n$ layers of relations, the next module is free. The number of variables is a hard ceiling on how tangled the relations can become. The first syzygy is the module of relations among generators; the second is relations among those; Hilbert says the chain terminates.

A regular sequence is a list of equations each of which genuinely cuts the module down — each is a non-zero-divisor on what is left. The depth counts how long such a chain can run. The Auslander-Buchsbaum formula is a conservation law: the length of the shortest resolution plus the depth always equals the depth of the ring itself.

Together these three results — the syzygy theorem, projective dimension, and the Auslander-Buchsbaum formula — explain why polynomial rings are homologically tame and why regular local rings behave like smooth points of a space.

Visual Beginner

A resolution reads left to right as a staircase of free modules, each kernel feeding the next map; Hilbert's theorem caps the number of stairs, and the Auslander-Buchsbaum formula balances two quantities against the fixed depth of the ring.

Layer	Name	What it captures
$F_0\to M$	generators	a free cover of $M$
$F_1\to F_0$	first syzygy	relations among the generators
$F_2\to F_1$	second syzygy	relations among the relations
$F_n\to F_{n-1}$	$n$-th syzygy	Hilbert: stops by layer $n$ over $k[x_1,\dots ,x_n]$

The balance line beneath the staircase is the conservation law: $\mathrm{pd}(M)+\mathrm{depth}(M)=\mathrm{depth}(R)$. Each extra layer of resolution trades for exactly one unit of depth.

Worked example Beginner

Take $R=k[x,y]$, polynomials in two variables, and the module $M=k$, the simplest module over $R$. It is generated by the single element $1$, so a free module $F_0=R$ maps onto $k$ by sending each polynomial to its constant term.

Step 1. The kernel is the ideal of polynomials with zero constant term, which is $(x,y)$. This ideal needs two generators, $x$ and $y$, so the first syzygy layer is $F_1=R\oplus R$ mapping onto $F_0$ by the row $(xy)$: a pair $(a,b)$ lands on $ax+by$.

Step 2. The single relation between $x$ and $y$ is that $y\cdot x-x\cdot y=0$. Written as a pair this is $(-y,x)$, so $F_2=R$ maps into $F_1$ by the column $\left(\genfrac{}{}{0ex}{}{-y}{x}\right)$.

Step 3. Check the composition: multiply the row $(xy)$ by the column $\left(\genfrac{}{}{0ex}{}{-y}{x}\right)$. The result is $x\cdot (-y)+y\cdot x=-xy+xy=0$. The composition vanishes, as required, and there are no further relations, so the resolution ends at $F_2$.

What this tells us: the resolution $0\to R\to R^2\to R\to k\to 0$ has length $2$, exactly the number of variables. Hilbert's syzygy theorem guarantees this is the worst case over $k[x,y]$.

Check your understanding Beginner

Formal definition Intermediate+

Let $R$ be a ring and $M$ an $R$-module. A free resolution of $M$ is an exact sequence

$$\cdots ⟶F_2\stackrel{d_2}{\to }{F}_1\stackrel{d_1}{\to }{F}_0\stackrel{\epsilon }{\to }M⟶0$$

with each $F_i$ a free $R$-module ^{[Eisenbud Ch. 17]}. The $i$-th syzygy module is ${\mathrm{Syz}}_i^R(M)=\ker d_{i-1}$ (with $d_{-1}=\epsilon$ and ${\mathrm{Syz}}_0(M)=M$). Thus ${\mathrm{Syz}}_1(M)=\ker \epsilon$ is the module of relations among the chosen generators of $M$, and the resolution builds ${\mathrm{Syz}}_{i+1}$ as a free cover of ${\mathrm{Syz}}_i$. Syzygies depend on the chosen cover, but Schanuel's lemma controls the ambiguity: given two short exact sequences $0\to K\to P\to M\to 0$ and $0\to K^{\prime }\to P^{\prime }\to M\to 0$ with $P,P^{\prime }$ projective, there is an isomorphism $K\oplus P^{\prime }\cong K^{\prime }\oplus P$. Iterating, syzygies are well-defined up to projective summands.

The projective dimension of $M$ is

$${\mathrm{pd}}_R(M)=\inf \{n\ge 0:M\text{ admits a projective resolution of length }n\},$$

with ${\mathrm{pd}}_R(M)=\infty$ if no such resolution exists ^{[Weibel Ch. 4]}. Equivalently ${\mathrm{pd}}_R(M)=\sup \{n:{\mathrm{Ext}}_R^n(M,N)\ne 0\text{ for some }N\}$. Consequently ${\mathrm{pd}}_R(M)=0$ iff $M$ is projective, and ${\mathrm{pd}}_R(M)\le n$ iff ${\mathrm{Ext}}_R^{n+1}(M,N)=0$ for every module $N$. The global dimension of $R$ is $\mathrm{gldim}R={\sup }_M{\mathrm{pd}}_R(M)$.

Let $I\subseteq R$ be an ideal and $M$ a finitely generated $R$-module with $IM\ne M$. A sequence $x_1,\dots ,x_r\in I$ is $M$-regular if $x_i$ is a non-zero-divisor on $M/(x_1,\dots ,x_{i-1})M$ for each $i$, and $M/(x_1,\dots ,x_r)M\ne 0$. The $I$-depth of $M$ is the length of the longest $M$-regular sequence in $I$:

$${\mathrm{depth}}_I(M)=\sup \{r:\text{there exists an M-regular sequence of length }r\text{ in }I\}.$$

For a Noetherian local ring $(R,\mathfrak{m},k)$, write $\mathrm{depth}(M)={\mathrm{depth}}_{\mathfrak{m}}(M)$. Two homological characterisations make depth computable without exhibiting a regular sequence ^{[Eisenbud Ch. 17-18]}:

$${\mathrm{depth}}_I(M)=\inf \{n\ge 0:{\mathrm{Ext}}_R^n(R/I,M)\ne 0\},$$

and, for local $(R,\mathfrak{m},k)$, $\mathrm{depth}(M)=\inf \{n:{\mathrm{Ext}}_R^n(k,M)\ne 0\}$, equivalently ${\mathrm{depth}}_I(M)=n-\sup \{i:H_i(x_1,\dots ,x_n;M)\ne 0\}$ where $x_1,\dots ,x_n$ generate $I$ and $H_{\bullet }$ is Koszul homology. The inequality $\mathrm{depth}(M)\le \dim M$ always holds; when equality obtains, $M$ is Cohen-Macaulay.

Counterexamples to common slips

• Syzygies depend on the cover. Over $R=k[x,y]$, the residue field $k$ has the minimal free cover $R\twoheadrightarrow k$, whose first syzygy is the ideal $(x,y)$; using the non-minimal cover $R\oplus R\twoheadrightarrow k$ instead produces a first syzygy with an extra free summand. Schanuel's lemma records exactly this freedom.
• Order matters for regular sequences. In $R=k[x,y]$, the sequence $x,y$ is $R$-regular, but $xy,x$ is not: modulo $xy$, the element $x$ is killed by $y\ne 0$, so $x$ is a zero-divisor on $R/(xy)$. The condition "non-zero-divisor on the successive quotient" is stronger than "each element nonzero".
• Depth can be strictly smaller than dimension. In $R=k[[x,y]]/(xy)$, the maximal ideal consists of zero-divisors (since $x\cdot y=0$), so $\mathrm{depth}(R)=0$ while $\dim (R)=1$. The ring is not Cohen-Macaulay.

Key theorem with proof Intermediate+

Theorem (Auslander-Buchsbaum formula). Let $(R,\mathfrak{m},k)$ be a Noetherian local ring and $M$ a finitely generated $R$-module with ${\mathrm{pd}}_R(M)<\infty$. Then

\operatorname{pd}_R(M) \;+\; \operatorname{depth}(M) \;=\; \operatorname{depth}(R). \qquad\text{[ref: TODO_REF Auslander-Buchsbaum 1957]}

The proof rests on the depth lemma, the homological bookkeeping for a short exact sequence. For any short exact sequence $0\to M^{\prime }\to M\to M^{\prime \prime }\to 0$ of finite $R$-modules,

$$\mathrm{depth}(M^{\prime })\ge \min (\mathrm{depth}(M),\mathrm{depth}(M^{\prime \prime })+1),\mathrm{depth}(M^{\prime \prime })\ge \min (\mathrm{depth}(M),\mathrm{depth}(M^{\prime })-1),$$

which follows from the long exact sequence of ${\mathrm{Ext}}_R^{\bullet }(k,-)$ together with the identification $\mathrm{depth}(N)=\inf \{i:{\mathrm{Ext}}_R^i(k,N)\ne 0\}$ ^{[Eisenbud Prop. 18.4]}.

Proof. Induct on $d={\mathrm{pd}}_R(M)$.

Base case $d=0$. Then $M$ is projective, hence free over the local ring $R$ (a finite projective over a local ring is free). For a free module $F\cong R^{\oplus n}$, ${\mathrm{Ext}}_R^i(k,F)\cong {\mathrm{Ext}}_R^i(k,R{)}^{\oplus n}$, so $\mathrm{depth}(F)=\mathrm{depth}(R)$. The identity reads $0+\mathrm{depth}(R)=\mathrm{depth}(R)$.

Inductive step $d\ge 1$. Choose a finite free cover $0\to \Omega M\stackrel{\alpha }{\to }F\stackrel{\beta }{\to }M\to 0$. Splicing this short exact sequence with a projective resolution of $\Omega M$ of length $d-1$ gives one of $M$ of length $d$, so ${\mathrm{pd}}_R(\Omega M)=d-1$. The inductive hypothesis gives

$$(d-1)+\mathrm{depth}(\Omega M)=\mathrm{depth}(R),\text{\hspace{1em}(⋆)}$$

hence $\mathrm{depth}(\Omega M)=\mathrm{depth}(R)-(d-1)\le \mathrm{depth}(R)$.

Apply the depth lemma to $0\to \Omega M\to F\to M\to 0$, using $\mathrm{depth}(F)=\mathrm{depth}(R)$:

$$\mathrm{depth}(\Omega M)\ge \min (\mathrm{depth}(R),\mathrm{depth}(M)+1),\text{\hspace{1em}(1)}$$ $$\mathrm{depth}(M)\ge \min (\mathrm{depth}(R),\mathrm{depth}(\Omega M)-1).\text{\hspace{1em}(2)}$$

First observe $\mathrm{depth}(M)<\mathrm{depth}(R)$. Suppose instead $\mathrm{depth}(M)\ge \mathrm{depth}(R)$. Then (2) forces $\mathrm{depth}(\Omega M)-1\ge \mathrm{depth}(R)$, i.e. $\mathrm{depth}(\Omega M)>\mathrm{depth}(R)$, contradicting $(\star )$. Hence $\mathrm{depth}(M)<\mathrm{depth}(R)$.

With $\mathrm{depth}(M)<\mathrm{depth}(R)$, inequality (1) reads $\mathrm{depth}(\Omega M)\ge \mathrm{depth}(M)+1$, and inequality (2) becomes $\mathrm{depth}(M)\ge \mathrm{depth}(\Omega M)-1$ (since $\mathrm{depth}(\Omega M)-1<\mathrm{depth}(R)$). Together these give $\mathrm{depth}(\Omega M)=\mathrm{depth}(M)+1$. Substituting into $(\star )$,

$$(d-1)+(\mathrm{depth}(M)+1)=\mathrm{depth}(R),$$

that is, ${\mathrm{pd}}_R(M)+\mathrm{depth}(M)=\mathrm{depth}(R)$. $\square$

Bridge. This conservation law builds toward 04.03.06, where the $\mathrm{Ext}$ groups measuring depth here are the same derived functor that classifies extensions of coherent sheaves, and appears again in 04.02.07 as the homological engine behind dimension theory and regularity on schemes. The foundational reason the formula holds is that each syzygy layer of a minimal resolution over a local ring peels off exactly one unit of depth; this is exactly the equality $\mathrm{depth}(\Omega M)=\mathrm{depth}(M)+1$ extracted from the depth lemma, and the bridge is that projective dimension and depth are two complementary measurements of a single module, trading off against the fixed depth of the ambient ring.

Exercises Intermediate+

Lean formalization Intermediate+

lean_status: none is recorded for this unit. Mathlib carries the raw ingredients — Module.Projective and the in-progress ProjectiveDimension API, the Koszul complex on a sequence of ring elements, IsRegularSequence, and growing coverage of Noetherian local rings and Krull dimension — but the theorem-level wiring that this unit depends on is not yet assembled. Specifically absent is the pipeline that binds a minimal free resolution over a local ring, depth via $\mathrm{Ext}$, regular sequences, the Auslander-Buchsbaum conservation law, and the specialisation to $k[x_1,\dots ,x_n]$ yielding Hilbert's syzygy bound, together with the Auslander-Buchsbaum-Serre equivalence of regularity and finite global dimension. the Mathlib gap analysis records this as a coordinated formalization target across the commutative-algebra and homological-algebra strands; until it lands, no Lean module is declared and the Master-tier proofs below stand on human review.

Advanced results Master

Hilbert's syzygy theorem. Let $S=k[x_1,\dots ,x_n]$. Every finitely generated $S$-module $M$ admits a free resolution of length at most $n$; equivalently $\mathrm{gldim}S=n$ ^{[Hilbert 1890]}. The sharp example is the residue field $k=S/(x_1,\dots ,x_n)$: the Koszul complex $K_{\bullet }(x_1,\dots ,x_n;S)$ is a free resolution of $k$ of length $n$, so ${\mathrm{pd}}_S(k)=n$. The universal bound ${\mathrm{pd}}_S(M)\le n$ for all $M$ now drops out of the Auslander-Buchsbaum formula applied to the regular local ring $S_{\mathfrak{m}}$ at $\mathfrak{m}=(x_1,\dots ,x_n)$: there $\mathrm{depth}(S_{\mathfrak{m}})=n$, so ${\mathrm{pd}}_{S_{\mathfrak{m}}}(M_{\mathfrak{m}})=n-\mathrm{depth}(M_{\mathfrak{m}})\le n$, and localization cannot decrease what is already bounded. The graded version, closer to Hilbert's original statement, produces a graded free resolution of length $\le n$ with explicit Betti numbers read off from the Hilbert function of $M$.

The Auslander-Buchsbaum-Serre theorem. A Noetherian local ring $(R,\mathfrak{m},k)$ is regular when $\mathfrak{m}$ can be generated by $\dim R$ elements, i.e. the embedding dimension $\mu (\mathfrak{m})$ equals the Krull dimension. The Auslander-Buchsbaum-Serre theorem equates four conditions ^{[Serre 1955]}:

• $R$ is regular;
• $\mathrm{gldim}(R)<\infty$;
• ${\mathrm{pd}}_R(k)<\infty$;
• ${\mathrm{pd}}_R(k)=\dim R$.

When these hold, $\mathrm{gldim}(R)=\dim R$. The direction "regular $\Rightarrow$ ${\mathrm{pd}}_R(k)<\infty$" is the Koszul resolution: if $\mathfrak{m}$ is generated by a regular sequence $x_1,\dots ,x_d$, the Koszul complex resolves $k$ in length $d$. The reverse direction uses that a minimal free resolution of $k$ has $\mathrm{rank}{F}_i={\dim }_k{\mathrm{Tor}}_i^R(k,k)$, and the inequality ${\dim }_k{\mathrm{Tor}}_1^R(k,k)=\mu (\mathfrak{m})\ge \dim R$ together with the principal ideal theorem forces $\mu (\mathfrak{m})=\dim R$ when ${\mathrm{pd}}_R(k)$ is finite. The theorem makes regularity — geometrically, the smoothness of a point — into a purely homological property, which is the entry point to regular morphisms in scheme theory.

Minimal free resolutions. Over a local ring $(R,\mathfrak{m},k)$, a free resolution $F_{\bullet }\to M$ is minimal when each differential $d_i:F_i\to F_{i-1}$ has image inside $\mathfrak{m}{F}_{i-1}$, equivalently $d_i\otimes k=0$. A minimal resolution is unique up to isomorphism of complexes, and its ranks satisfy the Betti-number formula ${\beta }_i(M)=\mathrm{rank}{F}_i={\dim }_k{\mathrm{Tor}}_i^R(k,M)$. The Auslander-Buchsbaum formula then reads off the length of the minimal resolution as $\mathrm{pd}(M)=\mathrm{depth}(R)-\mathrm{depth}(M)$, and the resolution terminates exactly at the projective dimension. The Poincaré series $P_k^R(t)={\sum }_i{\dim }_k{\mathrm{Tor}}_i^R(k,k)t^i$ is a fundamental invariant: for a regular local ring it is the polynomial $(1+t{)}^{\dim R}$, and deviations from this polynomial measure how far $R$ sits from regularity.

Syzygy theorem as a flatness and duality tool. Over a regular local ring, the boundedness of projective dimension makes every finitely generated module a perfect complex, so derived duals $\mathbf{R}{\mathrm{Hom}}_R(M,-)$ commute with derived tensor in the derived category. The Auslander-Buchsbaum formula then specialises to the codimension formula of Exercise 8 for Cohen-Macaulay modules, which is the homological input to intersection multiplicities: Serre's intersection multiplicity $\chi (M,N)={\sum }_i(-1{)}^i\mathrm{length}{\mathrm{Tor}}_i^R(M,N)$ is well-defined precisely because projective dimension is finite over a regular ring.

Synthesis. Hilbert's syzygy theorem, the projective-dimension formalism, and the Auslander-Buchsbaum formula build toward 04.03.06, where $\mathrm{Ext}$ computes both depth and sheaf extension groups, and the regular-local-ring characterisation appears again in 04.02.07 as the homological definition of a smooth point on a scheme; this is exactly the conservation law that ties resolution length to depth, the central insight is that a minimal resolution peels off one unit of depth per syzygy, the regular-sequence viewpoint generalises to the Koszul complex and complete intersections, the projective-versus-flat dimension story is dual to itself across the derived category, and the bridge is that regularity, finite global dimension, and finite projective dimension of the residue field are three faces of one condition; putting these together, polynomial rings and regular local rings are precisely the rings on which homological algebra terminates in bounded length.

Full proof set Master

Proposition (Koszul complex resolves the residue field). Let $S=k[x_1,\dots ,x_n]$ and $\mathfrak{m}=(x_1,\dots ,x_n)$. The Koszul complex $K_{\bullet }(x_1,\dots ,x_n;S)$ is a free resolution of $k=S/\mathfrak{m}$ of length $n$.

Proof. The Koszul complex has $K_i={\bigwedge }^i{S}^n$ (free of rank $\left(\genfrac{}{}{0ex}{}{n}{i}\right)$) in degree $i$, with differential determined by exterior contraction against the vector $(x_1,\dots ,x_n)$:

$$d(e_{j_1}\wedge \cdots \wedge e_{j_i})=\sum _{\ell =1}^i(-1{)}^{\ell -1}{x}_{j_{\ell }}{e}_{j_1}\wedge \cdots \wedge \widehat{e_{j_{\ell }}}\wedge \cdots \wedge e_{j_i}.$$

It is a chain complex of free $S$-modules of length $n$, with $H_0(K)=S/\mathfrak{m}=k$. The key fact is that $x_1,\dots ,x_n$ is an $S$-regular sequence, which forces acyclicity in positive degrees: $H_i(K)=0$ for $i\ge 1$. This is proved by induction on $n$. For $n=1$, $K_{\bullet }$ is $0\to S\stackrel{x_1}{\to }S\to 0$, acyclic in positive degree because $x_1$ is a non-zero-divisor. For the inductive step, identify $K_{\bullet }(x_1,\dots ,x_n)$ as the mapping cone of the chain map "multiplication by $x_n$" on $K_{\bullet }(x_1,\dots ,x_{n-1})$; the associated long exact homology sequence, together with the fact that $x_n$ is a non-zero-divisor on the homology of $K_{\bullet }(x_1,\dots ,x_{n-1})$ (which is $k$ concentrated in degree $0$ by induction), yields $H_i(K_{\bullet }(x_1,\dots ,x_n))=0$ for $i\ge 1$. Hence $K_{\bullet }\twoheadrightarrow k$ is a free resolution of length $n$, and ${\mathrm{pd}}_S(k)=n$. $\square$

Proposition (Auslander-Buchsbaum-Serre, regularity $\Rightarrow$ finite global dimension). If $(R,\mathfrak{m},k)$ is a regular local ring, then ${\mathrm{pd}}_R(k)=\dim R$, and consequently $\mathrm{gldim}(R)=\dim R$.

Proof sketch. If $R$ is regular, $\mathfrak{m}$ is generated by a regular sequence $x_1,\dots ,x_d$ with $d=\dim R$. Localising the Koszul complex of the previous proposition gives a free resolution of $k$ over $R$ of length $d$, so ${\mathrm{pd}}_R(k)\le d$. Minimality shows this resolution is minimal, so ${\mathrm{Tor}}_d^R(k,k)\cong k\ne 0$ and ${\mathrm{pd}}_R(k)\ge d$; hence equality. For any finitely generated $R$-module $M$, the Auslander-Buchsbaum formula gives ${\mathrm{pd}}_R(M)=d-\mathrm{depth}(M)\le d$, so $\mathrm{gldim}(R)\le d$, while ${\mathrm{pd}}_R(k)=d$ forces $\mathrm{gldim}(R)\ge d$. Thus $\mathrm{gldim}(R)=d$. The converse (finite ${\mathrm{pd}}_R(k)$ implies regularity) requires the inequality $\mu (\mathfrak{m})={\dim }_k{\mathrm{Tor}}_1^R(k,k)\le {\mathrm{pd}}_R(k)$ against Krull's principal ideal theorem; see ^{[Serre 1955]}. $\square$

Connections Master

• Homological algebra foundations 01.06.01. Syzygies, free resolutions, and projective dimension are built directly on the chain complexes, exact sequences, and derived functors established in 01.06.01; the projective dimension here is the length of the projective resolutions introduced there, and $\mathrm{Ext}$ — the right derived functor of $\mathrm{Hom}$ from that unit — is the device that computes both projective dimension and depth.

• Rings and modules 01.03.01. Free and projective resolutions live inside the module category of a ring from 01.03.01; syzygies are submodules of free modules, and Schanuel's lemma is a statement about the ambiguity of short exact sequences of the kind classified there.

• Commutative algebra foundations 01.05.01. The Auslander-Buchsbaum formula operates over Noetherian local rings, the localisation and primary-decomposition machinery of 01.05.01; the commutative algebra of Noetherian rings is the soil in which depth, regular sequences, and regular local rings grow.

• Derived functors and Ext in geometry 04.03.06. The $\mathrm{Ext}$ used here to measure depth and projective dimension is specialised to coherent sheaves in 04.03.06, where the same derived functor classifies sheaf extensions and underpins local-to-global spectral sequences on schemes.

• Dimension theory on schemes 04.02.07. The Auslander-Buchsbaum-Serre characterisation of regular local rings is the homological definition of a smooth point; in 04.02.07 it becomes the algebraic-geometric criterion for a scheme to be regular, connecting projective dimension to Krull dimension and the Nullstellensatz.

Historical & philosophical context Master

Hilbert proved the syzygy theorem in 1890 in Über die Theorie der algebraischen Formen, as the closing step of his reform of invariant theory; the word "syzygy," an astronomical term for a conjunction of bodies, had been borrowed into algebra by Sylvester to denote a relation among relations ^{[Hilbert 1890]}. Hilbert's theorem showed that the "relations among relations among relations" over a polynomial ring must terminate in at most $n$ steps, a finiteness result that settled the generation of invariants and foreshadowed the modern theory of free resolutions. The homological reformulation had to await Cartan and Eilenberg's Homological Algebra (1956), which reframed syzygies as kernels in a long exact sequence.

The depth notion crystallised in the early 1950s through work of Rees, Samuel, and Auslander-Buchsbaum. The Auslander-Buchsbaum formula was announced in 1957 ^{[Auslander-Buchsbaum 1957]}, and the full equivalence between regularity of a local ring and finite global dimension is due independently to Auslander-Buchsbaum and Serre, the latter in his 1955 Sur la dimension homologique des anneaux et des modules noethériens ^{[Serre 1955]}. Serre's paper introduced the Koszul complex — originally a tool of Koszul (1950) in the cohomology of Lie algebras — into commutative algebra, and proved that a local ring is regular exactly when its residue field has finite projective dimension. This made smoothness of a point a homological property, the entry point to Grothendieck's definition of a regular morphism in scheme theory. The Auslander-Buchsbaum-Serre theorem, the Koszul complex, and the Auslander-Buchsbaum formula together converted the finiteness Hilbert had observed in 1890 into the structural backbone of modern commutative algebra.

Bibliography Master

@book{Eisenbud1995,
  author = {Eisenbud, David},
  title = {Commutative Algebra with a View Toward Algebraic Geometry},
  publisher = {Springer GTM 150},
  year = {1995},
}

@book{Matsumura1989,
  author = {Matsumura, Hideyuki},
  title = {Commutative Ring Theory},
  publisher = {Cambridge University Press},
  series = {Cambridge Studies in Advanced Mathematics},
  volume = {8},
  year = {1989},
}

@book{Weibel1994,
  author = {Weibel, Charles A.},
  title = {An Introduction to Homological Algebra},
  publisher = {Cambridge University Press},
  year = {1994},
}

@book{BrunsHerzog1998,
  author = {Bruns, Winfried and Herzog, Jurgen},
  title = {Cohen-Macaulay Rings},
  publisher = {Cambridge University Press},
  edition = {revised},
  year = {1998},
}

@article{AuslanderBuchsbaum1957,
  author = {Auslander, Maurice and Buchsbaum, David A.},
  title = {Homological dimension in local rings},
  journal = {Proceedings of the National Academy of Sciences USA},
  volume = {42},
  pages = {364--367},
  year = {1957},
}

@article{Serre1955,
  author = {Serre, Jean-Pierre},
  title = {Sur la dimension homologique des anneaux et des modules noeth\'eriens},
  journal = {Nagoya Mathematical Journal},
  volume = {7},
  pages = {1--75},
  year = {1955},
}

@article{Hilbert1890,
  author = {Hilbert, David},
  title = {Ueber die Theorie der algebraischen Formen},
  journal = {Mathematische Annalen},
  volume = {36},
  pages = {473--534},
  year = {1890},
}

@article{Koszul1950,
  author = {Koszul, Jean-Louis},
  title = {Homologie et cohomologie des alg\`ebres de Lie},
  journal = {Bulletin de la Soci\'et\'e Math\'ematique de France},
  volume = {78},
  pages = {65--127},
  year = {1950},
}