Cotangent complex and deformation-obstruction theory

Intuition Beginner

A smooth surface carries a tangent plane at each point, and the assembly of these planes — the module of differentials — is an instrument that measures how functions on the surface change. This instrument suffices whenever the shape is smooth. Shapes with corners, crossings, or self-intersections defeat it: at a singular point the plain tangent picture breaks. The cotangent complex is an extended, two-storey instrument that keeps working at singular points, storing in its lower floor the ordinary differentials and in its upper floor a correction term that detects the singularity.

The extended instrument converts the vague question "can this shape be deformed?" into a clean count. The number of independent first-order bending directions is the deformation space $T^1$, and the upper floor supplies an obstruction space $T^2$ that records whether a tentative bend extends to a full family. When the obstruction space is zero, every bend extends and the moduli space of shapes is smooth.

The reward is uniformity. Smooth shapes, singular shapes, shapes over fields of positive characteristic, and families of maps between shapes all answer to the same counting rule: build the cotangent complex, read off $T^1$ and $T^2$. One machine organizes the entire local theory of moduli.

Visual Beginner

The diagram shows the cotangent complex as a two-storey instrument attached to a shape with one singular crossing. The lower floor carries the ordinary differentials ${\Omega }_X^1$, the same data a smooth shape already records. The upper floor carries the singularity-correction $H_1(L_X)$, which is zero for smooth shapes and nonzero exactly at the crossing. Two output arrows leave the instrument: one labelled $T^1$ counts the bending directions, one labelled $T^2$ measures the obstructions.

For a smooth shape the upper floor is empty and the instrument collapses to the familiar one-storey ruler of differentials. For a singular shape the upper floor is the part that the plain ruler missed, and it is exactly the part that decides which bends are realizable.

Worked example Beginner

Take the crossing shape $X$ defined by $xy=0$ in the plane — a node, two lines meeting at one point. At the crossing the plain differentials misbehave, so the cotangent complex $L_X$ acquires a correction term in its upper floor.

Step 1. Build the cotangent complex of the node. Its lower floor is the usual differentials; its upper floor is a one-dimensional correction concentrated at the crossing.

Step 2. Compute the count of bending directions $T^1$ from the whole instrument. The answer is the single number $1$.

Step 3. Interpret the number. The node has exactly one smoothing direction, realized by the family $xy=t$ as the parameter $t$ moves away from $0$: the two lines merge into one smooth curve.

What this tells us: the cotangent complex turns the question "can the crossing be smoothed, and in how many ways?" into the concrete count $1$. A rigid smooth shape of genus $0$, by contrast, has count $0$ — its upper floor is empty and there is nothing to bend.

Check your understanding Beginner

Formal definition Intermediate+

Let $k$ be an algebraically closed field and $X$ a scheme of finite type over $k$. Write $\mathbf{D}({\mathcal{O}}_X)$ for the derived category of complexes of ${\mathcal{O}}_X$-modules. The module of Kähler differentials ${\Omega }_{X/k}^1$ is the ${\mathcal{O}}_X$-module representing derivations: for every ${\mathcal{O}}_X$-module $M$, ${\mathrm{Der}}_k({\mathcal{O}}_X,M)\cong {\mathrm{Hom}}_{{\mathcal{O}}_X}({\Omega }_{X/k}^1,M)$ ^{[Hartshorne 2010 §5]}.

The cotangent complex (Illusie). The cotangent complex $L_{X/k}\in \mathbf{D}({\mathcal{O}}_X)$ is the left-derived functor of Kähler differentials: resolve $X\to \mathrm{Spec}k$ on the left by a simplicially free (or polynomial) algebra $P_{\bullet }\twoheadrightarrow {\mathcal{O}}_X$ and set $L_{X/k}:={\Omega }_{P_{\bullet }/k}^1{\otimes }_{P_{\bullet }}{\mathcal{O}}_X$, a complex concentrated in homological degrees $\ge 0$ ^{[Illusie 1971]}. Its zeroth homology sheaf is $H_0(L_{X/k})={\Omega }_{X/k}^1$; the higher homology sheaves $H_i(L_{X/k})$, $i\ge 1$, vanish when $X$ is smooth over $k$ and are the singularity-correction terms detected at singular points.

Universal property (André-Quillen-Illusie). The cotangent complex is characterized by its Ext groups ^{[Quillen 1970] [André 1974]}:

$${\mathrm{Ext}}_{{\mathcal{O}}_X}^i(L_{X/k},M)\cong D^i(X/k,M),i\ge 0,$$

where $D^0={\mathrm{Der}}_k({\mathcal{O}}_X,M)$ is the module of derivations, $D^1$ is the group of $k$-split square-zero extensions $0\to M\to \mathcal{A}\to {\mathcal{O}}_X\to 0$ up to isomorphism, and $D^2$ is the obstruction group governing the lift of such an extension across a small extension of the base. This is the defining identification: Ext of $L_X$ is the deformation theory of $X$.

Transitivity (Jacobi-Zariski) triangle. For a composite of morphisms $X\stackrel{f}{\to }Y\stackrel{g}{\to }Z$ there is a distinguished triangle in $\mathbf{D}({\mathcal{O}}_X)$ ^{[Illusie 1971]}:

$$L_{X/Y}{\otimes }_{f^{-1}{\mathcal{O}}_Y}^{\mathbf{L}}{\mathcal{O}}_X⟶L_{X/Z}⟶f^{\ast }{L}_{Y/Z}\stackrel{+1}{\to },$$

the transitivity triangle. For a closed immersion $X\hookrightarrow Y$ cut out by the ideal $\mathcal{I}$, the first leg specialises to the conormal sheaf: when the immersion is a regular embedding (locally complete intersection), $L_{X/Y}\simeq \mathcal{I}/{\mathcal{I}}^2[1]$.

Deformation and obstruction groups. The deformation-theoretic groups of $X$ are

$$T^i(X):={\mathrm{Ext}}_{{\mathcal{O}}_X}^i(L_{X/k},{\mathcal{O}}_X),i\ge 0.$$

Then $T^0$ is the space of infinitesimal automorphisms, $T^1$ is the tangent space to the deformation functor, and $T^2$ is the obstruction space ^{[Sernesi 2006 §4]}.

The deformation functor and Schlessinger conditions. Fix $X/k$. The deformation functor is ${\mathrm{Def}}_X:{\mathbf{Art}}_k\to \mathbf{Set}$ sending an Artin local $k$-algebra $A$ to the set of flat families $\mathcal{X}\to \mathrm{Spec}A$ with closed fibre ${\mathcal{X}}_0\cong X$, modulo isomorphism. A functor of Artin rings $F$ with $F(k)=\{\ast \}$ satisfies Schlessinger's conditions when ^{[Schlessinger 1968]}:

• (H1) for every diagram of small extensions $A^{\prime }\twoheadrightarrow A\twoheadleftarrow A^{\prime \prime }$, the natural map $F(A^{\prime }{\times }_A{A}^{\prime \prime })\to F(A^{\prime }){\times }_{F(A)}F(A^{\prime \prime })$ is bijective (glueing);
• (H2) the kernel of $F(A^{\prime })\to F(A)$ for a small extension with principal kernel is a transitive $F(k[\epsilon ]/{\epsilon }^2)$-set;
• (H3) the comparison $F(k[{\epsilon }_1,{\epsilon }_2]/({\epsilon }_1,{\epsilon }_2{)}^2)\to F(k[\epsilon ]/{\epsilon }^2)\times F(k[\epsilon ]/{\epsilon }^2)$ is injective.

A prorepresentable hull (semiversal deformation) for $F$ is a complete local $k$-algebra $R$ with a smooth natural transformation $\mathrm{Hom}(R,-)\to F$ inducing an isomorphism on tangent spaces; $F$ is prorepresentable when this transformation is an isomorphism. Schlessinger's theorem (proved in the sibling unit 04.19.01 pending) states that (H1) + (H2) give a hull and (H1) + (H2) + (H3) give prorepresentability.

$T^1$-lifting. A deformation functor ${\mathrm{Def}}_X$ has a tangent-obstruction theory with coefficients $(T^1,T^2)$ when, for every small extension $0\to k\to A^{\prime }\to A\to 0$ and every $\xi \in {\mathrm{Def}}_X(A)$, the fibre of ${\mathrm{Def}}_X(A^{\prime })\to {\mathrm{Def}}_X(A)$ over $\xi$ is either empty or a torsor under $T^1(X)$, with emptiness detected by a class $o(\xi )\in T^2(X)$.

Counterexamples to common slips Intermediate+

• $L_X$ is not ${\Omega }_X^1$ in general. Only the zeroth homology agrees: $H_0(L_X)={\Omega }_X^1$. The higher homology is the correction that singularities force. Using ${\Omega }_X^1$ in place of $L_X$ for a singular $X$ computes the wrong $T^i$.
• $T^2(X)=0$ does not mean $X$ is rigid. Rigidity is $T^1(X)=0$. The vanishing of $T^2$ says deformations are unobstructed, not that they are absent.
• Prorepresentable hull $\ne$ prorepresentable. A hull is a smooth cover; prorepresentability requires the additional rigidity (H3), which fails in the presence of infinitesimal automorphisms.

Key theorem with proof Intermediate+

Theorem (Illusie, 1971). Let $X$ be a scheme of finite type over $k$ with cotangent complex $L_X=L_{X/k}$.

(i) Tangent. There is a natural bijection ${\mathrm{Def}}_X(k[\epsilon ]/{\epsilon }^2)\cong {\mathrm{Ext}}_{{\mathcal{O}}_X}^1(L_X,{\mathcal{O}}_X)=T^1(X)$.

(ii) Obstruction. For every small extension $e:0\to k\to A^{\prime }\to A\to 0$ in ${\mathbf{Art}}_k$ and every deformation $\xi \in {\mathrm{Def}}_X(A)$, there is an obstruction class $o_e(\xi )\in {\mathrm{Ext}}_{{\mathcal{O}}_X}^2(L_X,{\mathcal{O}}_X)=T^2(X)$ whose vanishing is equivalent to $\xi$ lifting to ${\mathrm{Def}}_X(A^{\prime })$. When the obstruction vanishes, the set of lifts is a torsor under $T^1(X)$.

Proof of (i). Let $D=\mathrm{Spec}k[\epsilon ]/{\epsilon }^2$ be the scheme of dual numbers, with closed point $i:\mathrm{Spec}k\hookrightarrow D$. A first-order deformation of $X$ is a cartesian square

\xymatrix{ X \ar[r] \ar[d] & \mathcal{X} \ar[d]^{\pi} \\ \mathrm{Spec}\,k \ar[r]^-{i} & D }

with $\pi$ flat. On the underlying topological space $|X|=|\mathcal{X}|$, the structure sheaf ${\mathcal{O}}_{\mathcal{X}}$ sits in an exact sequence of sheaves of $k$-algebras

$$0⟶{\mathcal{O}}_X\cdot \epsilon ⟶{\mathcal{O}}_{\mathcal{X}}⟶{\mathcal{O}}_X⟶0,$$

and because ${\epsilon }^2=0$ the ideal ${\mathcal{O}}_X\cdot \epsilon$ squares to zero. Thus ${\mathcal{O}}_{\mathcal{X}}$ is a $k$-split square-zero extension of ${\mathcal{O}}_X$ by the ${\mathcal{O}}_X$-module $M:={\mathcal{O}}_X$ (the ${\mathcal{O}}_X$-structure on $M$ coming through the quotient ${\mathcal{O}}_{\mathcal{X}}\twoheadrightarrow {\mathcal{O}}_X$). Conversely, every $k$-split square-zero extension $0\to {\mathcal{O}}_X\to \mathcal{A}\to {\mathcal{O}}_X\to 0$ that is $k$-flat defines such a deformation over $D$ by setting ${\mathcal{O}}_{\mathcal{X}}=\mathcal{A}$; isomorphisms of deformations agree with isomorphisms of extensions. Hence

$${\mathrm{Def}}_X(D)\cong \{\text{k-split square-zero extensions of OX by OX}\}/\cong .$$

By the universal property of the cotangent complex, the group of such extensions is ${\mathrm{Ext}}_{{\mathcal{O}}_X}^1(L_X,{\mathcal{O}}_X)$. Combining the two bijections yields ${\mathrm{Def}}_X(k[\epsilon ]/{\epsilon }^2)\cong T^1(X)$, natural in $X$ because both identifications are. $\square$

Proof of (ii). Let $I_A=\ker (A\to k)$ and $I_{A^{\prime }}=\ker (A^{\prime }\to k)$, and write ${\mathcal{O}}_{X_A}$, ${\mathcal{O}}_{X_{A^{\prime }}}$ for the structure sheaves of representatives of $\xi$ and a hypothetical lift. The deformation $\xi$ over $A$ corresponds, by the same recognition as in (i), to a $k$-split square-zero extension class

$$[\xi ]\in {\mathrm{Ext}}_{{\mathcal{O}}_X}^1(L_X,I_A{\otimes }_k{\mathcal{O}}_X).$$

A lift of $\xi$ to $A^{\prime }$ is the data of a square-zero extension of ${\mathcal{O}}_X$ by $I_{A^{\prime }}{\otimes }_k{\mathcal{O}}_X$ mapping to $[\xi ]$ under the surjection $I_{A^{\prime }}\twoheadrightarrow I_A$. Apply the long exact Ext sequence of the short exact sequence of ${\mathcal{O}}_X$-modules $0\to {\mathcal{O}}_X\to I_{A^{\prime }}\otimes {\mathcal{O}}_X\to I_A\otimes {\mathcal{O}}_X\to 0$ (the small-extension kernel is $k$):

$${\mathrm{Ext}}^1(L_X,I_{A^{\prime }}\otimes {\mathcal{O}}_X)\stackrel{\alpha }{\to }{\mathrm{Ext}}^1(L_X,I_A\otimes {\mathcal{O}}_X)\stackrel{\delta }{\to }{\mathrm{Ext}}^2(L_X,{\mathcal{O}}_X).$$

Define $o_e(\xi ):=\delta ([\xi ])\in T^2(X)$. Exactness gives $o_e(\xi )=0$ if and only if $[\xi ]$ lies in the image of $\alpha$, that is, if and only if $\xi$ lifts to ${\mathrm{Def}}_X(A^{\prime })$. When $o_e(\xi )=0$ the set of lifts is the fibre of $\alpha$ over $[\xi ]$, a torsor under $\ker \alpha =\mathrm{coker}{\mathrm{Ext}}^0(L_X,I_{A^{\prime }}\otimes {\mathcal{O}}_X\to I_A\otimes {\mathcal{O}}_X)={\mathrm{Ext}}^1(L_X,{\mathcal{O}}_X)=T^1(X)$, the last identification using that $I_{A^{\prime }}\to I_A$ has one-dimensional $k$-kernel. $\square$

This theorem identifies $T^1$ and $T^2$ as the two Ext groups that the cotangent complex feeds into the deformation functor, and it shows that obstruction theory is the boundary map of the Ext long exact sequence — a single homological construction.

Bridge. The Ext identification builds toward 04.10.20, where the abstract $T^1={\mathrm{Ext}}^1(L_X,{\mathcal{O}}_X)$ specialises to $H^1(C,T_C)$ for a smooth curve, and it appears again in 04.02.05 as the fact that $L_X={\Omega }_X^1[0]$ exactly when $X/k$ is smooth — the infinitesimal lifting criterion for smoothness is the degree-zero shadow of this theorem. The foundational reason the deformation functor admits a tangent-obstruction theory at all is that flatness converts a small extension into a short exact sequence of sheaves, and the boundary map of its Ext sequence is the obstruction; this is exactly the mechanism that generalises from schemes to locally complete intersections, to morphisms (via $L_{X/Y}$), and to perfect obstruction theories on derived stacks. The central insight of Illusie is that a single object $L_X$ governs every level — automorphisms, deformations, obstructions — and the bridge is that these are the three Ext groups ${\mathrm{Ext}}^0$, ${\mathrm{Ext}}^1$, ${\mathrm{Ext}}^2$ of one complex.

Exercises Intermediate+

Lean formalization Intermediate+

lean_status: none — Mathlib has no named cotangent complex. The gap is recorded in the unit metadata Mathlib gap analysis: missing are $L_{X/S}$ as the left-derived functor of Kähler differentials, the transitivity distinguished triangle, the identification of square-zero extensions with ${\mathrm{Ext}}^1(L_X,M)$, the obstruction class in ${\mathrm{Ext}}^2(L_X,M)$, the groups $T^i={\mathrm{Ext}}^i(L_X,{\mathcal{O}}_X)$, the $T^1$-lifting criterion, and the Schlessinger conditions with the prorepresentable hull. A first formalization target is the smooth-case identity $L_{X/k}\simeq {\Omega }_{X/k}^1[0]$ together with the transitivity triangle. Human review by an algebraic-geometry expert is the correctness gate until that infrastructure lands in Mathlib.

Advanced results Master

Construction via simplicial resolutions. Illusie ^{[Illusie 1971]} and, independently, Quillen ^{[Quillen 1970]} and André ^{[André 1974]} construct $L_{X/S}$ by choosing a simplicial resolution $P_{\bullet }\twoheadrightarrow {\mathcal{O}}_X$ by free ${\mathcal{O}}_S$-algebras and setting $L_{X/S}={\Omega }_{P_{\bullet }/S}^1{\otimes }_{P_{\bullet }}{\mathcal{O}}_X$. The resulting object is independent of the resolution up to canonical isomorphism in $\mathbf{D}({\mathcal{O}}_X)$, concentrated in homological degrees $\ge 0$, and the André-Quillen cohomology $D^n(X/S,M)=H^n(\mathrm{Hom}(L_{X/S},M))$ recovers the classical André-Quillen groups. This is the sense in which $L_X$ is the derived functor of differentials.

The Hodge-cotangent and conjugate spectral sequences. The exterior powers of $L_X$ support the Hodge-cotangent spectral sequence $E_{p,q}^2=H_p({\Lambda }^q{L}_X)\Rightarrow H_{p+q}(L_X)$, and Illusie constructs the Hodge-to-de Rham spectral sequence $E_1^{p,q}=H^q(X,{\Omega }_X^p)\Rightarrow H_{\mathrm{dR}}^{p+q}(X/k)$ by filtering ${\tau }_{\ge 0}{L}_X$ ^{[Illusie 1971]}. For smooth $X$ this recovers the classical Hodge-de Rham sequence; for singular $X$ the higher homology of $L_X$ contributes the correction terms. Deligne-Illusie's positive-characteristic degeneracy result — a Kodaira-Akizuki-Nakano vanishing in characteristic $p>\dim X$ lifted to characteristic zero — is proved by analyzing this spectral sequence.

$T^1$-lifting and unobstructedness. The $T^1$-lifting criterion ^{[Fantechi-Göttsche-van Straten 1999]} states that a locally complete intersection scheme $X$ whose first-order deformations lift compatibly through every small extension has vanishing obstruction space $T^2$, so ${\mathrm{Def}}_X$ is unobstructed. The criterion applies to finite subschemes of smooth surfaces, proving the smoothness of the Hilbert scheme ${\mathrm{Hilb}}^{(n)}(S)$ of $n$ points, and to Hilbert schemes of points on a wide class of smooth threefolds with at worst ordinary double points.

Schlessinger conditions and prorepresentability. Schlessinger's functor-of-Artin-rings framework ^{[Schlessinger 1968]}, proved in the sibling unit 04.19.01 pending, packages the Ext groups into a representability statement: conditions (H1), (H2) yield a prorepresentable hull, and (H3) upgrades it to prorepresentability. The cotangent complex computes the tangent $T^1$ and obstruction $T^2$ of this hull directly, so the hull is a quotient of a formal power series ring in $\dim T^1$ variables by an ideal whose embedding dimension is governed by $T^2$.

Deformations of morphisms. For a morphism $f:X\to Y$, deformations of $f$ with $X$, $Y$ fixed are controlled by $L_{X/Y}$: the tangent space is ${\mathrm{Ext}}^0(L_{X/Y},f^{\ast }{\mathcal{O}}_Y)=\mathrm{Hom}({\Omega }_{X/Y}^1,f^{\ast }{\mathcal{O}}_Y)$ and the obstruction space is ${\mathrm{Ext}}^1(L_{X/Y},f^{\ast }{\mathcal{O}}_Y)$. The transitivity triangle reduces these to combinations of $L_X$, $L_Y$, and the pullback $f^{\ast }{L}_Y$, recovering the Kodaira-Spencer map when $X$ and $Y$ are smooth.

Derived deformation theory and dg-Lie algebras. Pridham and Lurie, building on Hinich and Quillen, recast deformation functors as controlled by differential-graded Lie (or $L_{\infty }$) algebras whose linear term is $\mathrm{R}\mathrm{Hom}(L_X,{\mathcal{O}}_X)$ ^{[Illusie 1971]}. The Maurer-Cartan equation of the dg-Lie algebra is the equation of the deformation functor; the bracket is the obstruction. This identifies the cotangent complex as the linearization of the deformation problem and is the entry point to derived algebraic geometry.

Shifted symplectic structures and perfect obstruction theories. In derived algebraic geometry, Pantev-Toën-Vaquié-Vezzosi ^{[PTVV 2013]} show that the shifted cotangent complex ${\mathbf{T}}_X^{\ast }[-1]$ of a derived stack carries a canonical $(-1)$-shifted symplectic structure when $X$ is derived locally of finite presentation. Behrend-Fantechi ^{[Behrend-Fantechi 1997]} extract from a perfect obstruction theory — a morphism $E\to L_X$ with $E$ of perfect amplitude $[-1,0]$ — the intrinsic normal cone and the virtual fundamental class. The obstruction sheaf $h^1(E^{\vee })$ is the degree-one truncation of $E^{\vee }$, and the virtual dimension is $\mathrm{rk}{h}^0(E)-\mathrm{rk}{h}^1(E)$.

Synthesis. The cotangent complex is the foundational reason that a single homological object governs every layer of deformation: ${\mathrm{Ext}}^0$ for automorphisms, ${\mathrm{Ext}}^1$ for deformations, ${\mathrm{Ext}}^2$ for obstructions. This is exactly the central insight of Illusie — that the left-derived functor of Kähler differentials packages the full tangent-obstruction calculus — and putting these together with the transitivity triangle identifies deformations of a scheme, of a morphism, and of a map into a fixed target as three instances of one Ext computation. The bridge is that the boundary map of the Ext long exact sequence of a small extension is the obstruction class, and this construction generalises from schemes to locally complete intersections, to derived stacks with perfect obstruction theories, and to shifted symplectic moduli. The pattern recurs: the cotangent complex is the linear term of the governing dg-Lie algebra, and the deformation problem is its Maurer-Cartan equation.

Full proof set Master

Proposition (smooth case). If $X$ is smooth over $k$, then $L_{X/k}\simeq {\Omega }_{X/k}^1[0]$ in $\mathbf{D}({\mathcal{O}}_X)$, concentrated in homological degree zero.

Proof. Smoothness of $X\to \mathrm{Spec}k$ means that locally on $X$ there is an étale morphism $U\to {\mathbb{A}}_k^d$ for some $d$. Étale morphisms are formally étale: they satisfy the infinitesimal lifting criterion with a unique lift. By the universal property, the cotangent complex of an étale morphism vanishes: $L_{U/{\mathbb{A}}_k^d}=0$, since derivations into any module lift uniquely and the defining ${\mathrm{Ext}}^0=\mathrm{Hom}(L_{U/{\mathbb{A}}_k^d},M)=0$ forces $L_{U/{\mathbb{A}}_k^d}=0$. The transitivity triangle for $U\to {\mathbb{A}}_k^d\to \mathrm{Spec}k$ then gives $L_{U/k}\cong L_{{\mathbb{A}}_k^d/k}{\otimes }^{\mathbf{L}}{\mathcal{O}}_U$. For affine space, $L_{{\mathbb{A}}_k^d/k}={\Omega }_{{\mathbb{A}}_k^d/k}^1[0]={⨁}_{i=1}^d{\mathcal{O}}_{{\mathbb{A}}^d}d{x}_i[0]$, since polynomial algebras are free and hence already resolved. Tensoring with ${\mathcal{O}}_U$ preserves degree-zero concentration, giving $L_{U/k}\cong {\Omega }_{U/k}^1[0]$. These local identifications glue over an étale cover of $X$ because the cotangent complex is local for the étale topology, yielding $L_{X/k}\simeq {\Omega }_{X/k}^1[0]$. $\square$

Lemma (transitivity and the conormal sequence). For a composite $X\to Y\to Z$ the transitivity triangle is functorial in the composite, and for a closed immersion $X\hookrightarrow Y$ with ideal $\mathcal{I}$ cut out by a regular sequence (regular embedding), $L_{X/Y}\simeq \mathcal{I}/{\mathcal{I}}^2[1]$.

Proof. Functoriality is immediate from the construction: a commutative diagram of morphisms induces a morphism of simplicial resolutions and hence a morphism of the resulting distinguished triangles, compatible with the rotation. For the regular-embedding claim, take a free resolution $P_{\bullet }\twoheadrightarrow {\mathcal{O}}_X$ adapted to the factorization $X\hookrightarrow Y$; when $X\hookrightarrow Y$ is a regular embedding of codimension $c$, locally cut out by a regular sequence $f_1,\dots ,f_c$, the module $\mathcal{I}/{\mathcal{I}}^2$ is free of rank $c$ on $X$, and the Koszul resolution provides a length-one resolution of ${\mathcal{O}}_X$ over ${\mathcal{O}}_Y$. Applying ${\Omega }_{(-/Y}^1$ gives ${\Omega }_{{\mathcal{O}}_X/Y}^1=0$ (the $f_i$ are killed in ${\mathcal{O}}_X$) and the first homology is $\mathcal{I}/{\mathcal{I}}^2$, so $L_{X/Y}$ is quasi-isomorphic to $\mathcal{I}/{\mathcal{I}}^2[1]$. $\square$

Proposition (obstruction via the connecting morphism). In the notation of the Key theorem, the obstruction class $o_e(\xi )=\delta ([\xi ])\in {\mathrm{Ext}}^2(L_X,{\mathcal{O}}_X)$ is the image of the deformation class $[\xi ]\in {\mathrm{Ext}}^1(L_X,I_A\otimes {\mathcal{O}}_X)$ under the connecting morphism of the long exact Ext sequence of $0\to {\mathcal{O}}_X\to I_{A^{\prime }}\otimes {\mathcal{O}}_X\to I_A\otimes {\mathcal{O}}_X\to 0$. The class $o_e(\xi )$ is independent of the representative of $\xi$.

Proof. Independence of representative: two representatives of $\xi$ differ by an automorphism of the square-zero extension, hence by an element of ${\mathrm{Ext}}^0(L_X,I_A\otimes {\mathcal{O}}_X)=\mathrm{Der}({\mathcal{O}}_X,I_A\otimes {\mathcal{O}}_X)$. The boundary map $\delta$ factors through the cokernel of ${\mathrm{Ext}}^0(L_X,I_{A^{\prime }}\otimes {\mathcal{O}}_X)\to {\mathrm{Ext}}^0(L_X,I_A\otimes {\mathcal{O}}_X)$, so a change of representative acts by an element in the image of ${\mathrm{Ext}}^1(L_X,{\mathcal{O}}_X)\to {\mathrm{Ext}}^1(L_X,I_A\otimes {\mathcal{O}}_X)$, which lies in $\ker \delta$ by exactness; thus $\delta ([\xi ])$ is unchanged. The independence makes $o_e(\xi )$ a well-defined obstruction, and its vanishing is equivalent to liftability by exactness of the Ext sequence, as in the theorem. $\square$

Connections Master

• Deformation theory — overview 04.19.01 pending. The sibling unit develops the $(T^1,T^2)$ slogan, the Hilbert-scheme tangent $H^0(X,N_{X/Y})$, and proves Schlessinger's representability theorem. The present unit specialises inward: it constructs the cotangent complex that computes $T^1$ and $T^2$ and recovers the sibling's Schlessinger framework as the functor-of-Artin-rings shadow of the Ext identifications.

• Homological algebra 01.06.01. The cotangent complex lives in the derived category $\mathbf{D}({\mathcal{O}}_X)$, and every deformation-theoretic statement here is an Ext computation. The transitivity triangle is a distinguished triangle, and the obstruction class is a connecting morphism — the homological algebra of derived categories, long exact sequences, and derived tensor products is the substrate on which the entire theory runs.

• Scheme 04.02.01. The cotangent complex is the derived functor of the Kähler differentials ${\Omega }_{X/k}^1$ attached to the structure morphism $X\to \mathrm{Spec}k$; the fibre products, nilpotent thickenings (dual numbers), and ideal sheaves of closed immersions are the scheme-theoretic input the construction consumes.

• Sheaf cohomology 04.03.01. For smooth $X$, $T^1(X)=H^1(X,T_X)$ and $T^2(X)=H^2(X,T_X)$, so the vanishing and finiteness theorems of sheaf cohomology directly govern smoothness and dimension of moduli. The cotangent complex extends these cohomological counts to singular schemes.

• Deformation theory of smooth curves 04.10.20. The abstract identification $T^1={\mathrm{Ext}}^1(L_X,{\mathcal{O}}_X)$ specialises there to $H^1(C,T_C)$ for a smooth curve, recovering Riemann's $3g-3$ moduli count; the smooth-case reduction $L_X={\Omega }_X^1[0]$ proved here is the bridge to that computation.

• Smooth, étale, and unramified morphisms 04.02.05. The identity $L_X={\Omega }_X^1[0]$ characterises smoothness, and its proof runs through the vanishing of $L$ for étale morphisms; the infinitesimal lifting criterion for smoothness is the degree-zero shadow of the cotangent-complex identification.

Historical & philosophical context Master

The cotangent complex was constructed independently by Daniel Quillen in On the (co)homology of commutative rings (1970) ^{[Quillen 1970]} and Michel André in Méthode simpliciale en algèbre homologique et algèbre commutative (1967) and Homologie des Algèbres Commutatives (1974) ^{[André 1974]}, as the derived functor of Kähler differentials within the then-new framework of simplicial homotopy theory for commutative rings. The André-Quillen (co)homology groups $D^n(A,M)={\mathrm{Ext}}^n(L_A,M)$ generalized the cotangent-space formalism of Grothendieck's EGA from smooth to arbitrary algebras.

Luc Illusie's Complexe Cotangent et Déformations I, II (Springer Lecture Notes 239, 283, 1971–1972) ^{[Illusie 1971]} extended the construction to morphisms of schemes, established the transitivity (Jacobi-Zariski) distinguished triangle, and proved the deformation-theoretic identifications ${\mathrm{Def}}_X(k[\epsilon ]/{\epsilon }^2)\cong {\mathrm{Ext}}^1(L_X,{\mathcal{O}}_X)$ and the obstruction theory in ${\mathrm{Ext}}^2(L_X,{\mathcal{O}}_X)$. Illusie's treatment unified the Kodaira-Spencer analytic deformation theory of complex manifolds with the Grothendieck-Schlessinger functorial algebraic theory and supplied the technical language — the cotangent complex, the transitivity triangle, the $T^i$ groups — in which all subsequent deformation theory is written.

Michael Schlessinger's Functors of Artin rings (Trans. AMS 130, 1968) ^{[Schlessinger 1968]} had already reframed deformation theory as a representability problem for functors on Artin local rings, with the (H1)–(H3) criterion and the prorepresentable hull; Illusie's cotangent complex supplied the functor of Artin rings ${\mathrm{Def}}_X$ with its computable tangent and obstruction spaces. The bridge to derived geometry, where the cotangent complex is the linearization of a governing dg-Lie or $L_{\infty }$ algebra and carries shifted symplectic structures on moduli, was built by Hinich, Pridham, Lurie, and Pantev-Toën-Vaquié-Vezzosi ^{[PTVV 2013]}; Behrend-Fantechi ^{[Behrend-Fantechi 1997]} extracted the virtual fundamental class from a perfect obstruction theory $E\to L_X$, opening the route to Gromov-Witten and Donaldson-Thomas theory.

Bibliography Master

1. Illusie, L. (1971, 1972). Complexe Cotangent et Déformations I, II. Lecture Notes in Mathematics 239, 283. Springer-Verlag.

2. Schlessinger, M. (1968). Functors of Artin rings. Transactions of the American Mathematical Society 130, 208–222.

3. Hartshorne, R. (2010). Deformation Theory. Graduate Texts in Mathematics 257. Springer-Verlag.

4. Sernesi, E. (2006). Deformations of Algebraic Schemes. Grundlehren der mathematischen Wissenschaften 334. Springer-Verlag.

5. Quillen, D. G. (1970). On the (co)homology of commutative rings. In Applications of Categorical Algebra, Proceedings of Symposia in Pure Mathematics 17, 65–87. American Mathematical Society.

6. André, M. (1974). Homologie des Algèbres Commutatives. Grundlehren der mathematischen Wissenschaften 206. Springer-Verlag.

7. Behrend, K. and Fantechi, B. (1997). The intrinsic normal cone. Inventiones Mathematicae 128, 45–88.

8. Pantev, T., Toën, B., Vaquié, M., and Vezzosi, V. (2013). Shifted symplectic structures. Publications Mathématiques de l'IHÉS 117, 271–328.

9. Fantechi, B., Göttsche, L., and van Straten, D. (1999). Euler number of the compactifications of the arrangement of lines and a $T^1$-lifting theorem. Mathematische Zeitschrift 230, 41–60.