← All writingMathematics blog

Quillen–Suslin 1: Serre’s reduction

Serre’s reduction of the Quillen–Suslin theorem to stably free modules, using finite free resolutions and prime filtrations.

The Quillen–Suslin theorem is easy to state: if AA is a principal ideal domain, then every finite projective module over A[x1,…,xn]A[x_1,\dots,x_n] is free. Equivalently, every vector bundle over AAn\mathbb A_A^n is trivial. Here and throughout, “finite” means finitely generated.

The proof is broken into two parts:

  1. Every finite projective module over A[x1,…,xn]A[x_1,\dots,x_n] is stably free.
  2. Every stably free module over A[x1,…,xn]A[x_1,\dots,x_n] is free.

As will become clear, the method of proving each part will be quite different: the first consists mostly of manipulating exact sequences and resolutions, while the second deals with matrices of elements of rings. This post gives Serre’s reduction, the first part of the proof; we will complete the argument in Quillen–Suslin 2: Vaserstein’s proof.

Stably free modules and resolutions

In this section, all modules without qualification will be over some fixed commutative ring AA.

Definition 1. A module MM is stably free if there is a finite free module FF such that M⊕FM\oplus F is finite free. Two modules M,M′M,M' are stably isomorphic if there are finite free modules F,F′F,F' such that

M⊕F≅M′⊕F′.M\oplus F\cong M'\oplus F'.

In this case we write M≃stM′M\simeq_{\st}M'.

When MM is stably free it is finite projective, as a summand of a finite free module. It is also clear that if MM is stably isomorphic to a stably free module, then MM is stably free. Indeed, if M⊕F≅M′⊕F′M\oplus F\cong M'\oplus F' and M′⊕GM'\oplus G is finite free, then M⊕(F⊕G)M\oplus(F\oplus G) is finite free as well.

Throughout this post, we will generally reserve the letter FF for finite free modules and EE for stably free modules. Much of the theory of stably free modules is about resolutions.

Definition 2. A finite free resolution of a module MM is a finite exact sequence

0⟶Fn⟶Fn−1⟶⋯⟶F0⟶M⟶00\longrightarrow F_n\longrightarrow F_{n-1}\longrightarrow\cdots \longrightarrow F_0\longrightarrow M\longrightarrow0

with each FiF_i finite free. A finite stably free resolution is defined in the same way, with each term stably free.

The finite free dimension dim⁡ff(M)\dimff(M), respectively the finite stably free dimension dim⁡sf(M)\dimsf(M), is the minimal length nn of such a resolution, and is ∞\infty if no such resolution exists. Thus dimension zero means that MM itself is finite free, respectively stably free.

We start with a comparison of presentations.

Lemma 3 (Schanuel’s lemma). Suppose that for i=1,2i=1,2 the sequences

0⟶Ki⟶Pi⟶M⟶00\longrightarrow K_i\longrightarrow P_i\longrightarrow M\longrightarrow0

are exact, with PiP_i projective. Then

K1⊕P2≅K2⊕P1.K_1\oplus P_2\cong K_2\oplus P_1.

Proof. Consider the pullback P1×MP2P_1\times_M P_2. Projection onto either factor gives exact sequences

0⟶K1⟶P1×MP2⟶P2⟶0,0⟶K2⟶P1×MP2⟶P1⟶0.\begin{aligned} 0&\longrightarrow K_1\longrightarrow P_1\times_M P_2\longrightarrow P_2\longrightarrow0,\\ 0&\longrightarrow K_2\longrightarrow P_1\times_M P_2\longrightarrow P_1\longrightarrow0. \end{aligned}

Both split by projectivity, so the pullback is isomorphic to both K1⊕P2K_1\oplus P_2 and K2⊕P1K_2\oplus P_1. □

Proposition 4. Let M1,M2M_1,M_2 be modules such that M1≃stM2M_1\simeq_{\st}M_2, and suppose that

0⟶Ni⟶Ei⟶Mi⟶0,i=1,2,0\longrightarrow N_i\longrightarrow E_i\longrightarrow M_i\longrightarrow0, \qquad i=1,2,

are exact with EiE_i stably free. Then N1≃stN2N_1\simeq_{\st}N_2.

Proof. Choose finite free F1,F2F_1,F_2 and an isomorphism M1⊕F1≅M2⊕F2M_1\oplus F_1\cong M_2\oplus F_2. Adding the identity of FiF_i to the iith presentation gives two projective presentations of the same module. Schanuel’s lemma gives

N1⊕E2⊕F2≅N2⊕E1⊕F1.N_1\oplus E_2\oplus F_2\cong N_2\oplus E_1\oplus F_1.

Choose finite free GiG_i such that Ei⊕GiE_i\oplus G_i is finite free. Adding G1⊕G2G_1\oplus G_2 to both sides makes the summands accompanying N1N_1 and N2N_2 finite free. This is the required stable isomorphism. □

Lemma 5. Let n≥1n\geq1 and suppose dim⁡sf(M)≤n\dimsf(M)\leq n. If

En−1⟶⋯⟶E0⟶M⟶0E_{n-1}\longrightarrow\cdots\longrightarrow E_0\longrightarrow M\longrightarrow0

is exact with each EiE_i stably free, then

K=ker⁡(En−1⟶En−2)K=\ker(E_{n-1}\longrightarrow E_{n-2})

is stably free, where E−1=ME_{-1}=M when n=1n=1.

Proof. It suffices to show that KK is stably isomorphic to a stably free module. By assumption there is a resolution

0⟶En′⟶⋯⟶E0′⟶M⟶00\longrightarrow E_n'\longrightarrow\cdots\longrightarrow E_0'\longrightarrow M\longrightarrow0

of length nn, extending by zeros if necessary. Write E−1=E−1′=ME_{-1}=E_{-1}'=M, and set

Ki=ker⁡(Ei⟶Ei−1),Ki′=ker⁡(Ei′⟶Ei−1′).K_i=\ker(E_i\longrightarrow E_{i-1}), \qquad K_i'=\ker(E_i'\longrightarrow E_{i-1}').

In particular, Kn−1=KK_{n-1}=K and Kn−1′≅En′K_{n-1}'\cong E_n'.

For i=0i=0, Proposition 4 gives K0≃stK0′K_0\simeq_{\st}K_0'. For i≥1i\geq1 we have exact sequences

0⟶Ki⟶Ei⟶Ki−1⟶00\longrightarrow K_i\longrightarrow E_i\longrightarrow K_{i-1}\longrightarrow0

and similarly for the primed modules. The same proposition now gives Ki≃stKi′K_i\simeq_{\st}K_i' inductively. In particular K≃stEn′K\simeq_{\st}E_n', as desired. □

Corollary 6. Assume AA is Noetherian, and suppose

0⟶N⟶E⟶M⟶00\longrightarrow N\longrightarrow E\longrightarrow M\longrightarrow0

is exact with EE stably free. If dim⁡sf(M)=n<∞\dimsf(M)=n<\infty, then

dim⁡sf(N)≤max⁡{n−1,0}.\dimsf(N)\leq\max\{n-1,0\}.

Proof. If n=0n=0, then MM is stably free, hence projective, so E≅N⊕ME\cong N\oplus M. Adding finite free complements for EE and MM shows that NN is stably free.

For n≥1n\geq1, the module NN and the successive kernels of finite free presentations are finite by Noetherianity. Choose enough such presentations to extend E→ME\to M to a partial stably free resolution with terms in degrees 0,…,n−10,\dots,n-1. Its last kernel is stably free by Lemma 5. Removing E→ME\to M gives a stably free resolution of NN of length at most n−1n-1. □

Resolvable modules

Definition 7. A module is resolvable if it admits a finite free resolution, i.e. if dim⁡ff(M)<∞\dimff(M)<\infty. A ring AA is pointwise resolvable if every prime ideal p⊂A\mathfrak p\subset A is resolvable as an AA-module.

The distinction between prime ideals and their quotient modules is harmless here: the sequence

0⟶p⟶A⟶A/p⟶00\longrightarrow\mathfrak p\longrightarrow A\longrightarrow A/\mathfrak p\longrightarrow0

and the next proposition will show that one is resolvable if and only if the other is.

First, a small observation lets us use stably free terms whenever convenient. A finite stably free resolution can always be changed into a finite free resolution. Choose a finite free G0G_0 such that E0⊕G0E_0\oplus G_0 is finite free, and add the exact complex G0→idG0G_0\xrightarrow{\id}G_0 in degrees 11 and 00. Now the term in degree 00 is free and the term in degree 11 is still stably free. Repeat in degrees 22 and 11, then 33 and 22, and so on. After the last step, all terms are finite free, with at most one extra term at the left.

Proposition 8 (The 2-of-3 property). Assume AA is Noetherian. If

0⟶M′⟶M⟶M′′⟶00\longrightarrow M'\longrightarrow M\longrightarrow M''\longrightarrow0

is exact and any two of the modules are resolvable, then so is the third.

Proof. We give the three cases explicitly.

Suppose first that M′M' and M′′M'' have finite free resolutions F∙′F_\bullet' and F∙′′F_\bullet''. Lift F0′′→M′′F_0''\to M'' through M→M′′M\to M'', using freeness. Together with F0′→M′→MF_0'\to M'\to M, this gives a surjection

F0′⊕F0′′⟶M.F_0'\oplus F_0''\longrightarrow M.

Its kernel fits into a short exact sequence whose outer terms are the kernels of F0′→M′F_0'\to M' and F0′′→M′′F_0''\to M''. Indeed, projecting a pair in the middle kernel to F0′′F_0'' gives a map onto the second kernel; a preimage can be corrected using F0′→M′F_0'\to M', and its kernel is the first kernel. Repeating this construction gives a resolution of MM with terms Fi′⊕Fi′′F_i'\oplus F_i''. It is finite because the original resolutions are finite. This is the horseshoe construction.

For the other cases, recall the following explicit construction. Given a homomorphism u:L→Nu:L\to N and finite free resolutions F∙→LF_\bullet\to L and G∙→NG_\bullet\to N, projectivity lets us lift uu to maps fi:Fi→Gif_i:F_i\to G_i commuting with the differentials. At degree 00 this is the lifting property for G0→NG_0\to N. At each later degree, the map already constructed lands in the appropriate kernel, which is the image of the next differential, so the same lifting property applies.

The mapping cone has terms

Ci=Gi⊕Fi−1(i≥1),C0=G0,C_i=G_i\oplus F_{i-1}\quad(i\geq1), \qquad C_0=G_0,

and differential

dC(g,v)=(dGg+fi−1(v),−dFv),d_C(g,v)=\bigl(d_Gg+f_{i-1}(v),-d_Fv\bigr),

with the second component omitted when i=1i=1. Its homology is

H1(C∙)≅ker⁡u,H0(C∙)≅coker⁡u,Hi(C∙)=0(i≥2).H_1(C_\bullet)\cong\ker u, \qquad H_0(C_\bullet)\cong\coker u, \qquad H_i(C_\bullet)=0\quad(i\geq2).

For completeness, these identities follow from the degreewise split exact sequence of complexes

0⟶G∙⟶C∙⟶F∙[1]⟶0,0\longrightarrow G_\bullet\longrightarrow C_\bullet \longrightarrow F_\bullet[1]\longrightarrow0,

where F∙[1]F_\bullet[1] denotes the complex with Fi−1F_{i-1} in degree ii and differential −dF-d_F. The connecting map in homology is precisely uu, and the resolutions have no positive homology. Equivalently, the same identities follow by substituting the displayed formula for dCd_C and chasing kernels and images.

If M′M' and MM are resolvable, apply this construction to the injection u:M′→Mu:M'\to M. The cone has no positive homology and has H0=M′′H_0=M'', so it is a finite free resolution of M′′M''.

Finally, suppose MM and M′′M'' are resolvable, and apply the construction to u:M→M′′u:M\to M''. This time H0(C∙)=0H_0(C_\bullet)=0, so C1→C0C_1\to C_0 is surjective. Since C0C_0 is finite free, it splits. Thus

K=ker⁡(C1⟶C0)satisfiesK⊕C0≅C1,K=\ker(C_1\longrightarrow C_0) \quad\text{satisfies}\quad K\oplus C_0\cong C_1,

so KK is stably free. The homology calculation gives an exact sequence

⋯⟶C3⟶C2⟶K⟶M′⟶0.\cdots\longrightarrow C_3\longrightarrow C_2 \longrightarrow K\longrightarrow M'\longrightarrow0.

This is a finite stably free resolution of M′M', and the preceding observation converts it to a finite free resolution. □

In fact, the proof of the 2-of-3 property did not need Noetherianity. It will be crucial in what follows, however, because it gives finite filtrations. This is to be expected: our results concern finite resolutions, and Serre’s problem asks about sheaves which are locally free of finite rank.

Prime filtrations

To prove the following lemma, we will need associated primes. For an AA-module MM, write

Ann⁡A(M)={a∈A:ax=0 for all x∈M}.\Ann_A(M)=\{a\in A:ax=0\text{ for all }x\in M\}.

An associated prime of MM is a prime ideal p=Ann⁡A(x)\mathfrak p=\Ann_A(x) for some nonzero x∈Mx\in M. Equivalently, MM has a submodule isomorphic to A/pA/\mathfrak p.

If AA is Noetherian and M≠0M\neq0, then MM always has an associated prime. Indeed, choose a maximal element of

{Ann⁡A(x):0≠x∈M}\{\Ann_A(x):0\neq x\in M\}

under inclusion, say I=Ann⁡A(x)I=\Ann_A(x). This ideal is proper. If ab∈Iab\in I and b∉Ib\notin I, then bx≠0bx\neq0 and I⊆Ann⁡A(bx)I\subseteq\Ann_A(bx). Maximality gives equality, so a∈Ia\in I. Thus II is prime.

Lemma 9 (Prime filtration). Let AA be Noetherian and MM a finite AA-module. Then there are prime ideals pi\mathfrak p_i and a finite filtration

M=Mn⊃Mn−1⊃⋯⊃M1⊃M0=0M=M_n\supset M_{n-1}\supset\cdots\supset M_1\supset M_0=0

such that Mi/Mi−1≅A/piM_i/M_{i-1}\cong A/\mathfrak p_i for each ii.

Proof. By Noetherianity, MM satisfies the ascending chain condition on submodules. Start with M0=0M_0=0. If Mi≠MM_i\neq M, choose an associated prime pi+1\mathfrak p_{i+1} of M/MiM/M_i. The corresponding submodule of M/MiM/M_i has an inverse image Mi+1⊆MM_{i+1}\subseteq M satisfying

Mi+1/Mi≅A/pi+1.M_{i+1}/M_i\cong A/\mathfrak p_{i+1}.

In particular Mi⊊Mi+1M_i\subsetneq M_{i+1}. This process must stop by the ascending chain condition, and it can only stop when Mn=MM_n=M. □

Adjoining one variable

Our next goal is the following theorem.

Theorem 10. Suppose that AA is Noetherian and pointwise resolvable. Then A[x]A[x] is pointwise resolvable.

From this Serre’s reduction will follow relatively easily, but its proof requires more than the exact sequence manipulation we have used so far. We first prove an intermediate step. For an ideal p⊆A\mathfrak p\subseteq A, write p[x]=pA[x]\mathfrak p[x]=\mathfrak pA[x].

Lemma 11. Let AA be Noetherian, let P⊂A[x]P\subset A[x] be prime, and put p=P∩A\mathfrak p=P\cap A. Suppose P≠p[x]P\neq\mathfrak p[x]. Then there is a finite A[x]A[x]-module NN with the following properties:

  1. There is an exact sequence
0⟶A[x]/p[x]⟶P/p[x]⟶N⟶0.0\longrightarrow A[x]/\mathfrak p[x] \longrightarrow P/\mathfrak p[x]\longrightarrow N\longrightarrow0.
  1. The module NN admits a finite filtration
N=Nm⊃Nm−1⊃⋯⊃N0=0N=N_m\supset N_{m-1}\supset\cdots\supset N_0=0

such that Ni/Ni−1≅A[x]/QiN_i/N_{i-1}\cong A[x]/Q_i, where each QiQ_i is prime and

p⊊Qi∩A.\mathfrak p\subsetneq Q_i\cap A.

Proof. Let A0=A/pA_0=A/\mathfrak p, a Noetherian domain, and let kk be its fraction field. Since p[x]⊆P\mathfrak p[x]\subseteq P, the quotient

P0=P/p[x]P_0=P/\mathfrak p[x]

is a nonzero ideal of A0[x]≅A[x]/p[x]A_0[x]\cong A[x]/\mathfrak p[x]. The ring A0[x]A_0[x] is Noetherian, so P0P_0 is generated by finitely many polynomials g1,…,grg_1,\dots,g_r.

Choose a nonzero f∈P0f\in P_0 of minimal degree. Division in k[x]k[x] gives

gi=qif+ri,deg⁡ri<deg⁡fg_i=q_i f+r_i, \qquad \deg r_i<\deg f

for each nonzero remainder. Let d0∈A0∖{0}d_0\in A_0\setminus\{0\} be a product of denominators such that d0qi,d0ri∈A0[x]d_0q_i,d_0r_i\in A_0[x] for all ii. Then

d0ri=d0gi−(d0qi)f∈P0.d_0r_i=d_0g_i-(d_0q_i)f\in P_0.

If this polynomial were nonzero, it would have degree less than ff, contradicting minimality. Hence ri=0r_i=0 for every ii, and

d0P0⊆(f).d_0P_0\subseteq(f).

Set N=P0/(f)N=P_0/(f), considered as an A[x]A[x]-module through the quotient map A[x]→A0[x]A[x]\to A_0[x]. There is an exact sequence

0⟶(f)⟶P0⟶N⟶0.0\longrightarrow(f)\longrightarrow P_0\longrightarrow N\longrightarrow0.

Multiplication by ff identifies A0[x]A_0[x] with (f)(f): it is onto by definition and injective because A0[x]A_0[x] is a domain and f≠0f\neq0. This gives the first required sequence.

Choose a lift d∈Ad\in A of d0d_0, so d∉pd\notin\mathfrak p. The inclusion d0P0⊆(f)d_0P_0\subseteq(f) shows that dN=0dN=0, and pN=0\mathfrak pN=0 by the definition of the A[x]A[x]-module structure. Thus

(p,d)A[x]⊆Ann⁡A[x](N).(\mathfrak p,d)A[x]\subseteq\Ann_{A[x]}(N).

The module NN is finite, so Lemma 9 gives a filtration with quotients A[x]/QiA[x]/Q_i. Every element annihilating NN also annihilates these quotients, hence

(p,d)A[x]⊆Qi.(\mathfrak p,d)A[x]\subseteq Q_i.

Therefore p⊆Qi∩A\mathfrak p\subseteq Q_i\cap A, and the inclusion is strict since d∈Qi∩Ad\in Q_i\cap A but d∉pd\notin\mathfrak p. □

We can now prove Theorem 10.

Proof. First, each extended prime p[x]\mathfrak p[x] is resolvable as an A[x]A[x]-module. Indeed, pointwise resolvability of AA gives a finite free resolution of p\mathfrak p. The extension A→A[x]A\to A[x] is flat, and

A⊕r⊗AA[x]≅A[x]⊕r,p⊗AA[x]≅p[x].A^{\oplus r}\otimes_A A[x]\cong A[x]^{\oplus r}, \qquad \mathfrak p\otimes_A A[x]\cong\mathfrak p[x].

Tensoring the resolution with A[x]A[x] therefore gives the required finite free resolution of p[x]\mathfrak p[x]. The 2-of-3 property also shows that A[x]/p[x]A[x]/\mathfrak p[x] is resolvable.

Suppose now that some prime of A[x]A[x] is not resolvable. By Noetherianity of AA, the set

{P∩A:P∈Spec⁡(A[x]), P is not resolvable}\{P\cap A:P\in\Spec(A[x]),\ P\text{ is not resolvable}\}

has a maximal element. Choose such a PP and write p=P∩A\mathfrak p=P\cap A. The preceding paragraph implies P≠p[x]P\neq\mathfrak p[x], so we may apply Lemma 11.

In its filtration of NN, each prime QiQ_i has contraction strictly larger than p\mathfrak p. By maximality, QiQ_i is resolvable. The exact sequence

0⟶Qi⟶A[x]⟶A[x]/Qi⟶00\longrightarrow Q_i\longrightarrow A[x]\longrightarrow A[x]/Q_i\longrightarrow0

then shows that A[x]/QiA[x]/Q_i is resolvable. Starting from N0=0N_0=0 and applying the 2-of-3 property to

0⟶Ni−1⟶Ni⟶A[x]/Qi⟶00\longrightarrow N_{i-1}\longrightarrow N_i\longrightarrow A[x]/Q_i\longrightarrow0

shows inductively that N=NmN=N_m is resolvable.

We already know that A[x]/p[x]A[x]/\mathfrak p[x] is resolvable, so the first sequence of Lemma 11 shows that P/p[x]P/\mathfrak p[x] is resolvable. Finally,

0⟶p[x]⟶P⟶P/p[x]⟶00\longrightarrow\mathfrak p[x]\longrightarrow P\longrightarrow P/\mathfrak p[x]\longrightarrow0

shows that PP is resolvable, a contradiction. □

Corollary 12 (Serre). Suppose that AA is Noetherian and pointwise resolvable. Then every finite A[x]A[x]-module is resolvable.

Proof. By Theorem 10, all prime ideals of A[x]A[x] are resolvable. If MM is a finite A[x]A[x]-module, take a prime filtration

M=Mr⊃⋯⊃M0=0,Mi/Mi−1≅A[x]/Pi.M=M_r\supset\cdots\supset M_0=0, \qquad M_i/M_{i-1}\cong A[x]/P_i.

The sequence 0→Pi→A[x]→A[x]/Pi→00\to P_i\to A[x]\to A[x]/P_i\to0 shows that each quotient is resolvable. The sequences

0⟶Mi−1⟶Mi⟶A[x]/Pi⟶00\longrightarrow M_{i-1}\longrightarrow M_i\longrightarrow A[x]/P_i\longrightarrow0

then show by induction that every MiM_i, and in particular MM, is resolvable. □

Serre’s reduction

It remains to connect resolutions back to stably free modules.

Lemma 13. Let MM be a projective module. Then MM is stably free if and only if it is resolvable.

Proof. If M⊕F≅GM\oplus F\cong G with F,GF,G finite free, then

0⟶F⟶G⟶M⟶00\longrightarrow F\longrightarrow G\longrightarrow M\longrightarrow0

is a finite free resolution.

Conversely, suppose MM has a finite free resolution, and induct on its length. In length zero there is nothing to prove. Otherwise write its last part as

0⟶K⟶F0⟶M⟶0.0\longrightarrow K\longrightarrow F_0\longrightarrow M\longrightarrow0.

Since MM is projective, this sequence splits, so F0≅K⊕MF_0\cong K\oplus M and KK is projective. The remaining terms form a shorter finite free resolution of KK, so KK is stably free by induction. Choose finite free F,GF,G with K⊕F≅GK\oplus F\cong G. Then

M⊕G≅M⊕K⊕F≅F0⊕FM\oplus G\cong M\oplus K\oplus F\cong F_0\oplus F

is finite free. Thus MM is stably free. □

Theorem 14 (Serre’s reduction). Let AA be a principal ideal domain. Then every finite projective module over A[x1,…,xn]A[x_1,\dots,x_n] is stably free.

Proof. A PID is Noetherian and every ideal is finite free: the zero ideal is free of rank zero, and a nonzero principal ideal is isomorphic to AA because AA is a domain. In particular AA is pointwise resolvable.

Repeatedly applying Theorem 10 shows that A[x1,…,xn−1]A[x_1,\dots,x_{n-1}] is Noetherian and pointwise resolvable. Hence, for n≥1n\geq1, Corollary 12 shows that every finite A[x1,…,xn]A[x_1,\dots,x_n]-module is resolvable. Its finite projective modules are therefore stably free by Lemma 13.

For n=0n=0, every finite projective AA-module is already free. To recall the elementary reason, any submodule M⊆ArM\subseteq A^r is finite free: project onto the last factor, whose image is a principal ideal, and use induction on rr on the kernel. The resulting short exact sequence splits because the image is free. A finite projective module is a summand, hence a submodule, of some ArA^r. □

Notice that Serre’s reduction works more generally over any Noetherian pointwise resolvable ring: its finite projective modules, and those over its polynomial rings, are stably free. The argument applies in particular when every prime ideal is stably free, since stably free modules are resolvable.

A natural question is where in this argument the ring AA may be replaced by a scheme XX and the modules by quasicoherent sheaves. The biggest issue is that finite freeness, which we would like to replace with triviality, is not local. In addition, vector bundles need not be projective objects of QCoh⁡(X)\operatorname{QCoh}(X). For these reasons, the best setting for this sequence of results is the affine case.

We have now reduced the Quillen–Suslin theorem to a cancellation problem: if adding a finite free summand makes a module free, must it already be free? We answer this for polynomial rings over a PID in Quillen–Suslin 2: Vaserstein’s proof. For the original exposition and its geometric context, see Serre Schemes and the Quillen–Suslin Theorem.

Back to the blog →