Journal of Pure and Applied Algebra 216 (2012) 2159–2169

Contents lists available at SciVerse ScienceDirect

Journal of Pure and Applied Algebra

journal homepage: www.elsevier.com/locate/jpaa

Excision in algebraic obstruction theory

Satya Mandal a,∗ , Yong Yang b
a
Department of Mathematics, University of Kansas, 1460 Jayhawk Blvd., Lawrence, KS 66045, USA
b
Department of Mathematics and Systems Science, National University of Defense Technology, Changsha, Hunan, 410073, PR China

article info abstract

Article history: In this paper the relative algebraic obstruction groups (also known as Euler class groups)
Received 20 June 2011 were defined and some excision exact sequences were established. In particular, for a
Received in revised form 20 November regular domain A, essentially of finite type over an infinite field k, and a rank one projective
2011
A-module L0 , it was proved that
Available online 24 February 2012
Communicated by R. Parimala E n (A[T ], L0 ⊗ A[T ]) ≈ E n (A, L0 ) whenever 2n ≥ dim A + 4.

MSC: 13c10 © 2012 Elsevier B.V. All rights reserved.

1. Introduction

In topology, there is a well established obstruction theory for vector bundles (see [11]). A germ for a parallel obstructions
theory for projective modules over noetherian commutative rings was given by M.V. Nori around 1990 (see [6,8]). Originally,
this program was focused on the top rank case. For projective modules P of rank d, over noetherian commutative rings A with
dim A = d, an obstruction class e(P ) was defined. Bhatwadekar and Sridharan [3] proved that e(P ) = 0 if and only if P has a
free direct summand. A theory of obstructions for all projective modules, as complete as that of topological vector bundles,
is possible. In fact, a parallel K -theoretic approach was initiated by Barge and Morel [1] in 2000. This was given a more
complete shape by Fasel [5]. However, this approach does not seem very descriptive and two approaches must reconcile. In
this paper, we are concerned with the approach of Nori.
Following [4], obstruction groups E n (A, L) were defined in [9], for all integers n ≥ 1 and rank one projective A-modules
L. There has been only a limited success in defining obstruction classes e(P ) ∈ E n (A, L) for projective A-modules P with
rank(P ) = n < dim A and det(P ) ≈ L, as would be desired. When det P ≈ L ≈ A, we say that P is oriented and the situation
is referred to as the oriented case. Otherwise, it is referred to as the non-oriented case. Recently, in the oriented case, Yang
[12] defined relative obstructions groups E n (A, I, A), with respect to ideals I of A, when n ≥ 1. When 2n ≥ dim A + 3, he
established some exact sequences of these groups. The purpose of this paper is to extend the results of Yang [12] to the non-
oriented case. As in [12], first we define pull-back homomorphisms f ∗ : E n (A, L) → E n (B, B ⊗ L) of the obstructions groups,
corresponding to certain ring homomorphisms f : A → B and some integers n. Then, we define the relative obstruction
groups, E n (A, I, L) (see 4.1), and establish some exact sequences in Theorems 4.2 and 4.3. In particular, we establish an
excision exact sequence as follows.
Theorem 1.1. Let A be noetherian commutative ring with dim(A) = d and I be an ideal of A. Write A0 = IA . Assume that the
quotient homomorphism q : A  A0 has a splitting β : A0 → A such that for each locally n-generated ideal I0 of A0 , of height
n, we have height (β(I0 )A) ≥ n. Suppose L0 is a projective A0 -module of rank one and L = L0 ⊗β A. Then, for integers n with
2n ≥ d + 3, we have a split exact sequence as follows:

0 / E n (A, I, L) / E n (A, L) / E n ( A , L0 ) / 0.
I

∗ Corresponding author.
E-mail addresses: mandal@math.ku.edu (S. Mandal), yangyong@nudt.edu.cn (Y. Yang).

0022-4049/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.jpaa.2012.02.008
2160 S. Mandal, Y. Yang / Journal of Pure and Applied Algebra 216 (2012) 2159–2169

As an application, we prove that if A is a regular domain, essentially of finite type over an infinite perfect field k, and L0
is a projective A-module of rank one, then

E n (A[T ], L0 ⊗ A[T ]) ≈ E n (A, L0 ) whenever 2n ≥ dim A + 4.

The authors would like to thank the referee for careful reviewing and many valuable suggestions. The exposition of the paper
improved due to these suggestions. The Remark 3.6 due to the referee is particularly appreciated.

2. Preliminaries

For the definition of the obstructions groups the readers are referred to [9]. We recall some notations from [9].

Notations 2.1. Throughout this paper, A will denote a commutative noetherian ring with dim A = d and L will denote a
projective A-module of rank one.

(1) For integers n ≥ 1, write F = Fn = L ⊕ An−1 .

(2) A local L-orientation (of codimension n) is an equivalence class of surjective homomorphisms ω : F /IF  I /I 2 , where I
is an ideal of height n. When it is clear from the context, we just call them orientations.
(3) The free abelian group generated by the local orientations (I , ω), where the ideal I is connected, is denoted by Gn (A, L).
Gn (A,L)
(4) The obstruction group of codimension n is defined by E n (A, L) := R
, where R is the subgroup generated by the
global orientations.
(5) A local orientation ω : F /IF  I /I 2 , defines an element (I , ω) ∈ Gn (A, L). We will take the liberty to use the same
notation (I , ω) to denote its image in E n (A, L).

The following lemma is proved by using standard basic element theory along with generalized dimensions (see [7]).

Lemma 2.2. Suppose A is a noetherian commutative ring with dim A = d. Suppose I , J are two ideals of A with height (I ) = n
and J ⊆ I 2 . Let F be a projective A-module of rank n and ω : F  I /J be a surjective homomorphism. Also suppose I1 , . . . , Ir are
finitely many ideals of A. Then
 there is a surjective lift f : F  I ∩ K , such that (1) J + K = A, (2) height (K ) ≥ n and (3) for
1 ≤ i ≤ r, ≥ n.
K +I i
height Ii

Proof. Similar to [9, Lemma 4.3]. 

Lemma 2.3. Suppose A is a noetherian commutative ring with dim A = d. Suppose I , J are two ideals of A and F is a projective
A-module of rank n. Let ω : F  I /I 2 and ϕ : F  J be two surjective homomorphisms such that
I +J

A A F I
ω⊗ =ϕ⊗ :  .
I + J I + J (I + J )F I2 + IJ
Then, there is a surjective homomorphism Ω : F  I
I2∩ J
that lifts both ω and ϕ.

Proof. Consider the fiber product diagrams:

F / F
(I ∩ J )F
C IF EE
C EE
Ω
C EEω
EE
C! ! EE"
"
A / A I / I
I∩ J I J ∩ I2 I2

and
   
A / A F / F
J I +J JF CC (I + J )F
EE
CC EE
CC EE
ϕ CC EE
C! !  " 
J +I / I
.
J I2 + IJ

By the properties of fiber product diagrams, the desired homomorphism Ω is defined in the later diagram. This completes
the proof. 

Following is the non-oriented version of the theorem of Bhatwadekar and Sridharan [4, Theorem 4.2].
 S. Mandal, Y. Yang / Journal of Pure and Applied Algebra 216 (2012) 2159–2169 2161

Theorem 2.4. Let A be a noetherian commutative ring with dim A = d and 2n ≥ d + 3. Let L be a projective A-module of rank
one and Fn = An−1 ⊕ L. Let J be an ideal of height n and ω : JFFn  J /J 2 be a local L-orientation. Assume
n

(J , ω) = 0 ∈ E n (A, L).
Then there is a surjective lift θ : Fn  J of ω.

Proof. Similar to that of [4, Theorem 4.2]. 

2.1. Double of a ring

In this subsection, we define the Double of a ring A along an ideal I and summarize some of the facts about it. Let
A be a noetherian commutative ring with dim A = d and I be an ideal of A. The double of A along the ideal I, is
defined as

D = D(A, I) = {(x, y) ∈ A × A : x − y ∈ I}.

This will be considered as a subring of A × A. The projection to the first and second coordinates from D will be denoted by
p1 and p2 .
Similarly, for an A-module M , the double of M along the ideal I, is defined as

D(M , I) = {(m, n) ∈ M × M : m − n ∈ IM }.

The following are some facts:

(1) The following diagrams

D(A, I)
p1
/A D(M , I)
p1
/M

p2 q and p2

   
A / A
M / M
q I IM

are fiber product diagrams, where q denotes the quotient homomorphism.

(2) The kernel(p1 ) = 0 × I and kernel(p2 ) = I × 0.
(3) In fact, the diagonal homomorphism ∆ : A → D splits both p1 , p2 .
(4) We have D = ∆(A)+ 0 × I = ∆(A)+ I × 0. So, D = ∆(A)+ ∆(A)(0, xi ), where I = Axi . So, D is finitely generated
 
A-module.
(5) So, D is noetherian and integral over A. Therefore, dim A = dim D.
(6) For any ideal I of A, we have

1 (I ) = {(x, x + z ) : x ∈ I , z ∈ I} = ∆(I ) + 0 × I
p− 1

and

2 (I ) = {(x + z , x) : x ∈ I , z ∈ I} = ∆(I ) + I × 0.
p− 1

(7) If ℘ ∈ Spec (A), then P = ∆(℘) + 0 × I = p−

1 (℘) ∈ Spec (D). Similarly, P = ∆(℘) + I × 0 = p2 (℘) ∈ Spec (D).
1 −1

Lemma 2.5. Let A, I, L, M , D, ∆ be as above. Then,

(1) There is a natural surjective homomorphism τ : D(A, I) ⊗ M → D(M , I), where D is considered as an A-algebra via the
diagonal ∆.
(2) If Q is projective, then τ : D(A, J ) ⊗ Q ≈ D(Q , J ).

Proof. It is easy to see that τ (m ⊗ (x, x + z )) = (xm, (x + z )m) is a well defined homomorphism from D ⊗ M → D(M , J ).
This establishes (1).
If Q = An is free, then it is obvious that

τ : D(A, J ) ⊗ An ≈ D(An , J ).
In the general case, note that there is a split exact sequence:

0 /Q g
/F f
/P /0
2162 S. Mandal, Y. Yang / Journal of Pure and Applied Algebra 216 (2012) 2159–2169

where F is free. Correspondingly, we have the following commutative diagram:

0 / D(A, J ) ⊗ Q / D(A, J ) ⊗ F / D(A, J ) ⊗ P /0

τ ≀ τ
  
0 / D(Q , J ) / D(F , J ) / D(P , J ) / 0.
γ ϕ

Here the rows are exact, while one needs a proof that the bottom row is exact. It is easy to see that γ is injective and ϕ is
surjective and ϕγ = 0. Suppose ϕ(m, m + z ) = 0 for some m ∈ F , z ∈ IF . Then f (m) = f (z ) = 0. Therefore, g (u) = m and
g (v) = z for some u, v ∈ Q . Let ϵ : F → Q be a splitting of g . Then v = ϵ g (v) = ϵ(z ) ∈ IQ . Hence (u, u + v) ∈ D(Q , I)
and γ (u, u + v) = (m, m + z ). This establishes that the bottom row is exact.
Since the middle vertical map is an isomorphism, the first vertical map is injective. This completes the proof. (Alternately,
one could use the Snake lemma to prove the same.) 

The following lemma will be of our interest subsequently.

Lemma 2.6. With the notations as above, let J be an ideal of the double D = D(A, I). If height (J ) = n, then height (p1 (J )) ≥ n.

Proof. Suppose p1 (J ) ⊆ ℘ ∈ Spec (R) are minimal. We will prove that height (℘) ≥ n. We have J ⊆ p−
1 (p1 (J )) ⊆ p1 (℘).
1 −1

There is a prime ideal P ∈ Spec (D) such that P is minimal over J and P ⊆ p1 (℘). Write m = height (P ). Then
−1

m ≥ n = height (J ). Let

P0 ⊆ P1 ⊆ P2 · · · ⊆ Pm = P

be a strictly increasing chain of primes in Spec (D). Write, ℘i = Pi ∩ R = ∆−1 (Pi ). Since, R → D is integral, there is no
inclusion relationship between two primes in D over the same prime ℘ ∈ Spec (R) (see [10, Theorem 9.3, pp. 66]). So,

℘0 ⊆ ℘1 ⊆ ℘2 · · · ⊆ ℘m
is a strictly increasing chain in Spec (R). Further, ℘m ⊆ p1 (P ) ⊆ p1 (p−
1 (℘)) = ℘. Therefore, height (℘) ≥ m ≥ n. The proof
1

is complete. 

3. Pull-back and functoriality

In this section, we define some pull-back homomorphisms of the obstruction groups, corresponding to some suitable
ring homomorphisms. First one will correspond to the quotient homomorphisms q : A → A/J .

Definition 3.1. Let A be a noetherian commutative ring with dim A = d and J ⊆ A be an ideal. Let L be a rank one projective
A-module. For integers n, with 2n ≥ d + 3, there is a group homomorphism
 
A L
ρ = ρJ : E n (A, L) → E n ,
J JL
defined as follows:
Write F = L ⊕ An−1 . Let ω : F /IF  I /I 2 be local L-orientation.
 We can find an ideal I1 and a local orientation
ω1 : F /I1 F  I1 /I12 such that (I , ω) = (I1 , ω1 ) and height I1 + J
J
≥ n. Then, ω1 induces an orientation β as in the following
commutative diagram:
ω1
F /I1 F / / I1 /I 2
1

 
F // I1 + J
.
(I1 + J )F β I12 + J

Define
 
I1 + J
ρ(I , ω) = ,β .
J

We use the notations q∗ = E n (q) = ρ = ρJ , corresponding to notation for the quotient map q : A → A/J . This
homomorphism will be called a pull-back homomorphism.
 S. Mandal, Y. Yang / Journal of Pure and Applied Algebra 216 (2012) 2159–2169 2163

Proof that ρ Well Defined. First, we define a homomorphism

     
A L I1 + J A L
ϕ : Gn (A, L) → E n , by ϕ(I , ω) = ,β ∈ En , ,
J JL J J JL

where β is as above. Two different representatives of ω1 leads to the same, because transvections of F /I1 F will induce
transvections of F1 /(J + I1 )F .
First, we prove that if (I1 , ω1 ) is global, then so is the image. If it is global, then ω1 lifts to a surjective homomorphism
f : F  I1 . Then, f induces a surjective lift g of β, as demonstrated by the following diagram:

GG
f
/ / I1
F
GG EE
EE
GG EE
GG EE
GG# E" "
#
F /JF
g
// I1 + J
J

 
ω1
F /I1 F / / I1 / I 2
EE 1
CC
EE CC
EE CC
EE CC! 
E" "  ! 
β
F / / I12 + J .
(I1 + J )F I1 + J

This establishes that β is global. Now, suppose

(I , ω) = (I1 , ω1 ) = (I2 , ω2 ) ∈ E n (A, L), with height (I1 /J ) = height (I2 /J ) = n.

There is an ideal K and a surjective lift f : F  I1 ∩ K of ω1 , such that I1 + K = A, height (K ) ≥ n. Since 2n > d, we can also
assume K + I2 = A. Let ωK : F /KF  K /K 2 be induced by f . We have,

(I1 , ω1 ) + (K , ωK ) = (I2 , ω2 ) + (K , ωK ) = 0.
By Theorem 2.4, there is a surjective homomorphism f2 : F  I2 ∩ K that lifts ω2 and ωK . So, both (I1 ∩ K , ω1 ⊗ ωK ), (I2 ∩
K , ω2 ⊗ ωK ) are global. Since the image of a global orientation is global, the images of both (I1 , ω1 ), (I2 , ω2 ) are negative of
that of (K , ωK ).
Therefore, the homomorphism ϕ is well defined. Again, since image of a global orientation is global, ϕ factors through a
homomorphism
 
A L
ρ : E n (A, L) → E n , .
J JL

This completes the proof that ρ is well defined. 

Proposition 3.2. Let A, J , d, n, L be as in Definition 3.1. Let J1 be another ideal. Then the diagram
ρJ
/ En
 
E n (A, L)
KK
A
J
, JLL
KK
KK
KK ρ(J1 +J )/J
ρJ +J1 KK
K% 
 
En A
, L
J +J1 (J +J1 )L

commutes.

Proof. Write ρ0 = ρJ +J1 , ρ = ρJ , ρ1 = ρA/(J +J1 ) . Write I = J + J1 and F = L ⊕ An−1 . Let ω : F /IF  I /I 2 be a local L-
 
orientation. We can find I1 and a local L-orientation ω1 : F /I1 F  I1 /I12 such that (1) (I , ω) = (I1 , ω1 ), (2) height
I1 + J
J
≥n
   
and (3) height
I1 + J +J1
J +J1
= height I 1 +I
I
≥ n. We have,
   
I1 + J I1 + J + J 1
ρ1 ρ(I , ω) = ρ1 ,β = ,γ = ρ0 (I , ω)
J J + J1
2164 S. Mandal, Y. Yang / Journal of Pure and Applied Algebra 216 (2012) 2159–2169

where β, γ are induced by ω1 , according to the following commutative diagram:

F / I1 F // F / F
(I1 + J )F (I1 + J +J1 )F

ω1 β γ
  
I1 /I12 / I1 + J
/ I1 + J + J1
.
I12 + J I12 + J + J1

This completes the proof. 

Next, we define pull-back homomorphisms of the obstruction groups, corresponding to some suitable ring
homomorphisms R → A.
Definition 3.3. Suppose f : R → A is a homomorphism of two noetherian commutative rings, with dim R = d1 , dim A = d2 .
Let n ≥ 1 be an integer. For an ideal I of R, we will denote IA := f (I )A. Assume that for any ideal I of R, which is locally
generated by n elements and height (I ) = n, we have height (IA) ≥ n.
Let L be a projective R-module of rank one and L′ = L ⊗ A. Then, there is a homomorphism
f ∗ = E (f ) = E n (f ) : E n (R, L) → E n (A, L′ )
defined below. This homomorphism will be called a pull-back homomorphism.
To define E n (f ), we proceed as follows. Write F = Fn = L ⊕ Rn−1 , F ′ = F ⊗ A. Let I be an ideal of R of height n and
ω : F /IF  I /I 2 be a local L-orientation. Write J = IA. Then ω induces a local L′ -orientation ω′ by the following commutative
diagram:

F /IF
ω / I /I 2

 
F ⊗A / IA
.
I ( F ⊗ A) I2A
ω′

The association (I , ω) → (J , ω′ ) ∈ E n (A, L′ ) defines a group homomorphism

ϕ0 : Gn (R, L) → E n (A, L′ ).
If (I , ω) is global, ω lifts to a surjection Ω : F  I . It is easy to see that Ω ⊗ IdA induces a surjective lift Ω ′ : F ⊗ A  J of
ω . So, ϕ0 ((I , ω)) = 0. Therefore, ϕ0 factors through a homomorphism
′

E n (f ) : E n (R, L) → E n (A, L′ ).
This completes the definition of E n (f ) and establishes that it is well defined. 
Proposition 3.4. Let f : R → A, g : A → B be homomorphisms of noetherian commutative rings of finite dimension, satisfying
the properties of Definition 3.3. Let L be a rank one projective R-module. Then the diagram
E (f )
E n (R, L) / E n (A, L ⊗ A)
MMM
MMM
M
E (gf ) MMM
E (g )
& 
E n (B, L ⊗ B)
commutes.
Proof. Obviously from the Definition 3.3. 
Proposition 3.5. Let f : A → R be a homomorphism of commutative noetherian rings of finite dimension, and L be a rank
one projective A-module. Let r , s ≥ 1 be two integers be such that r ≥ 2, if L ̸= A. Assume that f satisfies the properties of
Definition 3.3 for n = r , s, r + s. Then, for x ∈ E r (A, L), y ∈ E s (A, A), we have
E (f )(x) ∩ E (f )(y) = E (f )(x ∩ y) ∈ E r +s (A, L ⊗ A),
where ∩ is defined as in [9, Theorem 4.5].
Proof. Write F = L ⊕ Ar −1 , F ′ = As . Using bilinearity, we can assume that x = (I , ω) and y = (J , ω′ ), where I , J ⊆ A are
ideals with height (I ) = r , height (J ) = s and ω : F /IF  I /I 2 , ω′ : F ′ /JF ′  J /J 2 are local orientations. We can assume that
height (I + J ) = r + s. So,
F ⊕ F′ (I + J )
x ∩ y = (I + J , ω0 ) where ω0 : 
(I + J )F ⊕ F ′ (I + J )2
is induced by ω, ω′ .
 S. Mandal, Y. Yang / Journal of Pure and Applied Algebra 216 (2012) 2159–2169 2165

Since f : A → R satisfies properties of Definition 3.3 for n = r , s, r + s the proposition follows by chasing the
definitions. 

Remark 3.6. The referee points out that the Definitions 3.1 and 3.3 can be combined to define pull-back homomorphisms
for more general ring homomorphisms f : A → B. For example, assume that both A, B contain a field k and f is a
k-algebra homomorphism. Then, the projection homomorphism p : A → A ⊗k B is flat, since B is flat over k and it
satisfies the conditions of Definition 3.3. So, for integers n ≥ 1 and any rank one projective A-module L, the pull-back
p∗ : E n (A, L) → E n (A ⊗k B, L ⊗k B) is defined. Now consider the surjective homomorphism q : A ⊗k B → B given by
q(x ⊗ y) = f (x)y. Then f = qp. If 2n ≥ dim(A ⊗ B) + 3 then the pull-back q∗ : E n (A ⊗ B, L ⊗k B) → E n (B, L ⊗A B) is defined.
So, the pull-back f ∗ := q∗ p∗ : E n (A, L) → E n (B, L ⊗A B) is defined. (In fact, the argument works, if k is any commutative ring
and k → B is flat.) The authors are thankful to the referee for this comment.

4. Exact sequences and excision

In this section, first we define the relative obstruction groups and then establish the exact sequences.

Definition 4.1. Let A be a commutative noetherian ring and I be an ideal of A. Let L be a projective A-module with
rank(L) = 1. Let n ≥ 1 be an integer. With D = D(A, I) and other notations as in Section 2.1, by Lemma 2.6 the pull-
back homomorphism

E (p1 ) : E n (D, D ⊗ L) → E n (A, L)

is well defined. The relative obstruction groups, E n (A, I, L) are defined as

E n (A, I, L) = Kernel (E (p1 )) .

The following theorem establishes an exact sequence.

Theorem 4.2. Let A be a noetherian commutative ring with dim(A) = d and I be an ideal. As before, p1 , p2 : D(A, I) → A
will denote the projections to the first and second coordinates. Suppose L is a projective A-module of rank one. For integers n with
2n ≥ d + 3, the following

E (p2 ) ρI
E n (A, I, L) / E n (A, L) / En( A , L
)
I IL

is an exact sequence.

Proof. Let q : A → A/I denote the quotient homomorphism and F = L ⊕ An−1 . First, we prove ρ E (p2 ) = 0. Suppose
D⊗F
(W , ω) ∈ E n (A, I, L) ⊆ E n (D, D ⊗ L) where ω : → W /W 2
W (D ⊗ F )

is a local orientation, with ideal W ⊆ D of height n. By Lemma 2.2, we can assume that height
 W +0×I 
0×I
≥ n and
∼ A
≥ n. Since p1 : 0×I → I ,
 W +I × 0  D
height I×0

p1 ( W ) + I
   
W +0×I
height = height ≥ n.
I 0×I
Write J1 = p1 (W ), and J2 = p2 (W ). Since, qp1 = qp2 , we have, J1 + I = J2 + I. For i = 1, 2 ω induces ωi , βi by the
commutative diagram:

D⊗F pi
/ F / F
W (D⊗F ) Ji F (Ji +I)F

ω ωi βi
  
W /W 2 / Ji /J 2 / Ji +I
.
pi i Ji2 +I

Then, for i = 1, 2 we have E (pi )(W , ω) = (Ji , ωi ) ∈ E n (A, L). It follows, β1 = β2 . Since height is consistent,
   
J2 + I J1 + I
ρI E (p2 )(W , ω) = , β2 = , β1 = ρI E (p1 )(W , ω) = 0.
I I
Conversely, suppose x ∈ E n (A, L), are such that ρI (x) = 0. Since, 2n > d, we can write x = (I, ω) where I ⊆ A is an
ideal of height n and ω : F /IF  I /I 2 is a local orientation. We can further assume that height I +II ≥ n. Now, ω induces
2166 S. Mandal, Y. Yang / Journal of Pure and Applied Algebra 216 (2012) 2159–2169

ω′ as in the following commutative diagram:

F /IF
ω / / I /I 2

 
F // I +I
.
(I +I)F ω′ I 2 +I

Since ρ(x) = , ω′ = 0, we have ω′ lifts to a surjection f ′ :  I +II . We have the following fiber product diagram:
 I +I  F
I IF

F /(I ∩ I)F / F /IF

M PPP
M PPP ω
Mf PPP
M PPP
M& & PP' '
I /I ∩ I 2 / I /I 2

 
F /IF / F /(I + I)F
MMM PPP
MMMf ′ PPPω′
MMM PPP
MM& &  PPP
' 
(I + I )/I / (I + I)/(I 2 + I)

where f is defined by properties of fiber product diagrams.  ByI Lemma 2.3, there is a surjective lift ϕ : F  I ∩ K , of f , such
that (1) K + I ∩ I 2 = A, (2) height (K ) ≥ n and (3) height K + ≥ n.

I
Now, ϕ defines an element (K , ωK ) = −(I , ω) ∈ E n (A, L). Define, W = ∆(K ) + I × 0 = {(x, y) ∈ D : y ∈ K }. In fact, W
is defined by the fiber product of A and K as follows:

W ⊆D
p1
/ A=K +I

p2

 
K / K +I
= A/I.
I

By Lemma 2.5, D ⊗ F = D(F , J ). We consider the following fiber product diagram:

D ⊗ FI
p1
/F C
I CC
IωW CC0
I CC
I$ $ CC!
p2 W /W 2
p1
/! 0

 
F HH ≀ / F /I F
HH BB
HH BB 0
H BB
ωK HHH BB
$$  ! 
K /K 2 / 0.

Here ωK is defined by properties of fiber product diagrams. (Alternately, one can prove elementwise that W /W 2 → K /K 2 is
an isomorphism, using the fact that x + y = 1 for some x ∈ K 2 , y ∈ I2 and (1, x) = (x, x) + (y, 0) ∈ W 2 .)
We consider ωW as an orientation. It follows E (p2 )(W , ωW ) = (K , ωK ) = −(I , ω) = −x. So, x ∈ imageE (p2 ). Since
p1 (W ) = A, we have (W , ωW ) ∈ ker (E (p1 )) = E n (A, L, I). This completes the proof of (4.2). 
With further conditions, we extend the above sequences as follows.
Theorem 4.3. Use notations as in Theorem 4.2 and assume 2n ≥ d + 3. Write A0 = IA . Assume that the quotient homomorphism
q : A  A0 has a splitting β : A0 → A and L = L0 ⊗β A0 for some rank one projective A0 -module L0 .
(1) Then, the sequence
E (p2 ) ρI
0 / E n (A, I, L) / E n (A, L) / En( A , L
)
I IL

is exact.
 S. Mandal, Y. Yang / Journal of Pure and Applied Algebra 216 (2012) 2159–2169 2167

(2) Assume that, for each locally n-generated ideal I0 of A0 , of height n, we have height (β(I0 )A) ≥ n. Then, the sequence

E (p2 ) ρI
E n (A, I, L) / E n (A, L) / En( A , L
) /0
I IL

is exact and ρI splits.

Proof. First, we prove (2). Because of Theorem 4.2, we only need to prove that ρ is split-surjective. Since E (β) is defined, it
is easy to check that ρI E (β) = Id. So, ρI is surjective and splits.
Again by Theorem 4.2, to prove (1), we need to prove that E (p2 ) is injective on E n (A, I, L). Let x ∈ E n (A, I, L)
⊆ E n (D, L ⊗ D) such that E (p2 )(x) = 0. Since 2n > d = dim(D), we can assume x = (W , ω), where F = L ⊕ An−1
and W is an ideal of D with height (W ) ≥ n and ω : WF(F⊗⊗DD)  WW2 is a local orientation.
Since L = L0 ⊗β A, we have L0 = L ⊗A A0 . We will use the notation F ′ = F ⊗ D. Consider the following fiber product
diagram:

F′ / F′
(W ∩ (I×I))F ′
K WF ′ KKK
K KKK
K
f
KωKK
K KKK
K% % K% %
W / W
W 2 ∩ (I×I) W2

 
F′ / F′
(I×I)F ′ (W +(I×I))F ′
KKK KKK
KKKf ′ KKωK⊗ I×D I
KKK KKK
K% %  K% 
W +(I×I) / W2+(I×I) .
(I×I) W +(I×I)

In this diagram, f ′ is a surjective lift of ω ⊗ I× D

, which exists, by Theorem 2.4, because WI+×II×I , ω ⊗ I×D I =
 
I
ρ E (p1 )(W , ω) = 0. Also, f is given by the properties of fiber product diagrams and is surjective.
 K +I×0  Ω : F  W ∩ K where K is an ideal with (1) W + K = D, (2) height (K ) ≥ n, (3)
′
By Lemma
 K +0×I 2.2, f lifts to a surjection,
height 0×I ≥ n (4) height I×0 ≥ n (5) K + I × I = D.
Ω induces a local orientation ωK : F ′ /KF ′  K /K 2 . We have (I , ω) + (K , ω′ ) = 0. Write K1 = p1 (K ), K2 = p2 (K2 ). Note
q(K1 ) = q(K2 ) = image(K ) = A0 , and height (Ki ) ≥ n for i = 1, 2. For i = 1, 2, let ωi : F /Ki F  Ki /Ki2 be induced by ωK .
The following are some observations:

1 (K1 ) = K + 0 ⊗ I and p2 (K2 ) = K + I ⊗ 0.

(1) Claim: p− 1 −1

To see this, note K + 0 ⊗ I ⊆ p− 1 (K ). Conversely, let (x, y) ∈ p (K ). Then, (x, y′ ) ∈ K for some y′ ∈ A. Therefore,
1 −1

(x, y) = (x, y ) + (0, (y − y )) = (x, y ) + (0, (y − x) + (x − y )) ∈ K + 0 ⊗ I. This establishes the claim.

′ ′ ′ ′

(2) Since K + I × I = D we have Ki + I = A for i = 1, 2.

(3) Also E (pi )(K , ωK ) = (Ki , ωi ) = 0 for i = 1, 2. Therefore, by Theorem 2.4, there are surjective lifts Ωi : F  Ki of ωi .

Write F0 = L0 ⊕ An0−1 . Then, F ⊗ A0 = L ⊗ A0 ⊕ An0−1 ≈ F0 . Since Ki + I = A, for i = 1, 2, we have, Ωi ⊗ A0 surjects on

to A0 . Write Ωi0 = Ωi ⊗A A0 : F0  A0 for i = 1, 2.
Write J1 = {(a, a + z ) : a ∈ K1 , z ∈ I} and J2 = {(b + z , b) : b ∈ K2 , z ∈ I}. Both J1 and J2 are ideals of D. Consider the
following two fiber product diagrams:

F′
p1
/F F′
p1
/F
> @@ @ AA
> @@ Ω1 @ AA Ω20 ⊗β A
> @@ @ AA
Γ1 >  @@ Γ2 @  AA

p2 J1 / K1 p2 J2 /A

   
/ F0 / F0
F>> ?? 0 F ?? AA 0
>> ??Ω1 ?? AAΩ2
>> ?? ?? AA
Ω10 ⊗β A >> ?   Ω2 ??   A 
 
A / A0 K2 / A0 .
2168 S. Mandal, Y. Yang / Journal of Pure and Applied Algebra 216 (2012) 2159–2169

Here Γ1 , Γ2 are obtained by the properties of fiber product diagrams and they are surjective by the same. We gather some
facts below:
(1) We have J1 + J2 = D.
Proof. We have K + I × I = D. So, (x, y) + (u, v) = (1, 1) for some (x, y) ∈ K and u, v ∈ I. So, (1, y) = (y + v, y) ∈ J2
and (x, 1) = (x, x + u) ∈ J1 . Therefore, (1, 1) = (1, y) + (x, 1)(0, v) ∈ J1 + J2 . This establishes J1 + J2 = D or (1). 
(2) We have K = J1 ∩ J2 .
Proof. Clearly, for (a, b) ∈ K , we have (a, b) = (a, a + (b − a)) ∈ J1 . Similarly, (a, b) ∈ J2 . Therefore, K ⊆ J1 ∩ J2 .
Now, let (a, b) ∈ J1 ∩ J2 . Looking at the description of Ji , it follows a ∈ K1 and b ∈ K2 . So, (a, c ) ∈ K for some c ∈ A.
So, (a, b) = 
(a, c ) + (0, b − c ). Since (a, c ) ∈ K ⊆ J1 ∩ J2 , we have (0, b − c ) ∈ J1 ∩ J2 . Therefore, b − c ∈ K2 ∩ I = K2 I.
So, b − c = yi zi for some (xi , yi ) ∈ K and zi ∈ I. Hence

(a, b) = (a, c ) + (0, b − c ) = (a, c ) + (xi , yi )(0, zi ) ∈ K .
This establishes (2). 
(3) Since K = J1 ∩ J2 , we have height (Ji ) ≥ n. So, Γi induce local orientation γi : F ′ /Ji F ′  Ji /Ji2 for i = 1, 2. Indeed, they
are global. So,
(J1 , γ1 ) = (J2 , γ2 ) = 0 ∈ E n (D, L ⊗ D).
(4) In fact, ωK ⊗ D/Ji = γi for i = 1, 2.
Proof. We will check for i = 1. We will check that Ω (m) − Γ1 (m) ∈ J12 , for m ∈ F ′ . Since K12 + I2 = A, we have
x + c = 1 for some (x, y) ∈ K 2 and c ∈ I2 . So, (x, 1) = (x, y) + (0, (1 − x) + (x − y)) ∈ J12 . Therefore, for any z ∈ I we
have (0, z ) = (x, 1)(0, z ) ∈ J12 .
Let m ∈ F ′ and Γ1 (m) = (a, b), Ω (m) = (x, y). Since the first coordinates of both agree in K1 /K12 , we have
c = x − a ∈ K12 . Therefore, (c , d) ∈ K 2 ⊆ J12 for some d ∈ A. So,

(x, y) − (a, b) = (c , y − b) = (c , d) + (0, y − b − d) ∈ J12 .

This establishes (4). 
It follows from above,
(K , ωK ) = (J1 , γ1 ) + (J2 , γ2 ) = 0.
Hence (W , ω) = −(K , ωK ) = 0. The proof is complete. 
As in the paper [12] of Yang, immediate application of Theorem 4.3 would provide exact sequences for Euler class groups
of polynomial rings and Laurent polynomial rings.
Corollary 4.4. Let R be a commutative ring with dim R = d. Let A = R[X ] be the polynomial ring and B = R[X , X −1 ] be the
Laurent polynomial ring. Let L0 be a projective R-module of rank one. Write L = L0 ⊗ A, L′ = L0 ⊗ B. Assume that 2n ≥ d + 4.
We have the following.
(1) The sequence,
E (p2 ) ρX
0 / E n (A, (X ), L) / E n (A, L) / E n (R, L0 ) /0
is a split exact sequence.
(2) The sequence,
E (p2 ) ρX −1
/ E n B, (X − 1), L′ / E n B, L′ / E n (R, L0 ) /0
   
0

is a split exact sequence.

(3) Further, if R is a regular domain that is essentially of the finite type over an infinite field k, then

ρX : E n (A, L)
∼ / E n (R, L0 )
is an isomorphism. In particular, the relative group
E n (A, (X ), L) = 0.

Proof. Obviously, (1) and (2) are direct consequences of Theorem 4.3. To prove (3), we need to show that the first
homomorphism is zero. As in Theorem 4.3, p1 , p2 will denote the projection maps and q : A  R will be the quotient
homomorphism. The first homomorphism in the sequence is E (p2 ). Suppose x ∈ E n (A, (X ), L) . We will prove E (p2 )(x) = 0.
We can assume that x = (I , ω) where I is an ideal of D = D(A, (X )), with height (I ) ≥ n, height (p1 (I )) ≥ n and
height (p2 (I )) ≥ n. Write p1 (I ) = I1 , p2 (I ) = I2 and I0 = q(I1 ) = q(I2 ). Therefore, I0 = (I1 , X ) = (I2 , X ).
 S. Mandal, Y. Yang / Journal of Pure and Applied Algebra 216 (2012) 2159–2169 2169

For i = 0, 1, 2, let ωi : F /Ii F  Ii /Ii2 be induced by ω. From the exactness of the sequence, we have
(I0 , ω0 ) = ρX (I2 , ω2 ) = ρX E (p2 )(I , ω) = 0.
Therefore, by Theorem 2.4, ω0 lifts to a surjective homomorphism f0 : L0 ⊕ Rn−1  I0 . Now, by [2, Theorem 4.13], there is a
surjective lift f2 : L ⊕ An−1  I2 such that f2 ⊗ A/(X ) = f0 . So, (I2 , ω2 ) = 0. This completes the proof. 

Acknowledgement

The first author was partially supported by a grant from General Research Fund of University of Kansas.

References

[1] Jean Barge, Morel, Fabien Groupe de Chow des cycles orients et classe d’Euler des fibres vectoriels (The Chow group of oriented cycles and the Euler
class of vector bundles), C. R. Acad. Sci. Paris Sr. I Math. 330 (4) (2000) 287–290 (in French).
[2] S.M. Bhatwadekar, M.K. Keshari, A question of Nori: projective generation of ideals, K-Theory 28 (2003) 329–351.
[3] S.M. Bhatwadekar, Raja Sridharan, The Euler class group of a Noetherian ring, Compos. Math. 122 (2) (2000) 183–222.
[4] S.M. Bhatwadekar, Raja Sridharan, On Euler classes and stably free projective modules, in: Algebra, Arithmetic and Geometry, Part I, II (Mumbai, 2000),
in: Tata Inst. Fund. Res. Stud. Math., vol. 16, Tata Inst. Fund. Res., Bombay, 2002, pp. 139–158.
[5] Jean Fasel, Groupes de Chow-Witt, Mém. Soc. Math. Fr. (N.S.) 113 (2008).
[6] Satya Mandal, Homotopy of sections of projective modules, J. Algebraic Geom. 1 (4) (1992) 639–646.
[7] Satya Mandal, Projective modules and complete intersections, in: Lecture Notes in Mathematics, vol. 1672, Springer-Verlag, Berlin, 1997, viii+114 pp.
[8] Satya Mandal, Raja Sridharan, Euler Classes and Complete Intersections, J. Math. Kyoto Univ. 36 (3) (1996) 453–470.
[9] Satya Mandal, Yong Yang, Intersection theory of algebraic obstructions, J. Pure Appl. Algebra 214 (2010) 2279–2293.
[10] Hideyuki Matsumura, Commutative ring theory, in: Cambridge Studies in Advanced Mathematics, vol. 8, Cambridge University Press, Cambridge,
1986, xiv+320 pp.
[11] N.E. Steenrod, The Topology of Fibre Bundles, Princeton University Press, 1951.
[12] Yong Yang, Homology sequence and excision theorem for Euler class group, Pacific J. Math. 249 (1) (2011) 237–254.