This action might not be possible to undo. Are you sure you want to continue?

r

X

i

v

:

1

2

0

3

.

5

5

7

8

v

1

[

m

a

t

h

.

A

C

]

2

6

M

a

r

2

0

1

2

VARIATION OF HILBERT COEFFICIENTS

LAURA GHEZZI, SHIRO GOTO, JOOYOUN HONG, AND WOLMER V. VASCONCELOS

Abstract. For a Noetherian local ring (R, m), the ﬁrst two Hilbert coeﬃcients, e0 and e1, of the

I-adic ﬁltration of an m-primary ideal I are known to code for properties of R, of the blowup of

Spec(R) along V (I), and even of their normalizations. We give estimations for these coeﬃcients

when I is enlarged (in the case of e1 in the same integral closure class) for general Noetherian local

rings.

1. Introduction

Let (R, m) be a Noetherian local ring of dimension d ≥ 1, and let I be an m-primary ideal. We

will consider multiplicative, decreasing ﬁltrations of R ideals,

/ = ¦I

n

[ I

0

= R, I

n+1

= II

n

, ∀n ≫0¦,

integral over the I-adic ﬁltration, conveniently coded in the corresponding Rees algebra and its

associated graded ring

!(/) =

n≥0

I

n

, gr

A

(R) =

n≥0

I

n

/I

n+1

.

Let ! =

n≥0

I

n

be the integral closure of the Rees algebra ! = !(I) with I

n

= I

n

for all n ≥ 0,

which we assume to be ﬁnite over !.

We will consider the Hilbert coeﬃcients e

i

(I) associated to m-primary ideals I, for i = 0, 1. These

integers play important roles in the corresponding blowup algebras. Some of these issues have a

long tradition in the context of Cohen-Macaulay local rings, but others are of a recent vintage for

general Noetherian local rings. From the several problem areas, we highlight the following:

(i) The comparison between e

0

and e

1

;

(ii) e

1

and normalization;

(iii) The structure of R associated to the values of e

1

;

(iv) Variation of e

i

, that is how e

i

(I) changes when I is enlarged.

We are concerned here with the last item but give brief comments on the others ﬁrst.

AMS 2010 Mathematics Subject Classiﬁcation. Primary 13A30; Secondary 13B22, 13H10, 13H15.

The ﬁrst author is partially supported by a grant from the City University of New York PSC-CUNY Research

Award Program-41. The second author is partially supported by Grant-in-Aid for Scientiﬁc Researches (C) in Japan

(19540054) and by a grant from MIMS (Meiji Institute for Advanced Study of Mathematical Sciences). The last

author is partially supported by the NSF.

1

2 LAURA GHEZZI, SHIRO GOTO, JOOYOUN HONG, AND WOLMER V. VASCONCELOS

(i) For Cohen-Macaulay rings, an uniform bound for e

1

(I) ﬁrst appeared for rings of dimension 1

in the work of D. Kirby ([K]),

e

1

(m) ≤

_

e

0

(m)

2

_

.

Progressively, quadratic bounds of this type were developed for arbitrary m-primary ideals in all

dimensions by several authors. As a basic source, [RV2] has a systematic development of these

formulas along with a comprehensive bibliography. It also treats more general ﬁltrations which

were helpful to us. Among the formulas which more directly inﬂuenced the authors here, we single

out two developed in the work of J. Elias ([E1, E2]) and M. E. Rossi and G. Valla ([RV1, RV2]).

For an d-dimensional Cohen-Macaulay local ring and an m-primary ideal I, the ﬁrst bound asserts

that if I is minimally generated by m = ν(I) elements,

e

1

(I) ≤

_

e

0

(I)

2

_

−

_

m−d

2

_

−λ(R/I) + 1.

The other bound uses the m-adic order of I, that if I ⊂ m

s

and I ,= m

s

, then

e

1

(I) ≤

_

e

0

(I) −s

2

_

.

Recently, K. Hanumanthu and C. Huneke ([HH]) brought a new parameter to bear on these formulas

with their proof that

e

1

(I) ≤

_

e

0

(I) −k

2

_

,

where k is the maximal length of chains of integrally closed ideals between I and m.

(ii) Since e

1

(I) ≤ e

1

(I) := e

1

(!), bounds with a diﬀerent character arise. A baseline is the fact

that when R is analytically unramiﬁed, but not necessarilly Cohen-Macaulay, one has e

1

(I) ≥ 0

([GHM]). An upper bound for e

1

(I) (see [PUV] for other bounds) is the following. Let (R, m) be

a reduced Cohen-Macaulay local ring of dimension d, essentially of ﬁnite type over a perfect ﬁeld,

and let I be an m-primary ideal. Let δ be a regular element of the Jacobian ideal of R. Then

e

1

(I) ≤ e

1

(I) ≤

t

t + 1

_

(d −1)e

0

(I) + e

0

((I + δR)/δR)

¸

,

where t is the Cohen-Macaulay type of R. In particular, if R is a regular local ring

e

1

(I) ≤

(d −1)e

0

(I)

2

.

(iii) When R is not Cohen-Macaulay, the issues become less structured since the values of e

1

(I)

may be negative. In fact, using the values of e

1

(I) for ideals generated by systems of parameters

led to the characterization of several properties (Cohen-Macaulay, Buchsbaum, ﬁnite cohomology)

of the ring R itself (see [GhGHOPV], [GhHV], [GO], [MV], [V2]).

(iv) We shall now outline the main results of this note.(We refer to [V1] for basic deﬁnitions and

Rees algebras theory.) Sections 2 and 3 are organized around a list of questions about the changes

that e

0

(I) and e

1

(I) undergo when I varies. An important case is

e

0

(J), e

1

(J) −→e

0

(I), e

1

(I), I = (J, x).

VARIATION OF HILBERT COEFFICIENTS 3

Clearly the optimal baseline is that of an ideal J generated by a system of parameters, but we

will consider very general cases. As will be seen, some relationships involve the multiplicity f

0

(J)

of the special ﬁber. To describe one of these estimates, let (R, m) be a Noetherian local ring of

dimension d ≥ 1, let J be an m-primary ideal and let I = (J, h

1

, . . . , h

m

) be integral over J of

reduction number s = red

J

(I). Then Theorem 2.6 asserts that

e

1

(I) −e

1

(J) ≤ λ(R/(J : I))

__

m + s

s

_

−1

_

f

0

(J),

where f

0

(J) is the multiplicity of the special ﬁber of !(J) =

n≥0

J

n

. We add a word of warning

in reading some of the formulas with terms like e

1

(I) − e

1

(J). When J is a minimal reduction of

I, e

1

(J) is always non-positive, according to [MSV], and vanishes when R is Cohen–Macaulay. In

fact, for unmixed local rings the vanishing characterizes Cohen-Macaulayness ([GhGHOPV]).

In Section 3, we address the need to link the value of red

J

(I) to other properties of J. This

is a well-known fact when R is a Cohen-Macaulay ring, but we give a general formulation in

Theorem 3.3: Let (R, m) be a Noetherian local ring of dimension d ≥ 1 and inﬁnite residue ﬁeld.

For an m-primary ideal I and a minimal reduction J of I, there exists a minimal reduction Q of I

such that

red

Q

(I) ≤ max¦dλ(R/J) −2d + 1, 0¦.

2. Upper bounds for the variations of e

0

(I) and e

1

(I)

In our calculations we make repeated use of the following elementary observation.

Lemma 2.1. If (R, m) is a Noetherian local ring and M is an R–module of ﬁnite length λ(M),

then

λ(M ⊗N) ≤ λ(M)ν(N)

for every ﬁnitely generated R–module N, where ν(N) denotes the minimal number of generators

for N.

Proof. Induct on n = λ(M). If n = 1, then M ≃ R/m and the assertion is clear. Suppose that

n ≥ 2 and choose an R–submodule L of M with λ(L) = 1. By tensoring 0 →L →M →M/L →0

with N, we get the exact sequence

L ⊗N →M ⊗N →(M/L) ⊗N →0.

Since λ(M/L) = n −1, the induction hypothesis shows

λ((M/L) ⊗N) ≤ (n −1)ν(N),

so that

λ(M ⊗N) ≤ λ(L ⊗N) + λ((M/L) ⊗N) ≤ (1 + (n −1))ν(N) = λ(M)ν(N).

✷

Theorem 2.2. Let (R, m) be a Noetherian local ring of dimension d and let J ⊂ I = (J, h) be

m–primary ideals of R. Then

e

0

(J) −e

0

(I) ≤ λ(R/(J : I))f

0

(J).

4 LAURA GHEZZI, SHIRO GOTO, JOOYOUN HONG, AND WOLMER V. VASCONCELOS

Proof. For n ∈ N, consider the following ﬁltration:

J

n

= M

0

⊂ M

1

= (M

0

, J

n−1

h) ⊂ ⊂ M

r−1

= (M

r−2

, J

n−r+1

h

r−1

)

⊂ M

r

= (M

r−1

, J

n−r

h

r

)

⊂ ⊂ M

n

= (M

n−1

, h

n

) = I

n

.

Then we obtain

λ(R/J

n

) −λ(R/I

n

) = λ(I

n

/J

n

) = λ(M

n

/M

0

) =

n

r=1

λ(M

r

/M

r−1

).

For each r, M

r

/M

r−1

is generated by the image of h

r

J

n−r

+M

r−1

. Consider the natural surjection

ζ : R/(J : h) ⊗J

n−r

։M

r

/M

r−1

= (h

r

J

n−r

+ M

r−1

)/M

r−1

.

Using Lemma 2.1, we have

λ(M

r

/M

r−1

) ≤ λ(R/(J : h) ⊗J

n−r

) ≤ λ(R/(J : h))ν(J

n−r

).

It follows that

λ(R/J

n

) −λ(R/I

n

) ≤ λ(R/(J : I))

n−1

r=0

ν(J

r

).

The iterated Hilbert function

n−1

r=0

ν(J

r

) is of polynomial type of degree d with leading (binomial)

coeﬃcient f

0

(J). Also, for n ≫ 0, λ(R/J

n

) −λ(R/I

n

) is the diﬀerence of two polynomials of

degree d and leading (binomial) coeﬃcients e

0

(J) and e

0

(I). Hence

e

0

(J) −e

0

(I) ≤ λ(R/(J : I))f

0

(J).

✷

Theorem 2.3. Let (R, m) be a Noetherian local ring of dimension d ≥ 1 and let J ⊂ I = (J, h) be

m–primary ideals. If h is integral over J, then

e

1

(I) −e

1

(J) ≤ red

J

(I)λ(R/(J : I))f

0

(J),

where red

J

(I) is the reduction number of I with respect to J.

Proof. Let s = red

J

(I). Then h

s+1

∈ JI

s

. For n ≥ s, we obtain the following ﬁltration:

J

n

= M

0

⊂ M

1

= (M

0

, J

n−1

h) ⊂ ⊂ M

r

= (M

r−1

, J

n−r

h

r

)

⊂ ⊂ M

s

= (M

s−1

, J

n−s

h

s

) = I

n

.

Therefore

λ(R/J

n

) −λ(R/I

n

) = λ(I

n

/J

n

) =

s

r=1

λ(M

r

/M

r−1

) ≤ λ(R/(J : I))

s

r=1

ν(J

n−r

).

VARIATION OF HILBERT COEFFICIENTS 5

Now for n ≫ 0, λ(R/J

n

) − λ(R/I

n

) is the diﬀerence of two polynomials of degree d and with

same leading (binomial) coeﬃcients e

0

(J) and e

0

(I), therefore it is at most a polynomial of degree

d −1 and leading coeﬃcient e

1

(I) −e

1

(J). On the other hand, for n ≫0, we have

λ(R/(J : h))

s

r=1

ν(J

n−r

) ≤ λ(R/(J : h))s

d−1

i=0

(−1)

i

f

i

(J)

_

n + d −i −2

d −i −1

_

,

which proves that

e

1

(I) −e

1

(J) ≤ red

J

(I)λ(R/(J : I))f

0

(J).

✷

Corollary 2.4. Let (R, m) be a Noetherian local ring of dimension d ≥ 1 and inﬁnite residue ﬁeld.

Let Q ⊂ I = (Q, h) be m–primary ideals such that Q is a minimal reduction of I. Then

e

1

(I) ≤ red

Q

(I)λ(R/(Q : I)).

Moreover, if R is Gorenstein, then

e

1

(I) ≤ red

Q

(I)(e

0

(I) −λ(R/I)).

Proof. The ﬁrst assertion follows from e

1

(Q) ≤ 0 [MSV] and f

0

(Q) = 1 for every parameter ideal

Q. Suppose that R is Gorenstein. Then it is enough to show that

λ(R/(Q : I)) = e

0

(I) −λ(R/I).

This follows from

λ(R/(Q : I)) = λ(R/Q) −λ((Q : I)/Q) = e

0

(Q) −λ((Q : I)/Q) = e

0

(Q) −λ(R/I)

because (Q : I)/Q is the canonical module of R/I. ✷

Example 2.5. ([V1, Example 7.36]) Let k[x, y, z] be the polynomial ring over an inﬁnite ﬁeld k.

Let R = k[x, y, z]

(x,y,z)

and let J and I be R–ideals such that

J = (x

a

, y

b

, z

c

) ⊂ (J, x

α

y

β

z

γ

) = I,

where

α

a

+

β

b

+

γ

c

< 1. This inequality ensures that h = x

α

y

β

z

γ

/ ∈ J. Then we have

e

0

(J) −e

0

(I) = abc −(abγ + bcα + acβ) = abc

_

1 −

α

a

−

β

b

−

γ

c

_

.

Since (J : I) = (J : x

α

y

β

z

γ

) = (x

a−α

, y

b−β

, z

c−γ

), we obtain

λ(R/(J : I))f

0

(J) = (a −α)(b −β)(c −γ)

= abc −bcα −acβ −abγ + aβγ + bαγ + cαβ −αβγ

= e

0

(J) −e

0

(I) + αβγ

_

a

α

+

b

β

+

c

γ

−1

_

> e

0

(J) −e

0

(I).

6 LAURA GHEZZI, SHIRO GOTO, JOOYOUN HONG, AND WOLMER V. VASCONCELOS

Let Q = (x

a

−z

c

, y

b

−z

c

, x

α

y

β

z

γ

) and suppose that a > 3α, b > 3β, c > 3γ. Note that I =

(Q, z

c

). Then Q is a minimal reduction of I and the reduction number red

Q

(I) ≤ 2. We can

estimate e

1

(I):

e

1

(I) = e

1

(I) −e

1

(Q) ≤ 2λ(R/(Q : I)).

✷

Now we treat a general case of Theorem 2.3. Let J be an m–primary ideal and H = (h

1

, . . . , h

m

)

a set of elements integral over J. Write I = (J, H), where ν(H) = ν(I/J), and consider the

diﬀerence of Hilbert functions

λ(R/J

n

) −λ(R/I

n

) = λ((J, H)

n

/J

n

) = λ((J

n

, HJ

n−1

, . . . , H

n−1

J, H

n

)/J

n

)

=

n

r=1

λ(M

r

/M

r−1

),

where M

r

= (J

n

, HJ

n−1

, . . . , H

r−1

J

n−r+1

, H

r

J

n−r

). Note that M

r

/M

r−1

is generated by the

image of H

r

J

n−r

. More precisely, if I = (J, h

1

, . . . , h

m

), then M

r

/M

r−1

is generated by batches of

elements, diﬃcult to control. This ﬁltration has been used by several authors when J is generated

by a system of parameters. As ν(I

n

/J

n

) is increasing, the method of iterating the assertion in

Theorem 2.3 tends to induce a bigger upper bound for e

1

(I) − e

1

(J) than necessary. Instead, our

formulation using the ﬁltration above wraps it diﬀerently to accommodate our data.

Theorem 2.6. Let (R, m) be a Noetherian local ring of dimension d ≥ 1, let J be an m-primary

ideal and let I = (J, h

1

, . . . , h

m

) be integral over J of reduction number s = red

J

I. Then

e

1

(I) −e

1

(J) ≤ λ(R/(J : I))

__

m + s

s

_

−1

_

f

0

(J).

Proof. We have already given parts of the proof. The remaining part is to estimate the growth

of the length of M

r

/M

r−1

= [(h

1

, . . . , h

m

)

r

J

n−r

+ M

r−1

] /M

r−1

. We note that this module is

annihilated by J : I and is generated by the ‘monomials’ in the h

i

of degree r, with coeﬃcients in

J

n−r

. There is a natural surjection

Φ : R/(J : I) ⊗R

br

⊗J

n−r

−→M

r

/M

r−1

,

where b

r

=

_

m + r −1

r

_

. Therefore for n ≫0,

λ(R/J

n

) −λ(R/I

n

) =

s

r=1

λ(M

r

/M

r−1

)

≤

s

r=1

λ(R/(J : I))ν(J

n

)

_

m + r −1

r

_

= λ(R/(J : I))ν(J

n

)

__

m + s

s

_

−1

_

,

which completes the proof. ✷

VARIATION OF HILBERT COEFFICIENTS 7

In Theorem 2.6, if J is a minimal reduction of I, then it is well–known that m = ν(I) − ν(J)

does not depend on J because mI ∩J = mJ. Moreover, if R is Cohen–Macaulay, then λ(R/(J : I))

does not depend on J either, because λ(R/(J : I)) = e

0

(I) −λ(H

m

(I)), where H

m

(I) is the m–th

Koszul homology of I.

Proposition 2.7. Let (R, m) be a Noetherian local ring of dimension d ≥ 1, let J be an m-primary

ideal and let I = (J, h

1

, . . . , h

m

) be integral over J of reduction number s = red

J

I. Then

f

0

(I) ≤

_

1 + λ(R/(J : I))

__

m + s

s

_

−1

__

f

0

(J).

Proof. By tensoring the following exact sequence with R/m

0 −→J

n

−→I

n

−→I

n

/J

n

−→0,

we obtain

J

n

/mJ

n

−→I

n

/mI

n

−→(I

n

/J

n

) ⊗R/m →0.

Therefore, using Lemma 2.1, we get

λ(I

n

/mI

n

) −λ(J

n

/mJ

n

) ≤ λ((I

n

/J

n

) ⊗R/m)

≤ λ(I

n

/J

n

) = λ(R/J

n

) −λ(R/I

n

).

This induces the inequalities of the leading coeﬃcients (in degree d −1)

f

0

(I) −f

0

(J) ≤ e

1

(I) −e

1

(J).

Using Theorem 2.6, we obtain

f

0

(I) −f

0

(J) ≤ e

1

(I) −e

1

(J) ≤ λ(R/(J : I))

__

m + s

s

_

−1

_

f

0

(J),

which completes the proof. ✷

Remark 2.8. Note that the formulas for the variations of e

1

(I) and f

0

(I) require that the ideal I

has the same integral closure as J.

The values of the ﬁrst Hilbert coeﬃcients are also related to the multiplicity of certain Sally

modules, according to [C, Proposition 2.8]. Let (R, m) be a Noetherian local ring of dimension

d ≥ 1 with inﬁnite residue ﬁeld. Let I be an m–primary ideal and Q a minimal reduction of I. If

dim(S

Q

(I)) = d and H

0

m

(R) ⊂ I, then the multiplicity s

0

(Q, I) of the Sally module S

Q

(I) is

s

0

(Q, I) = e

1

(I) −e

1

(Q) −e

0

(I) + λ(R/I).

Corollary 2.9. Let (R, m) be a Noetherian local ring of dimension d ≥ 1 with inﬁnite residue ﬁeld.

Let I be an m–primary ideal and Q a minimal reduction of I. Suppose that dim(S

Q

(I)) = d and

that H

0

m

(R) ⊂ I. Then the multiplicity s

0

(Q, I) of the Sally module S

Q

(I) satisﬁes

s

0

(Q, I) ≤ −e

0

(I) + λ(R/I) + λ(R/(Q : I))

__

ν(I) −d + s

s

_

−1

_

,

where s = red

Q

(I) is the reduction number.

8 LAURA GHEZZI, SHIRO GOTO, JOOYOUN HONG, AND WOLMER V. VASCONCELOS

Example 2.10. Let R = k[x, y]

(x,y)

where k[x, y] denotes the polynomial ring over an inﬁnite ﬁeld

k. Let m = (x, y) and I = m

n

= (a

1

, . . . , a

n

, a

n+1

) for some n ≥ 2. We assume that Q = (a

1

, a

2

)

is a minimal reduction of I. Let J = (a

1

, a

2

, . . . , a

n

). Then since Q ⊆ J, I is integral over J with

red

J

(I) = 1, because I ,= J and red

Q

(I) = 1. Using e

1

(I) = e

1

(m

n

) =

1

2

n(n −1) and

e

1

(I) −e

1

(J) ≤ λ(R/(J : a

n+1

))f

0

(J),

we obtain

e

1

(J) ≥

1

2

n(n −1) −λ(R/(J : a

n+1

))f

0

(J).

One situation that may be amenable to further analysis is when I = J : m, or more generally

I = J : m

s

for some values of s. We refer to I as a socle extension of J.

Remark 2.11. (Reduction number one) Let (R, m) be a Noetherian local ring of dimension

d ≥ 1 with inﬁnite residue ﬁeld. Let I be an m–primary ideal and Q a minimal reduction of I.

Suppose that I

2

= QI. Then by Theorem 2.6 we get

e

1

(I) −e

1

(Q) ≤ λ(R/(Q : I))(ν(I) −d) ≤ λ(R/(Q : I))λ(I/Q).

Suppose that R is Cohen–Macaulay. Then since e

1

(I) = e

0

(I) − λ(R/I) = λ(I/Q) ([Hu, 2.1]), it

follows that

e

1

(I) = e

1

(I) −e

1

(Q) ≤ λ(R/(Q : I))λ(I/Q) = λ(R/(Q : I))e

1

(I).

For example, if R is a Cohen–Macaulay local ring that is not regular and I = Q : m, then

e

1

(I) = e

1

(I) −e

1

(Q) ≤ λ(R/(Q : I))(ν(I) −d) ≤ λ(I/Q) = e

1

(I),

which is a case when the equality in Theorem 2.6 holds true.

Example 2.12. Let R be a Cohen–Macaulay local ring of dimension 1 with e

0

(R) = 2. For every

m–primary ideal I, there exists a ∈ I such that I

2

= aI ([SV, Theorem 2.5]). Hence

e

1

(I) = λ(R/(aR : I)) [ν(I) −1] .

Example 2.13. Let a and ℓ be integers such that a ≥ 4 and ℓ ≥ 2. Let H be the numerical

semigroup generated by a, aℓ − 1, ¦aℓ + i¦

1≤i≤a−3

, and put R = k[[t

a

, t

aℓ−1

, ¦t

aℓ+i

¦

1≤i≤a−3

]] in

the formal power series ring k[[t]] over a ﬁeld k. Let I = (t

2aℓ−a−1

, ¦t

3aℓ−2a−1−i

¦

1≤i≤a−3

) R and

Q = (t

2aℓ−a−1

) ⊂ I. Then I = ω

R

is a canonical ideal of R and Q is a reduction of I. We have

mI ⊆ Q and e

1

(I) = λ(R[t

aℓ−a−i

[ 1 ≤ i ≤ a −3]/R) = a −2 = r(R)

([GMP, Lemma 2.1]), where r(R) is the Cohen–Macaulay type. Hence

a −3 = r(R) −1 = ν(I/Q) = λ(I/Q) < e

1

(I) = a −2.

Since r(R) ≥ 2, the ring R is not Gorenstein. We have e

1

(I) = λ(I/Q) + 1, so that

e

1

(I) = e

0

(I) −λ(R/I) + 1.

Therefore, thanks to [S2], we get I

3

= QI

2

(hence red

Q

(I) = 2) and

VARIATION OF HILBERT COEFFICIENTS 9

S

Q

(I) ≃ B(−1)

as graded !(Q)–modules, where !(Q) denotes the Rees algebra of Q, S

Q

(I) the Sally module of I

with respect to Q, and B = !(Q)/m!(Q). We have

a −2 = e

1

(I) ≤ λ(R/(Q : I))

__

m + s

s

_

−1

_

=

_

a −1

2

_

−1,

since s = 2 and m = a − 3. The equality e

1

(I) = λ(R/(Q : I))

__

m+s

s

_

−1

¸

holds if and only if

a = 4. When this is the case, we have H = ¸4, 4ℓ −1, 4ℓ + 1¸.

3. The reduction number formula

In order to make use of Theorem 2.6, we need information about the reduction number of I in

terms related to multiplicity. Let us recall [V1, Theorem 2.45]:

Theorem 3.1. Let (R, m) be a Cohen-Macaulay local ring of dimension d ≥ 1 and inﬁnite residue

ﬁeld. For an m-primary ideal I,

red(I) ≤ max

_

de

0

(I)

o(I)

−2d + 1 , 0

_

where o(I) is the m-adic order of I.

To establish such a result for arbitrary Noetherian local rings, we proceed diﬀerently. The version

of the following lemma for Cohen-Macaulay rings can be found in [S1, Chapter 3, Theorem 1.1].

Lemma 3.2. Let (R, m) be a Noetherian local ring of dimension 1. Let x be a parameter of R.

Let E be a ﬁnitely generated R–module and U an R–submodule of E. Then we have the following.

(a) ν(U) ≤ λ(R/(x))ν(E). Hence ν(I) ≤ λ(R/(x)) for every ideal I of R.

(b) If R is Cohen–Macaulay and x belongs to m

s

, then ν(U) ≤

λ(R/(x))

s

ν(E).

Proof. (a) Let W = H

0

m

(E), E

′

= E/W, and U

′

= (U + W)/W. Then E

′

is a Cohen-Macaulay

R–module of dimension 1 and x is E

′

–regular. Moreover,

λ(U

′

/xU

′

) = e

0

((x), U

′

) ≤ e

0

((x), E

′

) = λ(E

′

/xE

′

).

Consider the following two short exact sequences:

0 −−−−→ W −−−−→ E −−−−→ E

′

−−−−→ 0

0 −−−−→ U ∩ W −−−−→ U −−−−→ U

′

−−−−→ 0

10 LAURA GHEZZI, SHIRO GOTO, JOOYOUN HONG, AND WOLMER V. VASCONCELOS

Then we obtain

ν(U) ≤ λ(U/xU) = λ(U

′

/xU

′

) + λ((U ∩ W)/x(U ∩ W))

= λ(U

′

/xU

′

) + λ((0 :

U∩W

x))

≤ λ(U

′

/xU

′

) + λ((0 :

W

x))

= λ(U

′

/xU

′

) + λ(W/xW)

≤ λ(E

′

/xE

′

) + λ(W/xW)

= λ(E/xE)

≤ λ(R/(x))ν(E).

(b) We may assume that the ﬁeld R/m is inﬁnite. Let yR be a minimal reduction of m. Then since

x ∈ m

s

⊆ y

s

R, we get

λ(R/(x)) = e

0

(xR) ≥ e

0

(y

s

R) = e

0

(y

s

R) = se

0

(yR) = sλ(R/(y)).

Hence λ(R/(y)) ≤

λ(R/(x))

s

, so that

ν(U) ≤ λ(R/(y))ν(E) ≤

λ(R/(x))

s

ν(E).

✷

Theorem 3.3. Let (R, m) be a Noetherian local ring of dimension d ≥ 1 with inﬁnite residue ﬁeld.

For an m-primary ideal I and a minimal reduction J of I, there exists a minimal reduction Q of I

such that

red

Q

(I) ≤ max¦dλ(R/J) −2d + 1, 0¦.

Proof. Let us start with a minimal reduction J = (x

1

, . . . , x

d

) of I. Let L = (x

1

, . . . , x

d−1

). Then

ν(I

n

) ≤ ν(L

n

) + ν(I

n

/L

n

).

We need to estimate ν(I

n

/L

n

). Set

M

i

=

(I

n

+ L

n−i+1

) ∩ L

n−i

L

n−i+1

and N

i

=

I

n

+ L

n−i+1

L

n−i+1

.

Then we obtain the following series of exact sequences :

0 −→M

i

−→N

i

−→N

i+1

−→0,

where i = 1, . . . , n−1. Note that for each i, M

i

is a submodule of L

n−i

/L

n−i+1

as an R/L–module.

Hence by Lemma 3.2, for each i = 1, . . . , n −1,

ν(M

i

) ≤ λ(R/J)ν(L

n−i

/L

n−i+1

) = λ(R/J)

_

d + n −2 −i

d −2

_

.

VARIATION OF HILBERT COEFFICIENTS 11

Since N

n

=

I

n

+ L

L

is a submodule of R/L, by Lemma 3.2, we get

ν(N

n

) ≤ λ(R/J).

Therefore we obtain

ν(I

n

) ≤ ν(L

n

) + ν(I

n

/L

n

)

≤ ν(L

n

) + ν(M

1

) + ν(M

2

) + + ν(M

n−1

) + ν(N

n

)

≤

_

d + n −2

d −2

_

+ λ(R/J)

n

i=1

_

d + n −2 −i

d −2

_

=

_

d + n −2

d −2

_

+ λ(R/J)

_

d + n −2

d −1

_

.

Recall that if

ν(I

n

) <

_

n + d

d

_

,

then there is a minimal reduction Q of I such that red

Q

(I) ≤ n − 1 ([ES], [V1, Theorem 2.36]).

Hence by solving the inequality

_

d + n −2

d −2

_

+ λ(R/J)

_

d + n −2

d −1

_

<

_

n + d

d

_

,

we obtain the desired relation. ✷

Corollary 3.4. Let (R, m) be a Noetherian local ring of dimension d ≥ 1 and inﬁnite residue ﬁeld.

Let Q be a minimal reduction of m such that red

Q

(m) = red(m). Then

e

1

(m) ≤ e

1

(m) −e

1

(Q) ≤ λ(R/(Q : m))

__

ν(m) + λ(R/Q)d −3d + 1

ν(m) −d

_

−1

_

.

Remark 3.5. It is worthwhile to point out that there are other known bounds for the reduction

number of an ideal in terms of some of its Hilbert coeﬃcients. One of these is a bound proved by

M. E. Rossi ([R, Corollary 1.5]): If (R, m) is a Cohen-Macaulay local ring of dimension at most 2

then for any m–primary ideal I with a minimal reduction Q

red

Q

(I) ≤ e

1

(I) −e

0

(I) + λ(R/I) + 1.

Several open questions arise. Foremost whether it extends to higher dimensional Cohen-Macaulay

rings (with a correction term depending on the dimension). Another question is which oﬀsetting

terms should be added in the non Cohen-Macaulay case. For instance, in dimension 2 whether the

addition of −e

1

(Q), a term that can be considered a non Cohen–Macaulayness penalty, would give

a valid bound.

12 LAURA GHEZZI, SHIRO GOTO, JOOYOUN HONG, AND WOLMER V. VASCONCELOS

4. Normalization

The following observation shows how the special ﬁber of the normalization impacts e

0

(I). Of

course, more interesting issue would be to obtain relationships going the other way.

Proposition 4.1. Let (R, m) be a normal local domain and let I be an m-primary ideal. Suppose

that ! =

∞

n=0

C

n

is ﬁnite over ! = !(I). We denote by f

0

(I) the multiplicity of !/m!. Then

e

0

(I) ≤ min¦f

0

(I)λ(R/I), f

0

(I)λ(R/I)¦.

Proof. We ﬁrst observe that C

n+1

= IC

n

= IC

n

, for n ≫ 0. In particular, in that range,

C

n+1

⊂ mC

n

. Consider now the corresponding exact sequence

0 →mC

n

/C

n+1

−→C

n

/C

n+1

−→C

n

/mC

n

→0.

Counting multiplicities, we have

e

0

(I) ≤ deg(m!/I !) + deg(!/m!) ≤ f

0

(I)(λ(m/I) + 1) = f

0

(I)λ(R/I)

as desired. The other inequality, e

0

(I) ≤ f

0

(I)λ(R/I), has a similar proof. ✷

References

[C] A. Corso, Sally modules of m-primary ideals in local rings, Comm. Algebra 37 (2009), 4503–4515.

[ES] P. Eakin and A. Sathaye, Prestable ideals, J. Algebra 41 (1976), 439–454.

[E1] J. Elias, On the ﬁrst normalized Hilbert coeﬃcient, J. Pure and Applied Algebra 201 (2005), 116–125.

[E2] J. Elias, Upper bounds of Hilbert coeﬃcients and Hilbert functions, Math. Proc. Camb. Phil. Soc. 145 (2008),

87–94.

[GhGHOPV] L. Ghezzi, S. Goto, J. Hong, K. Ozeki, T.T. Phuong and W. V. Vasconcelos, Cohen–Macaulayness

versus the vanishing of the ﬁrst Hilbert coeﬃcient of parameter ideals, J. London Math. Soc. 81 (2010), 679–695.

[GhHV] L. Ghezzi, J. Hong and W. V. Vasconcelos, The signature of the Chern coeﬃcients of local rings, Math.

Research Letters 16 (2009), 279–289.

[GHM] S. Goto, J. Hong and M. Mandal, The positivity of the ﬁrst normalized Hilbert coeﬃcients, Proc. Amer.

Math. Soc. 139 (2011), 2399–2406.

[GMP] S. Goto, N. Matsuoka and T. T. Phuong, Almost Gorenstein rings; arXiv:1106.1301v2 [Math.AC].

[GO] S. Goto and K. Ozeki, Buchsbaumness in local rings possessing constant ﬁrst Hilbert coeﬃcients of parameters,

Nagoya Math. J. 199 (2010), 95–105.

[HH] K. Hanumanthu and C. Huneke, Bounding the ﬁrst Hilbert coeﬃcient, Proc. Amer. Math. Soc. 140 (2012),

109-117.

[Hu] C. Huneke, Hilbert functions and symbolic powers, Michigan Math. J. 34 (1987), 293–318.

[K] D. Kirby, A note on superﬁcial elements of an ideal of a local ring, Q. J. Math. Oxford 14 (1963), 21–28.

[MV] M. Mandal and J. K. Verma, On the Chern number of an ideal, Proc. Amer. Math. Soc. 138 (2010), 1995–1999.

[MSV] M. Mandal, B. Singh and J. K. Verma, On some conjectures about the Chern numbers of ﬁltrations, J. Algebra

325 (2011), 147–162.

[PUV] C. Polini, B. Ulrich and W. V. Vasconcelos, Normalization of ideals and Brian¸ con-Skoda numbers, Math.

Research Letters 12 (2005), 827–842.

[R] M. E. Rossi, A bound on the reduction number of a primary ideal, Proc. Amer. Math. Soc. 128 (2000), 1325–1332.

[RV1] M. E. Rossi and G. Valla, The Hilbert function of the Rattliﬀ-Rush ﬁltration, J. Pure and Applied Algebra

201 (2005), 24–41.

[RV2] M. E. Rossi and G. Valla, On the Chern number of a ﬁltration, Rendiconti Seminario Matematico Padova 121

(2009), 201–222.

[S1] J. D. Sally, Numbers of Generators of Ideals in Local Rings, Lecture Notes in Pure and Applied Mathematics

36, Marcel Dekker, New York, 1978.

[S2] J. D. Sally, Hilbert coeﬃcients and reduction number 2, J. Algebraic Geometry 1 (1992), 325–333.

[SV] J. D. Sally and W. V. Vasconcelos, Stable rings, J. Pure and Applied Algebra 4 (1974), 319–336.

VARIATION OF HILBERT COEFFICIENTS 13

[V1] W. V. Vasconcelos, Integral Closure, Springer Monographs in Mathematics, Springer, Heidelberg, 2005.

[V2] W. V. Vasconcelos, The Chern coeﬃcients of local rings, Michigan Math. J. 57 (2008), 725–743.

Department of Mathematics, New York City College of Technology-Cuny, 300 Jay Street, Brook-

lyn, NY 11201, U.S.A.

E-mail address: lghezzi@citytech.cuny.edu

Department of Mathematics, School of Science and Technology, Meiji University, 1-1-1 Higashi-

mita, Tama-ku, Kawasaki 214-8571, Japan

E-mail address: goto@math.meiji.ac.jp

Department of Mathematics, Southern Connecticut State University, 501 Crescent Street, New

Haven, CT 06515-1533, U.S.A.

E-mail address: hongj2@southernct.edu

Department of Mathematics, Rutgers University, 110 Frelinghuysen Rd, Piscataway, NJ 08854-8019,

U.S.A.

E-mail address: vasconce@math.rutgers.edu

the issues become less structured since the values of e1 (I) may be negative. RV2]). [GO]. an uniform bound for e1 (I) ﬁrst appeared for rings of dimension 1 in the work of D. In fact. E2]) and M. . JOOYOUN HONG. the ﬁrst bound asserts that if I is minimally generated by m = ν(I) elements. AND WOLMER V. [GhHV]. In particular. 2 2 e0 (I) − s . (ii) Since e1 (I) ≤ e1 (I) := e1 (R).2 LAURA GHEZZI. 2 The other bound uses the m-adic order of I. As a basic source. essentially of ﬁnite type over a perfect ﬁeld. Let (R. For an d-dimensional Cohen-Macaulay local ring and an m-primary ideal I. one has e1 (I) ≥ 0 ([GHM]). Then t (d − 1)e0 (I) + e0 ((I + δR)/δR) . we single out two developed in the work of J. m) be a reduced Cohen-Macaulay local ring of dimension d. Among the formulas which more directly inﬂuenced the authors here. Let δ be a regular element of the Jacobian ideal of R. An important case is e0 (J). that if I ⊂ ms and I = ms . Elias ([E1. VASCONCELOS (i) For Cohen-Macaulay rings. Hanumanthu and C. I = (J. It also treats more general ﬁltrations which were helpful to us. x). e1 (I). E. [V2]). quadratic bounds of this type were developed for arbitrary m-primary ideals in all dimensions by several authors. (iv) We shall now outline the main results of this note. Huneke ([HH]) brought a new parameter to bear on these formulas with their proof that e0 (I) − k e1 (I) ≤ . Buchsbaum. e1 (I) ≤ e1 (I) ≤ t+1 where t is the Cohen-Macaulay type of R. using the values of e1 (I) for ideals generated by systems of parameters led to the characterization of several properties (Cohen-Macaulay. e1 (I) ≤ e0 (I) m−d − − λ(R/I) + 1.) Sections 2 and 3 are organized around a list of questions about the changes that e0 (I) and e1 (I) undergo when I varies. [RV2] has a systematic development of these formulas along with a comprehensive bibliography. 2 (iii) When R is not Cohen-Macaulay. SHIRO GOTO. K. but not necessarilly Cohen-Macaulay. An upper bound for e1 (I) (see [PUV] for other bounds) is the following. 2 where k is the maximal length of chains of integrally closed ideals between I and m. ﬁnite cohomology) of the ring R itself (see [GhGHOPV]. then e1 (I) ≤ Recently. if R is a regular local ring e1 (I) ≤ (d − 1)e0 (I) . Rossi and G. and let I be an m-primary ideal. e0 (m) e1 (m) ≤ . 2 Progressively. A baseline is the fact that when R is analytically unramiﬁed. bounds with a diﬀerent character arise. Valla ([RV1. [MV]. Kirby ([K]). e1 (J) −→ e0 (I).(We refer to [V1] for basic deﬁnitions and Rees algebras theory.

but we will consider very general cases. If n = 1. then M ≃ R/m and the assertion is clear. h) be m–primary ideals of R. .2. Lemma 2. . let (R. so that λ(M ⊗ N ) ≤ λ(L ⊗ N ) + λ((M/L) ⊗ N ) ≤ (1 + (n − 1))·ν(N ) = λ(M )ν(N ). m) be a Noetherian local ring of dimension d and let J ⊂ I = (J. . Suppose that n ≥ 2 and choose an R–submodule L of M with λ(L) = 1.VARIATION OF HILBERT COEFFICIENTS 3 Clearly the optimal baseline is that of an ideal J generated by a system of parameters.6 asserts that e1 (I) − e1 (J) ≤ λ(R/(J : I))· m+s − 1 ·f0 (J). Let (R. Proof. the induction hypothesis shows λ((M/L) ⊗ N ) ≤ (n − 1)ν(N ). As will be seen. Upper bounds for the variations of e0 (I) and e1 (I) In our calculations we make repeated use of the following elementary observation.1. Then e0 (J) − e0 (I) ≤ λ(R/(J : I))·f0 (J). and vanishes when R is Cohen–Macaulay. Then Theorem 2. we address the need to link the value of redJ (I) to other properties of J. 2. . m) is a Noetherian local ring and M is an R–module of ﬁnite length λ(M ). e1 (J) is always non-positive. where ν(N ) denotes the minimal number of generators for N . When J is a minimal reduction of I.3: Let (R. For an m-primary ideal I and a minimal reduction J of I. . according to [MSV]. but we give a general formulation in Theorem 3. s where f0 (J) is the multiplicity of the special ﬁber of R(J) = n≥0 J n . In Section 3. By tensoring 0 → L → M → M/L → 0 with N . Induct on n = λ(M ). This is a well-known fact when R is a Cohen-Macaulay ring. To describe one of these estimates. 0}. we get the exact sequence L ⊗ N → M ⊗ N → (M/L) ⊗ N → 0. let J be an m-primary ideal and let I = (J. there exists a minimal reduction Q of I such that redQ (I) ≤ max{d·λ(R/J) − 2d + 1. some relationships involve the multiplicity f0 (J) of the special ﬁber. ✷ Theorem 2. Since λ(M/L) = n − 1. m) be a Noetherian local ring of dimension d ≥ 1 and inﬁnite residue ﬁeld. m) be a Noetherian local ring of dimension d ≥ 1. In fact. h1 . hm ) be integral over J of reduction number s = redJ (I). If (R. for unmixed local rings the vanishing characterizes Cohen-Macaulayness ([GhGHOPV]). We add a word of warning in reading some of the formulas with terms like e1 (I) − e1 (J). then λ(M ⊗ N ) ≤ λ(M )·ν(N ) for every ﬁnitely generated R–module N .

J n−r hr ) ⊂ · · · ⊂ Mn = (Mn−1 . J n−r hr ) ⊂ · · · ⊂ Ms = (Ms−1 . Also. Proof. for n ≫ 0. JOOYOUN HONG. Then hs+1 ∈ JI s . AND WOLMER V. m) be a Noetherian local ring of dimension d ≥ 1 and let J ⊂ I = (J. λ(R/J n ) − λ(R/I n ) is the diﬀerence of two polynomials of degree d and leading (binomial) coeﬃcients e0 (J) and e0 (I). For n ≥ s. where redJ (I) is the reduction number of I with respect to J. VASCONCELOS Proof. J n−r+1 hr−1 ) ⊂ Mr = (Mr−1 . Mr /Mr−1 is generated by the image of hr J n−r + Mr−1 . Let s = redJ (I). Let (R. then e1 (I) − e1 (J) ≤ redJ (I)·λ(R/(J : I))·f0 (J). For n ∈ N. Therefore s s λ(R/J n ) − λ(R/I n ) = λ(I n /J n ) = r=1 λ(Mr /Mr−1 ) ≤ λ(R/(J : I)) r=1 ν(J n−r ). Consider the natural surjection ζ : R/(J : h) ⊗ J n−r ։ Mr /Mr−1 = (hr J n−r + Mr−1 )/Mr−1 . we obtain the following ﬁltration: J n = M0 ⊂ M1 = (M0 . Then we obtain n λ(R/J ) − λ(R/I ) = λ(I /J ) = λ(Mn /M0 ) = r=1 n n n n λ(Mr /Mr−1 ). J n−1 h) ⊂ · · · ⊂ Mr−1 = (Mr−2 .3. If h is integral over J. The iterated Hilbert function r=0 ν(J r ) is of polynomial type of degree d with leading (binomial) coeﬃcient f0 (J). Using Lemma 2. . we have λ(Mr /Mr−1 ) ≤ λ(R/(J : h) ⊗ J n−r ) ≤ λ(R/(J : h))·ν(J n−r ).1. consider the following ﬁltration: J n = M0 ⊂ M1 = (M0 . SHIRO GOTO. Hence e0 (J) − e0 (I) ≤ λ(R/(J : I))·f0 (J). J n−1 h) ⊂ · · · ⊂ Mr = (Mr−1 . For each r.4 LAURA GHEZZI. It follows that n−1 λ(R/J n ) − λ(R/I n ) ≤ λ(R/(J : I))· r=0 n−1 ν(J r ). ✷ Theorem 2. J n−s hs ) = I n . hn ) = I n . h) be m–primary ideals.

z] be the polynomial ring over an inﬁnite ﬁeld k. y b−β . The ﬁrst assertion follows from e1 (Q) ≤ 0 [MSV] and f0 (Q) = 1 for every parameter ideal Q. Then it is enough to show that λ(R/(Q : I)) = e0 (I) − λ(R/I). m) be a Noetherian local ring of dimension d ≥ 1 and inﬁnite residue ﬁeld.VARIATION OF HILBERT COEFFICIENTS 5 Now for n ≫ 0. then e1 (I) ≤ redQ (I)·(e0 (I) − λ(R/I)). ([V1. Example 7.y.36]) Let k[x. On the other hand. z c ) ⊂ (J. This follows from λ(R/(Q : I)) = λ(R/Q) − λ((Q : I)/Q) = e0 (Q) − λ((Q : I)/Q) = e0 (Q) − λ(R/I) because (Q : I)/Q is the canonical module of R/I.4. This inequality ensures that h = xα y β z γ ∈ J. Then e1 (I) ≤ redQ (I)·λ(R/(Q : I)). h) be m–primary ideals such that Q is a minimal reduction of I. therefore it is at most a polynomial of degree d − 1 and leading coeﬃcient e1 (I) − e1 (J). Suppose that R is Gorenstein. Moreover. b c a + + −1 α β γ . where α β γ + + < 1. z c−γ ). ✷ Corollary 2. Let R = k[x. Let Q ⊂ I = (Q. Proof. d−i−1 which proves that e1 (I) − e1 (J) ≤ redJ (I)·λ(R/(J : I))·f0 (J).z) and let J and I be R–ideals such that J = (xa . y. z](x. a b c Since (J : I) = (J : xα y β z γ ) = (xa−α . xα y β z γ ) = I. y b . if R is Gorenstein. Let (R. λ(R/J n ) − λ(R/I n ) is the diﬀerence of two polynomials of degree d and with same leading (binomial) coeﬃcients e0 (J) and e0 (I). Then we have / a b c α β γ e0 (J) − e0 (I) = abc − (abγ + bcα + acβ) = abc 1 − − − . ✷ Example 2. y. for n ≫ 0. we have s d−1 λ(R/(J : h))· r=1 ν(J n−r ) ≤ λ(R/(J : h))·s· i=0 (−1)i fi (J) n+d−i−2 .5. we obtain λ(R/(J : I))·f0 (J) = (a − α)(b − β)(c − γ) = abc − bcα − acβ − abγ + aβγ + bαγ + cαβ − αβγ = e0 (J) − e0 (I) + αβγ > e0 (J) − e0 (I).

. y b − z c . H)n /J n ) = λ((J n . if I = (J. Note that Mr /Mr−1 is generated by the image of H r J n−r . . c > 3γ. H). . . We have already given parts of the proof. . h1 . VASCONCELOS Let Q = (xa − z c . then Mr /Mr−1 is generated by batches of elements. . diﬃcult to control. s ✷ . xα y β z γ ) and suppose that a > 3α. Let J be an m–primary ideal and H = (h1 . More precisely. HJ n−1 . We note that this module is annihilated by J : I and is generated by the ‘monomials’ in the hi of degree r. hm ) be integral over J of reduction number s = redJ I. . AND WOLMER V. . the method of iterating the assertion in Theorem 2.6 LAURA GHEZZI. our formulation using the ﬁltration above wraps it diﬀerently to accommodate our data. . Therefore for n ≫ 0. H n−1 J. . . .3 tends to induce a bigger upper bound for e1 (I) − e1 (J) than necessary. where br = m+r−1 . . hm ) a set of elements integral over J. Note that I = (Q. The remaining part is to estimate the growth of the length of Mr /Mr−1 = [(h1 . .6. . HJ n−1 . Write I = (J. Then Q is a minimal reduction of I and the reduction number redQ (I) ≤ 2. . We can estimate e1 (I): e1 (I) = e1 (I) − e1 (Q) ≤ 2λ(R/(Q : I)). . z c ). . Theorem 2. m) be a Noetherian local ring of dimension d ≥ 1. b > 3β. SHIRO GOTO. . H r−1 J n−r+1 . H r J n−r ). m+s −1 . JOOYOUN HONG. hm ).3. s Proof. where ν(H) = ν(I/J). . Then e1 (I) − e1 (J) ≤ λ(R/(J : I))· m+s − 1 ·f0 (J). with coeﬃcients in J n−r . . and consider the diﬀerence of Hilbert functions λ(R/J n ) − λ(R/I n ) = λ((J. H n )/J n ) n = r=1 λ(Mr /Mr−1 ). There is a natural surjection Φ : R/(J : I) ⊗ Rbr ⊗ J n−r −→ Mr /Mr−1 . . This ﬁltration has been used by several authors when J is generated by a system of parameters. As ν(I n /J n ) is increasing. . ✷ Now we treat a general case of Theorem 2. hm )r J n−r + Mr−1 ] /Mr−1 . Instead. where Mr = (J n . Let (R. h1 . let J be an m-primary ideal and let I = (J. . r s λ(R/J n ) − λ(R/I n ) = r=1 s λ(Mr /Mr−1 ) m+r−1 r ≤ r=1 λ(R/(J : I))·ν(J n )· = λ(R/(J : I))·ν(J n )· which completes the proof.

I) = e1 (I) − e1 (Q) − e0 (I) + λ(R/I). then the multiplicity s0 (Q. Corollary 2. then λ(R/(J : I)) does not depend on J either. let J be an m-primary ideal and let I = (J. where Hm (I) is the m–th Koszul homology of I.7. we obtain J n /mJ n −→ I n /mI n −→ (I n /J n ) ⊗ R/m → 0. if R is Cohen–Macaulay. .6. then it is well–known that m = ν(I) − ν(J) does not depend on J because mI ∩ J = mJ.8]. we obtain f0 (I) − f0 (J) ≤ e1 (I) − e1 (J) ≤ λ(R/(J : I))· which completes the proof.6. using Lemma 2. according to [C. Let I be an m–primary ideal and Q a minimal reduction of I. Using Theorem 2. I) ≤ −e0 (I) + λ(R/I) + λ(R/(Q : I))· where s = redQ (I) is the reduction number. because λ(R/(J : I)) = e0 (I) − λ(Hm (I)). Then f0 (I) ≤ 1 + λ(R/(J : I))· m+s −1 s ·f0 (J). I) of the Sally module SQ (I) is s0 (Q. Proof. The values of the ﬁrst Hilbert coeﬃcients are also related to the multiplicity of certain Sally modules. m) be a Noetherian local ring of dimension d ≥ 1. Let I be an m–primary ideal and Q a minimal reduction of I. Moreover. Proposition 2. Let (R. Then the multiplicity s0 (Q. Therefore. I) of the Sally module SQ (I) satisﬁes s0 (Q. .8. if J is a minimal reduction of I. . h1 . hm ) be integral over J of reduction number s = redJ I.1. Suppose that dim(SQ (I)) = d and 0 that Hm (R) ⊂ I.VARIATION OF HILBERT COEFFICIENTS 7 In Theorem 2. s . m) be a Noetherian local ring of dimension d ≥ 1 with inﬁnite residue ﬁeld. Let (R. This induces the inequalities of the leading coeﬃcients (in degree d − 1) f0 (I) − f0 (J) ≤ e1 (I) − e1 (J). Proposition 2. Note that the formulas for the variations of e1 (I) and f0 (I) require that the ideal I has the same integral closure as J. . m+s − 1 ·f0 (J). Let (R. ν(I) − d + s −1 . s ✷ Remark 2. m) be a Noetherian local ring of dimension d ≥ 1 with inﬁnite residue ﬁeld.9. we get λ(I n /mI n ) − λ(J n /mJ n ) ≤ λ((I n /J n ) ⊗ R/m) ≤ λ(I n /J n ) = λ(R/J n ) − λ(R/I n ). If 0 dim(SQ (I)) = d and Hm (R) ⊂ I. By tensoring the following exact sequence with R/m 0 −→ J n −→ I n −→ I n /J n −→ 0.

there exists a ∈ I such that I 2 = aI ([SV. VASCONCELOS Example 2. . Then since Q ⊆ J. JOOYOUN HONG. We have e1 (I) = λ(I/Q) + 1. Using e1 (I) = e1 (mn ) = n(n − 1) and 2 e1 (I) − e1 (J) ≤ λ(R/(J : an+1 ))f0 (J). a2 ) is a minimal reduction of I.10. We assume that Q = (a1 . Remark 2. Then I = ωR is a canonical ideal of R and Q is a reduction of I. Then by Theorem 2. Since r(R) ≥ 2. Theorem 2. For every m–primary ideal I. we get I 3 = QI 2 (hence redQ (I) = 2) and . . {t3aℓ−2a−1−i }1≤i≤a−3 ) R and Q = (t2aℓ−a−1 ) ⊂ I. y](x. . Let R = k[x. it follows that e1 (I) = e1 (I) − e1 (Q) ≤ λ(R/(Q : I))·λ(I/Q) = λ(R/(Q : I))·e1 (I). Suppose that I 2 = QI. Let I be an m–primary ideal and Q a minimal reduction of I. Example 2. {taℓ+i }1≤i≤a−3 ]] in the formal power series ring k[[t]] over a ﬁeld k. (Reduction number one) Let (R. or more generally I = J : ms for some values of s. . . Therefore. and put R = k[[ta .12. AND WOLMER V. so that e1 (I) = e0 (I) − λ(R/I) + 1. an+1 ) for some n ≥ 2. where r(R) is the Cohen–Macaulay type. the ring R is not Gorenstein. We have mI ⊆ Q and e1 (I) = λ(R[taℓ−a−i | 1 ≤ i ≤ a − 3]/R) = a − 2 = r(R) ([GMP. because I = J and redQ (I) = 1. if R is a Cohen–Macaulay local ring that is not regular and I = Q : m. m) be a Noetherian local ring of dimension d ≥ 1 with inﬁnite residue ﬁeld.6 we get e1 (I) − e1 (Q) ≤ λ(R/(Q : I))·(ν(I) − d) ≤ λ(R/(Q : I))·λ(I/Q). SHIRO GOTO. . then e1 (I) = e1 (I) − e1 (Q) ≤ λ(R/(Q : I))·(ν(I) − d) ≤ λ(I/Q) = e1 (I). we obtain 1 e1 (J) ≥ n(n − 1) − λ(R/(J : an+1 ))f0 (J). Hence e1 (I) = λ(R/(aR : I))· [ν(I) − 1] . Let R be a Cohen–Macaulay local ring of dimension 1 with e0 (R) = 2. an . y) and I = mn = (a1 .13. Let m = (x.11. Let J = (a1 . Then since e1 (I) = e0 (I) − λ(R/I) = λ(I/Q) ([Hu. an ). 2. 2 One situation that may be amenable to further analysis is when I = J : m. Example 2. y] denotes the polynomial ring over an inﬁnite ﬁeld k. Suppose that R is Cohen–Macaulay.5]).y) where k[x. thanks to [S2]. Hence a − 3 = r(R) − 1 = ν(I/Q) = λ(I/Q) < e1 (I) = a − 2. Let H be the numerical semigroup generated by a.1]). Let a and ℓ be integers such that a ≥ 4 and ℓ ≥ 2. . taℓ−1 . which is a case when the equality in Theorem 2. For example. a2 .8 LAURA GHEZZI. We refer to I as a socle extension of J. Let I = (t2aℓ−a−1 . aℓ − 1. .6 holds true. I is integral over J with 1 redJ (I) = 1. {aℓ + i}1≤i≤a−3 . Lemma 2.1]).

1]. To establish such a result for arbitrary Noetherian local rings.1. we proceed diﬀerently. Let (R. E ′ ) = λ(E ′ /xE ′ ). s Proof. 4ℓ − 1. U ′ ) ≤ e0 ((x). Theorem 1. − 1 holds if and only if 3. Theorem 2. The version of the following lemma for Cohen-Macaulay rings can be found in [S1. Let (R.45]: Theorem 3. The reduction number formula In order to make use of Theorem 2. we have H = 4. we need information about the reduction number of I in terms related to multiplicity. (a) Let W = H0 (E). then ν(U ) ≤ ·ν(E). Consider the following two short exact sequences: 0 −−→ −− W − − → E − − → E′ − − → 0 −− −− −− d·e0 (I) − 2d + 1 .2. Let us recall [V1. λ(R/(x)) (b) If R is Cohen–Macaulay and x belongs to ms . Let E be a ﬁnitely generated R–module and U an R–submodule of E. 0 o(I) 0 −−→ U ∩ W −−→ U −−→ U′ −−→ 0 −− −− −− −− . Let x be a parameter of R. Lemma 3. red(I) ≤ max where o(I) is the m-adic order of I. Moreover. When this is the case. Hence ν(I) ≤ λ(R/(x)) for every ideal I of R. Then E ′ is a Cohen-Macaulay m R–module of dimension 1 and x is E ′ –regular.6. where R(Q) denotes the Rees algebra of Q.VARIATION OF HILBERT COEFFICIENTS 9 SQ (I) ≃ B(−1) as graded R(Q)–modules. m) be a Noetherian local ring of dimension 1. E ′ = E/W . Then we have the following. (a) ν(U ) ≤ λ(R/(x))·ν(E). The equality e1 (I) = λ(R/(Q : I)) a = 4. and B = R(Q)/mR(Q). 4ℓ + 1 . 2 m+s s a − 2 = e1 (I) ≤ λ(R/(Q : I)) since s = 2 and m = a − 3. Chapter 3. and U ′ = (U + W )/W . We have m+s −1 = s a−1 − 1. m) be a Cohen-Macaulay local ring of dimension d ≥ 1 and inﬁnite residue ﬁeld. SQ (I) the Sally module of I with respect to Q. λ(U ′ /xU ′ ) = e0 ((x). For an m-primary ideal I.

Hence λ(R/(y)) ≤ λ(R/(x)) . . . xd ) of I. Ln−i+1 Ln−i+1 Then we obtain the following series of exact sequences : Mi = 0 −→ Mi −→ Ni −→ Ni+1 −→ 0. Mi is a submodule of Ln−i /Ln−i+1 as an R/L–module. Let us start with a minimal reduction J = (x1 . for each i = 1. . For an m-primary ideal I and a minimal reduction J of I. . Let L = (x1 . Proof. n − 1. s ✷ ν(U ) ≤ λ(R/(y))·ν(E) ≤ Theorem 3. Let yR be a minimal reduction of m. . . Set (I n + Ln−i+1 ) ∩ Ln−i I n + Ln−i+1 and Ni = . xd−1 ). VASCONCELOS Then we obtain ν(U ) ≤ λ(U/xU ) = λ(U ′ /xU ′ ) + λ((U ∩ W )/x(U ∩ W )) = λ(U ′ /xU ′ ) + λ((0 :U ∩W x)) ≤ λ(U ′ /xU ′ ) + λ((0 :W x)) = λ(U ′ /xU ′ ) + λ(W/xW ) ≤ λ(E ′ /xE ′ ) + λ(W/xW ) = λ(E/xE) ≤ λ(R/(x))·ν(E). ν(Mi ) ≤ λ(R/J)·ν(Ln−i /Ln−i+1 ) = λ(R/J)· d+n−2−i . . there exists a minimal reduction Q of I such that redQ (I) ≤ max{d·λ(R/J) − 2d + 1. Hence by Lemma 3. Let (R. (b) We may assume that the ﬁeld R/m is inﬁnite. .10 LAURA GHEZZI. . SHIRO GOTO. 0}. . . s so that λ(R/(x)) ·ν(E). . m) be a Noetherian local ring of dimension d ≥ 1 with inﬁnite residue ﬁeld. n − 1. we get λ(R/(x)) = e0 (xR) ≥ e0 (y s R) = e0 (y s R) = s·e0 (yR) = s·λ(R/(y)).2. Then ν(I n ) ≤ ν(Ln ) + ν(I n /Ln ). . . . Then since x ∈ ms ⊆ y s R.3. where i = 1. . We need to estimate ν(I n /Ln ). Note that for each i. AND WOLMER V. d−2 . JOOYOUN HONG.

Rossi ([R. ν(m) − d Remark 3. Foremost whether it extends to higher dimensional Cohen-Macaulay rings (with a correction term depending on the dimension). d ✷ Corollary 3. < n+d . Let Q be a minimal reduction of m such that redQ (m) = red(m). would give a valid bound. For instance. m) is a Cohen-Macaulay local ring of dimension at most 2 then for any m–primary ideal I with a minimal reduction Q redQ (I) ≤ e1 (I) − e0 (I) + λ(R/I) + 1. Another question is which oﬀsetting terms should be added in the non Cohen-Macaulay case. m) be a Noetherian local ring of dimension d ≥ 1 and inﬁnite residue ﬁeld. It is worthwhile to point out that there are other known bounds for the reduction number of an ideal in terms of some of its Hilbert coeﬃcients. in dimension 2 whether the addition of −e1 (Q).4. a term that can be considered a non Cohen–Macaulayness penalty.5.5]): If (R.36]). d−2 d−1 n+d . [V1. Theorem 2.VARIATION OF HILBERT COEFFICIENTS 11 Since Nn = In + L is a submodule of R/L. by Lemma 3. . One of these is a bound proved by M. Hence by solving the inequality d+n−2 d+n−2 + λ(R/J) d−2 d−1 we obtain the desired relation. d ν(I n ) < then there is a minimal reduction Q of I such that redQ (I) ≤ n − 1 ([ES]. E. we get L ν(Nn ) ≤ λ(R/J). Corollary 1. Let (R. Then e1 (m) ≤ e1 (m) − e1 (Q) ≤ λ(R/(Q : m))· ν(m) + λ(R/Q)d − 3d + 1 −1 . Several open questions arise. Therefore we obtain ν(I n ) ≤ ν(Ln ) + ν(I n /Ln ) ≤ ν(Ln ) + ν(M1 ) + ν(M2 ) + · · · + ν(Mn−1 ) + ν(Nn ) ≤ d+n−2 + λ(R/J)· d−2 n i=1 d+n−2−i d−2 = Recall that if d+n−2 d+n−2 + λ(R/J)· .2.

Michigan Math. 81 (2010). we have e0 (I) ≤ deg(mR/I R) + deg(R/mR) ≤ f 0 (I)(λ(m/I ) + 1) = f 0 (I)·λ(R/I) as desired. Verma. Elias. arXiv:1106. Rossi. Marcel Dekker. [S1] J. [E2] J. 827–842. Proc. 128 (2000). The other inequality. Goto. 439–454. Polini. Math. Sally. Phil. J. 21–28. E. 199 (2010). Algebra 41 (1976). The signature of the Chern coeﬃcients of local rings. 1325–1332. has a similar proof. B. e0 (I) ≤ f0 (I)·λ(R/I). Math. Vasconcelos. In particular. Proof. Phuong and W. J. K. Sally and W. 201–222. Amer. Pure and Applied Algebra 201 (2005). SHIRO GOTO. 4503–4515. Numbers of Generators of Ideals in Local Rings. Goto and K. S. 34 (1987). [MSV] M. Amer. Math. [GO] S. 116–125. Mandal. Upper bounds of Hilbert coeﬃcients and Hilbert functions. ✷ . 140 (2012). Huneke. 325–333. Proposition 4. [GMP] S. J. Math. for n ≫ 0. [PUV] C. Math. more interesting issue would be to obtain relationships going the other way. [HH] K. Pure and Applied Algebra 201 (2005). [S2] J. Consider now the corresponding exact sequence 0 → mCn /Cn+1 −→ Cn /Cn+1 −→ Cn /mCn → 0. Hanumanthu and C. Soc. Ozeki. Bounding the ﬁrst Hilbert coeﬃcient. Pure and Applied Algebra 4 (1974). [GhHV] L. VASCONCELOS 4. Singh and J. 95–105. Counting multiplicities. Mandal. New York. N. J. Cohen–Macaulayness versus the vanishing of the ﬁrst Hilbert coeﬃcient of parameter ideals. B. 138 (2010). V. 109-117. Ghezzi. London Math. Normalization The following observation shows how the special ﬁber of the normalization impacts e0 (I). Elias. f 0 (I)·λ(R/I )}. Oxford 14 (1963). Goto. [E1] J. [MV] M. K. We ﬁrst observe that Cn+1 = ICn = ICn . Soc. 279–289. D. Q. Phuong. D. Algebra 37 (2009). Math. Vasconcelos. JOOYOUN HONG. Eakin and A. The positivity of the ﬁrst normalized Hilbert coeﬃcients. AND WOLMER V. Suppose that R = ∞ Cn is ﬁnite over R = R(I). Mandal and J. Verma. We denote by f 0 (I) the multiplicity of R/mR. J. 319–336.12 LAURA GHEZZI. Comm. 2399–2406. in that range. Soc. Goto. Soc. Ghezzi. Prestable ideals. Hong and M. c Research Letters 12 (2005). [GhGHOPV] L. J. The Hilbert function of the Rattliﬀ-Rush ﬁltration. 87–94. Of course. Valla. Corso. Rossi and G. Huneke. Amer. E. Rossi and G. J. Hong and W. Proc. J. V. 147–162. T. 1995–1999. V. Rendiconti Seminario Matematico Padova 121 (2009). Kirby. m) be a normal local domain and let I be an m-primary ideal. On the Chern number of an ideal.1301v2 [Math. Hilbert coeﬃcients and reduction number 2. Buchsbaumness in local rings possessing constant ﬁrst Hilbert coeﬃcients of parameters. Cn+1 ⊂ mCn . Algebraic Geometry 1 (1992). Research Letters 16 (2009). 679–695. D. J. Normalization of ideals and Brian¸on-Skoda numbers. Proc. Algebra 325 (2011). K. Then n=0 e0 (I) ≤ min{f0 (I)·λ(R/I). Proc. Valla. Sathaye. On the ﬁrst normalized Hilbert coeﬃcient. J. [ES] P. [RV2] M. 139 (2011). Proc. Vasconcelos. Math. Math. V. On some conjectures about the Chern numbers of ﬁltrations. Let (R. 293–318. On the Chern number of a ﬁltration. J.T. 24–41. Vasconcelos.AC]. Soc. Almost Gorenstein rings. [Hu] C. A note on superﬁcial elements of an ideal of a local ring. Ozeki. Hilbert functions and symbolic powers. J. T. Matsuoka and T. Ulrich and W. Lecture Notes in Pure and Applied Mathematics 36. [SV] J.1. [RV1] M. 145 (2008). 1978. J. A bound on the reduction number of a primary ideal. Sally. [GHM] S. References [C] A. Sally modules of m-primary ideals in local rings. Soc. Nagoya Math. E. Stable rings. Hong. Camb. [R] M. Amer. [K] D.

S.VARIATION OF HILBERT COEFFICIENTS 13 [V1] W. 501 Crescent Street. Rutgers University.A. Michigan Math. U. Meiji University. Southern Connecticut State University. NJ 08854-8019. Department of Mathematics. V. NY 11201. Brooklyn.S. Kawasaki 214-8571. 300 Jay Street. E-mail address: lghezzi@citytech. Springer. 1-1-1 Higashimita. Springer Monographs in Mathematics. 725–743.meiji. Integral Closure. 57 (2008). E-mail address: vasconce@math. U.edu . U.rutgers. Piscataway. Vasconcelos. 110 Frelinghuysen Rd. CT 06515-1533.ac.A.edu Department of Mathematics. Tama-ku.A. E-mail address: hongj2@southernct.jp Department of Mathematics.S. V. School of Science and Technology. J. Vasconcelos. 2005.edu Department of Mathematics. [V2] W. New York City College of Technology-Cuny. The Chern coeﬃcients of local rings. New Haven. Japan E-mail address: goto@math. Heidelberg.cuny.

- Sally Modules
- Analise Real Vol2 No Rn
- Tópico 09 Concurso Professor Maetemática
- Exame - Matematica Discreta
- Distribuição de Aulas - 2016_PROFMAT
- Tese Dissertacao Modelo Oficial 2012UEM
- Efeito estufa
- Lista de exercícios_2
- Lista de exercícios_1
- 1-s2.0-S0021869301987642-main
- 08-GotoNO
- Numeros de Fermat
- 10 Teorema de Green
- Programacao_2013_2_Turma_2013-3
- Exercícios resolvidos
- MEM2_solucoes_cap9
- IntroduçãoàAritmética
- Lema
- Projeto instruções
- Documentos acadêmicos
- Tese-Dissertacao-Modelo Oficial 2012 - UEM
- Ciência e ética
- Artigo
- 01_trabalhos_academicos
- TemaIV Mat 8a Serie EF PROF

- New Perelman
- mathgen-1100507790
- ON THE UNIVERSALITY OF SOME SMARANDACHE LOOPS
- ON THE UNIVERSALITY OF SOME SMARANDACHE LOOPS OF BOL-MOUFANG TYPE
- Deformation Quantization of Almost Kahler Models and Lagrange-Finsler Spaces
- endomorphism
- CASA KNOTS - Parts and Bights Explanations
- Sectional Curvatures and Characteristic Classes
- Morita
- Uncountability Methods in Topological Knot Theory
- Emmy Noether, Greatest Woman Mathematicianoether, Greatest Woman Mathematician
- Sail
- Saharon Shelah and Lutz Strungmann- A characterization of Ext(G,Z) assuming (V = L)
- Scott Morrison and Kevin Walker-Higher categories, colimits, and the blob complex
- 5Adj
- A Leaf Out of Great Lives - English - Subramanian A
- 6. IJAMSS-Some Properties- Suresh
- Common Fixed Points of Weakly Reciprocally Continuous Maps Using a Gauge Function
- International Future of Mathematics Conference
- 0710.0107
- International Future of Mathematics Conference
- Amazing Paper
- Freyd - Abelian Categories
- lero03
- 540-2329-1-PB
- lero07
- On Multi-color Partitions and the Generalized Rogers-Ramanujan Identities
- Fourier-Mukai Transforms & Appls to String Theory--Andreas-p
- On Convergence of Jungck Type Iteration for Certain Contractive Conditions
- 0-0-0-0
- Go To

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd