• Ingen resultater fundet

View of Stanley depth and symbolic powers of monomial ideals

N/A
N/A
Info
Hent
Protected

Academic year: 2022

Del "View of Stanley depth and symbolic powers of monomial ideals"

Copied!
12
0
0

Indlæser.... (se fuldtekst nu)

Hele teksten

(1)

STANLEY DEPTH AND SYMBOLIC POWERS OF MONOMIAL IDEALS

S. A. SEYED FAKHARI

Abstract

The aim of this paper is to study the Stanley depth of symbolic powers of a squarefree monomial ideal. We prove that for every squarefree monomial idealI and every pair of integersk, s 1, the inequalities sdepth(S/I(ks))sdepth(S/I(s))and sdepth(I(ks))sdepth(I(s))hold. If moreoverIis unmixed of heightd, then we show that for every integerk1, sdepth(I(k+d)) sdepth(I(k))and sdepth(S/I(k+d))sdepth(S/I(k)). Finally, we consider the limit behavior of the Stanley depth of symbolic powers of a squarefree monomial ideal. We also introduce a method for comparing the Stanley depth of factors of monomial ideals.

1. Introduction

LetKbe a field andS=K[x1, . . . , xn] be the polynomial ring innvariables over the fieldK. LetM be a nonzero finitely generatedZn-gradedS-module.

LetuMbe a homogeneous element andZ⊆ {x1, . . . , xn}. TheK-subspace uK[Z] generated by all elementsuvwithv∈K[Z] is called aStanley space of dimension |Z|, if it is a free K[Z]-module. Here, as usual, |Z| denotes the number of elements of Z. A decomposition D of M as a finite direct sum of Stanley spaces is called aStanley decompositionofM. The minimum dimension of a Stanley space inD is called theStanley depth of D and is denoted by sdepth(D). The quantity

sdepth(M):=max

sdepth(D)|D is a Stanley decomposition ofM is called theStanley depthofM. Stanley [10] conjectured that

depth(M)≤sdepth(M)

for allZn-gradedS-modulesM. As a convention, we set sdepth(M)=0, when Mis the zero module. For a reader friendly introduction to the Stanley depth, we refer to [8] and for a nice survey on this topic we refer to [3].

This research was in part supported by a grant from IPM (No. 93130422).

Received 24 February 2014.

DOI: https://doi.org/10.7146/math.scand.a-25501

(2)

In this paper, we generalize the technique which was used in [9] to introduce a method for comparing the Stanley depth of factors of monomial ideals (see Theorem 2.1). We show that our method implies the known results regarding the Stanley depth of radical, integral closure and colon of monomial ideals (see Propositions 2.2, 2.3, 2.4 and 2.5).

In Section 3, we apply our method to study the Stanley depth of symbolic powers of squarefree monomial ideals. We show that for every pair of integers k, s≥1 the Stanley depth of thekth symbolic power of a squarefree monomial idealI is an upper bound for the Stanley depth of the(ks)th symbolic power ofI (see Theorem 3.2). If moreoverI is unmixed of heightd, then we show that for every integerk ≥1, the Stanley depth of thekth symbolic power ofIis an upper bound for the Stanley depth of the(k+d)th symbolic power ofI(see Theorem 3.7). Finally, in Theorem 3.10 we show that the limit behavior of the Stanley depth of unmixed squarefree monomial ideals can be very interesting.

Indeed, we show that there exist finite setsL1andL2such that sdepth(S/I(k))L1and sdepth(I(k))L2, for everyk0.

2. A comparison tool for the Stanley depth

The following theorem is the main result of this section. Using this result, we deduce some known results regarding the Stanley depth of the radical, the integral closure and the colon of monomial ideals. We should mention that in the following theorem we use Mon(S)to denote the set of all monomials in the polynomial ringS.

Theorem2.1.LetI2 I1andJ2 J1be monomial ideals inS. Assume that there exists a functionφ: Mon(S) → Mon(S), such that the following conditions are satisfied:

(i) for every monomialu∈Mon(S),uI1if and only ifφ(u)J1; (ii) for every monomialu∈Mon(S),uI2if and only ifφ(u)J2; (iii) for every Stanley spaceuK[Z] ⊆Sand every monomialv ∈Mon(S),

vuK[Z]if and only ifφ(v)φ(u)K[Z].

Then

sdepth(I1/I2)≥sdepth(J1/J2).

Proof. Consider a Stanley decomposition D:J1/J2=

m i=1

tiK[Zi]

(3)

ofJ1/J2, such that sdepth(D)=sdepth(J1/J2). By our assumptions, for every monomialuI1\I2, we have

φ(u)J1\J2.

Then for each monomialuI1\I2, we defineZu :=Zi andtu :=ti, where i∈ {1, . . . , m}is the uniquely determined index, such thatφ(u)tiK[Zi]. It is clear that

I1\I2

uK[Zu],

where the sum asK-vector space is taken over all monomialsuI1\I2. For the converse inclusion note that for every uI1\I2 and every monomial h∈K[Zu], we clearly haveuhI1. By the choice oftuandZu, we conclude thatφ(u)tuK[Zu] and therefore, by (iii),

φ(uh)φ(u)K[Zu]⊆tuK[Zu].

This implies thatφ(uh) /J2and it follows from (ii) thatuh /I2. Thus I1/I2=

uK[Zu], where the sum is taken over all monomialsuI1\I2.

Now for every 1≤im, let

Ui = {uI1\I2:uis a monomial,Zu=Zi andtu =ti}.

Without loss of generality we may assume thatUi = ∅for every 1 ≤i andUi = ∅for every+1≤im. Note that

I1/I2=

i=1

uK[Zi],

where the second sum is taken over all monomialsuUi. For every 1≤i, letuibe the greatest common divisor of elements ofUi. We claim that for every 1≤i, we haveuiUi.

Proof of claim. It is enough to show thatφ(ui)tiK[Zi]. This, together with (i) and (ii) implies thatuiI1\I2,Zui =Zi,tui =ti and henceuiUi. So assume thattidoes not divideφ(ui). Then there exists 1≤jn, such that degxj(φ(ui)) <degxj(ti), where for every monomialvS, degxj(v)denotes the degree ofvwith respect to the variablexj. Also by the choice ofui, there exists a monomialuUi, such that degx

j(u)=degx

j(ui). We conclude that uuiK[x1, . . . , xj−1, xj+1, . . . , xn],

(4)

and hence by (iii) that

φ(u)φ(ui)K[x1, . . . , xj−1, xj+1, . . . , xn]. This shows that

degxj(φ(u))=degxj(φ(ui)) <degxj(ti).

It follows thatti does not divideφ(u), which is a contradiction, sinceφ(u)tiK[Zi]. Henceti dividesφ(ui). On the other hand, since ui divides every monomialuUi, (iii) implies that for every monomial uUi, φ(ui)di- videsφ(u). Note that by the definition ofUi, for every for every monomial uUi,φ(u)tiK[Zi]. It follows that

φ(ui)tiK[Zi] and this completes the proof of our claim.

Our claim implies that for every 1≤i, we have uiK[Zi]⊆

u∈Ui

uK[Zi].

On the other hand (iii) implies that, for every monomialuUi,φ(ui)divides φ(u). Since

φ(ui)tiK[Zi] and φ(u)tiK[Zi], we conclude that

φ(u)φ(ui)K[Zi] and it follows from (iii) that

uuiK[Zi] and thus

uiK[Zi]=

u∈Ui

uK[Zi]. Therefore

I1/I2=

i=1

uiK[Zi].

Next we prove that for every 1 ≤ i, j with i = j, the summands uiK[Zi] andujK[Zj] intersect trivially. For a contradiction, letvbe a monomial inuiK[Zi]∩ujK[Zj]. Then there existhi ∈K[Zi] andhj ∈K[Zj] such that

(5)

uihi = v = ujhj. Thereforeφ(uihi) = φ(v) = φ(ujhj). ButuiUi and henceφ(ui)tiK[Zi], which by (iii) implies that

φ(uihi)φ(ui)K[Zi]⊆tiK[Zi]. Similarlyφ(ujhj)tjK[Zj]. Thus

φ(v)tiK[Zi]∩tjK[Zj], which is a contradiction, becausem

i=1tiK[Zi] is a Stanley decomposition of J1/J2. Therefore

I1/I2=

i=1

uiK[Zi]

is a Stanley decomposition ofI1/I2which proves that sdepth(I1/I2)≥min

i=1 |Zi| ≥sdepth(J1/J2).

Using Theorem 2.1, we are able to deduce many known results regarding the Stanley depth of factors of monomial ideals. For example, it is known that the Stanley depth of the radical of a monomial idealI is an upper bound for the Stanley depth ofI. In the following proposition we show that this result follows from Theorem 2.1.

Proposition2.2 (see [1], [6]).LetJ I be monomial ideals inS such that

I =√ J. Then

sdepth(I /J )≤sdepth√ I /

J . Proof. LetG(

I ) = {u1, . . . , us}be the minimal set of monomial gen- erators of√

I. For every 1 ≤ is, there exists an integerki ≥1 such that ukiiI. LetkI =lcm(k1, . . . , ks)be the least common multiple ofk1, . . . , ks. Now for every 1≤ is, we haveukiII and this implies thatukII, for every monomialu∈√

I. It follows that for every monomialuS, we have u∈√

Iif and only ifukII. Similarly there exists an integerkJ, such that for every monomialuS,u∈√

J if and only ifukJJ. Letk = lcm(kI, kJ) be the least common multiple ofkI andkJ. For every monomialuS, we defineφ(u) = uk. It is clear that φ satisfies the hypothesis of Theorem 2.1.

Hence it follows from Theorem 2.1 that sdepth(I /J )≤sdepth√

I /J

.

(6)

LetISbe an arbitrary ideal. An elementfSisintegraloverIif there exists an equation

fk+c1fk−1+ · · · +ck−1f +ck =0 with ciIi.

The set of elementsIinSwhich are integral overIis theintegral closureofI. It is known that the integral closure of a monomial idealISis a monomial ideal generated by all monomialsuS for which there exists an integerk such thatukIk(see [4, Theorem 1.4.2]).

Let I be a monomial ideal inS and let k ≥ 1 be a fixed integer. Then for every monomialuS, we have uI if and only ifusIs, for some s ≥1, if and only ifuksIks, for somes ≥1, if and only ifukIk. This shows that by settingφ(u)=ukin Theorem 2.1 we obtain the following result from [9]. We should mention that the method used in the proof of Theorem 2.1 is essentially a generalization of the one used in [9].

Proposition2.3 ([9, Theorem 2.1]).LetJ Ibe two monomial ideals in Ssuch thatI =J. Then for every integerk ≥1,

sdepth(Ik/Jk)≤sdepth(I /J ).

Similarly, using Theorem 2.1 we can deduce the following result from [9].

Proposition2.4 ([9, Theorem 2.8]).LetI2 I1be two monomial ideals inSsuch thatI1 =I2. Then there exists an integerk ≥1, such that for every s≥1,

sdepth(I1sk/I2sk)≤sdepth(I1/I2).

Proof. Note that by [9, Remark 1.1], there exist integersk1, k2 ≥1, such that for every monomialuS, we haveuk1I1k1(resp.uk2I2k2) if and only ifuI1(resp.uI2). Letk= lcm(k1, k2)be the least common multiple of k1andk2. Then for every monomialuS, we haveukI1k (resp.ukI2k) if and only ifuI1 (resp.uI2). Hence for every monomial uS and everys ≥1, we haveuskI1sk (resp.uskI2sk) if and only ifuI1(resp.

uI2). Setφ(u)=usk, for every monomialuSand everys ≥1. Now the assertion follows from Theorem 2.1.

LetIbe a monomial ideal inSandvSbe a monomial. It can be easily seen that(I :v)is a monomial ideal. Popescu [7] proves that sdepth(I :v)≥ sdepth(I ). On the other hand, Cimpoea¸s [2] proves that sdepth(S/(I :v))≥ sdepth(S/I ). Using Theorem 2.1, we prove a generalization of these results.

(7)

Proposition2.5.LetJ I be monomial ideals inS and letvSbe a monomial such that(I :v) =(J :v). Then

sdepth(I /J )≤sdepth((I :v)/(J :v)).

Proof. It is enough to use Theorem 2.1 setting φ(u) = vu, for every monomialuS.

3. Stanley depth of symbolic powers

LetIbe a squarefree monomial ideal inSand suppose thatIhas the irredundant primary decomposition

I =ᒍ1. . .∩ᒍr,

where everyi is an ideal ofS generated by a subset of the variables ofS. Letk be a positive integer. Thekthsymbolic power ofI, denoted byI(k), is defined to be

I(k)=ᒍk1. . .∩ᒍkr.

As a convention, we define thekth symbolic power ofSto be equal toS, for everyk≥1.

We now use Theorem 2.1 to compare the Stanley depth of symbolic powers of squarefree monomial ideals.

Theorem 3.1.Let JI be squarefree monomial ideals inS. Then for every pair of integersk, s≥1

sdepth(I(ks)/J(ks))≤sdepth(I(s)/J(s)).

Proof. Suppose thatI = ri=1i is the irredundant primary decomposi- tion ofI and letuSbe a monomial. ThenuI(s) if and only if for every

1≤ir

xj∈Pi

degxjus

if and only if

xj∈Pi

degxj uksk

if and only if ukI(sk). By a similar argument, uJ(s) if and only if ukJ(sk). Thus for proving our assertion, it is enough to use Theorem 2.1, settingφ(u)=uk, for every monomialuS.

The following corollary is an immediate consequence of Theorem 3.1.

(8)

Corollary3.2.LetIbe a squarefree monomial ideal inS. Then for every pair of integersk, s ≥1, the inequalities

sdepth(S/I(ks))≤sdepth(S/I(s)) and

sdepth(I(ks))≤sdepth(I(s)) hold.

Remark 3.3. Let t ≥ 1 be a fixed integer. Also let I be a squarefree monomial ideal inSand suppose thatI = ri=1i is the irredundant primary decomposition ofI. Assume thatA⊆ {x1, . . . , xn}is a subset of variables of S, such that

|ᒍiA| =t,

for every 1 ≤ ir. We setv = xi∈Axi. It is clear that for every integer k ≥1 and every integer 1 ≤ ir, a monomialu∈ Mon(S)belongs toki if and only ifuvbelongs tok+ti . This implies that for every integerk ≥ 1, a monomialu∈Mon(S)belongs toI(k)if and only ifuvbelongs toI(k+t). This shows

(I(k+t) :v)=I(k) and thus Proposition 2.5 implies that

sdepth(I(k+t))≤sdepth(I(k)) and sdepth(S/I(k+t))≤sdepth(S/I(k)).

In particular, we conclude the following result.

Proposition3.4.LetI be a squarefree monomial ideal inSand suppose there exists a subsetA⊆ {x1, . . . , xn}of variables ofS, such that for every prime ideal∈Ass(S/I ),

|ᒍ∩A| =1. Then for every integerk ≥1, the inequalities

sdepth(I(k+1))≤sdepth(I(k)) and

sdepth(S/I(k+1))≤sdepth(S/I(k)) hold.

As an example of ideals which satisfy the assumptions of Proposition 3.4, we consider the cover ideal of bipartite graphs. LetGbe a graph with vertex

(9)

setV (G)= {v1, . . . , vn}and edge setE(G). A subsetCV (G)is aminimal vertex coverofGif, first, every edge ofGis incident with a vertex inCand, second, there is no proper subset ofC with the first property. For a graphG thecover idealofGis defined by

JG=

{vi,vj}∈E(G)

xi, xj.

For instance, unmixed squarefree monomial ideals of height two are just cover ideals of graphs. The name cover ideal comes from the fact thatJGis generated by squarefree monomialsxi1. . . xir with{vi1, . . . , vir}a minimal vertex cover ofG. A graphGisbipartiteif there exists a partitionV (G) = UW with UW =∅such that each edge ofGis of the form{vi, vj}withviU and vjW.

Corollary3.5.LetGbe a bipartite graph andJGbe the cover ideal ofG. Then for every integerk ≥1, the inequalities

sdepth(JG(k+1))≤sdepth(JG(k)) and sdepth(S/JG(k+1))≤sdepth(S/JG(k)) hold.

Proof. LetV (G)=U∪Wbe the partition for the vertex set of the bipartite graphG. Note that

Ass(S/JG)=

xi, xj:{vi, vj} ∈E(G) . Thus for everyᒍ∈Ass(S/JG), we have|ᒍ∩A| =1, where

A= {xi :viU}.

Now Proposition 3.4 completes the proof of the assertion.

It is known [5, Theorem 5.1] that for a bipartite graphGwith cover idealJG, we have JG(k) = JGk, for every integer k ≥ 1. Therefore we conclude the following result from Corollary 3.5.

Corollary3.6.LetGbe a bipartite graph andJGbe the cover ideal ofG. Then for every integerk ≥1, the inequalities

sdepth(JGk+1)≤sdepth(JGk) and sdepth(S/JGk+1)≤sdepth(S/JGk) hold.

(10)

LetGbe a non-bipartite graph and letJGbe its cover ideal. We do not know whether the inequalities

sdepth(JG(k+1))≤sdepth(JG(k)) and sdepth(S/JG(k+1))≤sdepth(S/JG(k))

hold for every integerk ≥ 1. However, we will see in Corollary 3.8 that we always have the following inequalities:

sdepth(JG(k+2))≤sdepth(JG(k)) and sdepth(S/JG(k+2))≤sdepth(S/JG(k)).

In fact, we can prove something stronger as follows.

Theorem3.7.LetIbe an unmixed squarefree monomial ideal and assume thatht(I )=d. Then for every integerk≥1the inequalities

sdepth(I(k+d))≤sdepth(I(k)) and

sdepth(S/I(k+d))≤sdepth(S/I(k)) hold.

Proof. LetA= {x1, . . . , xn}be the whole set of variables. Then for every prime idealᒍ∈Ass(S/I ), we have|ᒍ∩A| =d. Hence the assertion follows from Remark 3.3.

Since the cover ideal of every graphGis unmixed of height two, we con- clude the following result.

Corollary3.8. LetGbe an arbitrary graph andJG be the cover ideal ofG. Then for every integerk ≥1, the inequalities

sdepth(JG(k+2))≤sdepth(JG(k)) and

sdepth(S/JG(k+2))≤sdepth(S/JG(k)) hold.

Corollary3.9.Let I be an unmixed squarefree monomial ideal and as- sume thatht(I )=d. Then for every integer1≤dthe sequences

sdepth(S/I(kd+))

k∈Z≥0 and

sdepth(I(kd+))

k∈Z≥0

converge.

(11)

Proof. Note that by Theorem 3.7, the sequences sdepth(S/I(kd+))

k∈Z≥0 and

sdepth(I(kd+))

k∈Z≥0

are both nonincreasing and so stabilize.

We do not know whether the Stanley depth of symbolic powers of a square- free monomial ideal stabilizes. However, Corollary 3.9 shows that one can expect a nice limit behavior for the Stanley depth of symbolic powers of square- free monomial ideals. Indeed it shows that for unmixed squarefree monomial ideals of heightd, there exist two setsL1,L2of cardinalityd, such that

sdepth(S/I(k))L1 and sdepth(I(k))L2,

for everyk0. The following theorem shows that the situation is even better.

Theorem3.10.LetIbe an unmixed squarefree monomial ideal and assume thatht(I )=d. Suppose thatt is the number of positive divisors ofd. Then

(i) There exists a setL1of cardinalityt, such thatsdepth(S/I(k))L1, for everyk0.

(ii) There exists a setL2 of cardinalityt, such thatsdepth(I(k))L2, for everyk0.

Proof. (i) Based on Corollary 3.9, it is enough to prove that for every couple of integers 1≤1, 2d, with gcd(d, 1)=2, we have

k→∞lim sdepth(S/I(kd+1))= lim

k→∞sdepth(S/I(kd+2)).

Setm=1/2. Then by Corollary 3.2,

k→∞lim sdepth(S/I(kd+2))≥ lim

k→∞sdepth(S/I(mkd+m2))

= lim

k→∞sdepth(S/I(mkd+1))

= lim

k→∞sdepth(S/I(kd+1)), where the last equality holds, because the sequence

sdepth(S/I(mkd+1))

k∈Z≥0

is a subsequence of the convergent sequence sdepth(S/I(kd+1))

k∈Z≥0.

(12)

On the other hand, since gcd(d, 1)= 2, there exists an integerm ≥ 1, such thatm1is congruent to 2 modulod. Now by a similar argument as above, we have

k→∞lim sdepth(S/I(kd+1))≥ lim

k→∞sdepth(S/I(mkd+m1))

= lim

k→∞sdepth(S/I(kd+2)), and hence

k→∞lim sdepth(S/I(kd+1))= lim

k→∞sdepth(S/I(kd+2)).

(ii) The proof is similar to the proof of (i).

REFERENCES

1. Apel, J.,On a conjecture of R. P. Stanley. II. Quotients modulo monomial ideals, J. Algebraic Combin. 17 (2003), no. 1, 57–74.

2. Cimpoea¸s , M.,Several inequalities regarding Stanley depth, Rom. J. Math. Comput. Sci. 2 (2012), no. 1, 28–40.

3. Herzog, J.,A survey on Stanley depth, Monomial ideals, computations and applications, Lecture Notes in Math., vol. 2083, Springer, Heidelberg, 2013, pp. 3–45.

4. Herzog, J., and Hibi, T.,Monomial ideals, Graduate Texts in Mathematics, vol. 260, Springer- Verlag London, Ltd., London, 2011.

5. Herzog, J., Hibi, T., and Trung, N. V.,Symbolic powers of monomial ideals and vertex cover algebras, Adv. Math. 210 (2007), no. 1, 304–322.

6. Ishaq, M.,Upper bounds for the Stanley depth, Comm. Algebra 40 (2012), no. 1, 87–97.

7. Popescu, D.,Bounds of Stanley depth, An. ¸Stiin¸t. Univ. “Ovidius” Constan¸ta Ser. Mat. 19 (2011), no. 2, 187–194.

8. Pournaki, M. R., Seyed Fakhari, S. A., Tousi, M., and Yassemi, S.,What is. . .Stanley depth?, Notices Amer. Math. Soc. 56 (2009), no. 9, 1106–1108.

9. Seyed Fakhari, S. A.,Stanley depth of the integral closure of monomial ideals, Collect. Math.

64 (2013), no. 3, 351–362.

10. Stanley, R. P.,Linear Diophantine equations and local cohomology, Invent. Math. 68 (1982), no. 2, 175–193.

S. A. SEYED FAKHARI

SCHOOL OF MATHEMATICS, STATISTICS AND COMPUTER SCIENCE COLLEGE OF SCIENCE

UNIVERSITY OF TEHRAN TEHRAN

IRAN

E-mail:aminfakhari@ut.ac.ir

Referencer

RELATEREDE DOKUMENTER

We know that it is not possible to cover all aspects of the Great War but, by approaching it from a historical, political, psychological, literary (we consider literature the prism

The feedback controller design problem with respect to robust stability is represented by the following closed-loop transfer function:.. The design problem is a standard

H2: Respondenter, der i høj grad har været udsat for følelsesmæssige krav, vold og trusler, vil i højere grad udvikle kynisme rettet mod borgerne.. De undersøgte sammenhænge

The organization of vertical complementarities within business units (i.e. divisions and product lines) substitutes divisional planning and direction for corporate planning

Driven by efforts to introduce worker friendly practices within the TQM framework, international organizations calling for better standards, national regulations and

Her skal det understreges, at forældrene, om end de ofte var særdeles pressede i deres livssituation, generelt oplevede sig selv som kompetente i forhold til at håndtere deres

Her skal det understreges, at forældrene, om end de ofte var særdeles pressede i deres livssituation, generelt oplevede sig selv som kompetente i forhold til at håndtere deres

maripaludis Mic1c10, ToF-SIMS and EDS images indicated that in the column incubated coupon the corrosion layer does not contain carbon (Figs. 6B and 9 B) whereas the corrosion