• 検索結果がありません。

GENERALIZED EQUIVALENCE OF MATRICES OVER PREFER DOMAINS

N/A
N/A
Protected

Academic year: 2022

シェア "GENERALIZED EQUIVALENCE OF MATRICES OVER PREFER DOMAINS"

Copied!
10
0
0

読み込み中.... (全文を見る)

全文

(1)

Internat. J. Math. & Math. Scl.

VOL. 14 NO. 4 (1991) 665-674

665

GENERALIZED EQUIVALENCE OF MATRICES OVER PREFER DOMAINS

FRANK DEMEYERandHAINYAKAKAKHAIL

Department

of Mathenatics

Colorado

State

University

Fort

Collins,

CO U S A

80523

21A Victoria Park The

Mall,

Lahore Pakistan

(Received April 19, 1990)

ABSTRACT: Two

mx n matrices A,B over a commutative ring Rareequivalent

i,.ve,-tible nmtrices P,

O

over R withB PAQ. While anymxnmatrixovera principleideal dota.i,

ca, bediagonalized, thesame is not true for Dedekind domains. The first author and

T..I.

Ford ittroduced acoarserequivalencerelationonmatricescalledhomotopyandshowed any x mtrix over a. Dedekind domain is homotopic to a direct stun of x2 matrices.

In

this article wc giw, necessary and sufficient conditions on a Prefer domain that any mx n matrix be

homotolfic

to a.

directsumof x 2matrices.

l(cy Words andPhrases: Priiferdomain, Progenerator module, Bezoutdomain,matrixequivalence 1980 Subject classification codes 13F05, 13C10, 15A33.

1.

IN’I’RODUCTION

Let M,N be finitely generated projective faithful modules

(progenerators)

over a. co,nmuta.tivc rig R.

An

R homomorphism h M N is called image split in case h(M) is a faithful R-direct summand of N. If

I

:M N and g:P Q are homomorphisms of R progenerators then

are said to behomotopic ifthereare image split homomorphisms h A B and isomorphisms

,

makingthecommuting diagramofR-modules

MA I

N@B

PC

O

D

If

I P-’g

for isomorphisms p,v then

I

and g are homotopic

(where

R A B C O h. ln). Thus equivalent homomorphisms are homotopic but not conversely. The notion of homotopyofhomomorphismswintroduced in

[4]

toremovemost oftheobstructionobserved by

L.

Levy

in

[13]

todiagonalizationof matrix transformationsunder equivalenceoverDedekind domains.

Summarizingsomeof the results in

[4],

homotopyis anequivalence relationonhomomorphisms of progenerator modules and tensorproduct ofhomomorphis inducamultiplicationon homotopy classeswhich turns this setof cls intoamonoiddenoted M(R). Each

homotopy

classisreprescnted by at let onematrix transformation, and if R is a Dedekind domMn

by

a matrix transforma.tio which isadirectsumof x 2matrices,amatrixof the form

a b 0 0

0 0 a. be

m x2m

(2)

666 F. DeMEYER AND H. KAKAKHAIL

Moreover, if Ij ajR

+ bR

then

I

D 1._, D D IfR is a discrete valuation ritg lwn .(

PN[a.]. the mouoid of primitive polynofia,ls with coefficients iu N

{0,1,2,..} togctltcr

wil,l 0-1)olytomial. IfR is aDedekind domai tlen M(R) isnaturally isomorphicto

(],,,w,st,m

(

atttl tltis isomorphism gives an

isomorl)hisn

between M(R)

and

primitive

polynomials

over N ideterminates indexed byMazSpec(R).

q’le purpose of this paperis todetermine the extent to which theseresults can begctwralizcd toarbitrary doma.ins.

In fact,

theycomeclosetocharacterizingDedekind domains. Wefirst olsct tltat ifli’ isacommutativering containinga maximal ideal Psuch that dinn/t.( P/

P")

>2tlt(;t, isahonotopy class in M(R) which contains nomatrix transformation which is adirectsutnof x

tmtriccs.

Thus,

if R is a Noetherian domain andevery

homotopy

class in 3A(R) contains a ml,rix wlich isa direct sumof x2 matrices then direr

<

1. The inclusion mapfrom a domain R to its itcgralclosure

R

inducesamonoidhomomorphism.1(/)-

(/)

whichwassteadied in

[6].

llcrc

relax l,]eNoetherian condition andstudy.(/) for Prfifer domains. IfRisaPriiferdoma,io1"Krll ditnension orif Ris a Priiferdomain of finite character

(each

nonzero element,of Ris

i otly finitely many maximal ideals) weshow every class in .A(R) contains a represet|ting natrix wlicl is adirectsumof x2matrices. IfR isany valuation domain with valuegt’oup(:;<(R, +) (;+ is 1,]e nonoidofnonnegativeelements ofGformthemonoid PN(G+)of"primitive l)olynomials"

:.,,,:+ o.x.

with

a

N,ahnostall

a

0and 9cd{%lgeG+ 1.

We

showj4(R)

PN(G+).

After givi,g

asliglt generalizationof

L. Levy’s

"Separated DivisorTheorem" formatricesoverDedekittd donains

[13],

wecan show for Priifer domains that .I(R) is naturally isomorphic to

t.,nt.sr.(n).,’vl(Rt,)

it"

and o.ly if R isof finite characterand the valuation ringsat the maximal ideals of Rarc pairwisc independent. The principal examples ofPrfifer domainsoffinitecharacter whosevaltation rings at

m,ximal idealsarepairwise independentareDedekind domains and valuation donmins.

Partof this paper appeared in the first author’s Ph.D. dissertation written at Colorado Uniw:rsity. This paper was

completed

while the second author was a visitor at Florida Atlantic University. Hewishes to thank departmentchairman Jim

Brewer

for hishospitality. We would also like to thank

L. Levy

for hishelp withthe proofofthe generalized

Separated

DivisorTheorem.

"2.

SEPARATED DIVISOR THEOREM

PROPOSITION

1.

Let

R be a conunutative ring containirg a maxinal ideal p wil.l dimn/t,(P/P

’)

> 1. Then there is a matrix transformationover R which is not homotopic to a di- rect sumof x9.matrixtransformations.

PROOF: Let .

R Sbeahomomorphismof commutativerings,so Sisan

R-algebra.

Then itducesamonoidhomomorphismM(q)’M(R)-M(S) byM()(lf[)= [I(R)Yl, whereif

y

Homn(N,U.) the 13" Hotns(SC)U,,S(R)U,.)

(Theorem

of

[4]).

Sinceeach class in

M(S)is represented

byamatrix transformation, if0isanepimorphism then M() isanepimorphism. If isanepimorphism and if every class in M(R)isrepresented byamatrix whichisadirectsumof 2matrices,then ew.ry class i M(S) is

represented by

amatrix which isadirectsumof x :2matrices.

Thus,

it suffices to check tle conclusionof the propositionforahomomorphic image ofR.

Let

{ax

+ P,a+ P-} u

{c,,

+

P"},t be

a basis for

P/P"

over

RIP.

LetJbe the idealin/

generated

by

P

and {a,},,t. The ringS R/J isa local ring with maximal ideal M P

+

J/J.

Moreover,

M (0) and

dinslM(M/M)

2.

Let

a,a.., M be linearly independent over S]M

We

check the matrix aa

a:/

is not homotopicover S toally

0 ax

(3)

GENERALIZED EQUIVALENCE OF MATRICES OVER PRUFER DOMAINS 667

i.at,’ixof the form

H 0

..,

SinceS islocal,with respecttoasuitable basischoice, eachimagesplit homomorphismhasa ,natrix

representation ofthe formdiag(l,..,1,0,..,0) (Proposition

3(9)

of

[4]). We

needtocheck

F=

aO aa’2]

diag(1 1,0, ,0)=

mx2m

is ,lot equivalent over Sto H. View Fas the relation matrix of thefactor module

SO")/Lr

where l,r is thesubmodule ofS(2")

generated

by the rows ofF. Then

S(2)/Lr

is isomorphic to adirect

sul, of modules of the form A S

S

< (a,a2),(O, al) >

together

with

0-summands. A

(lit’oct

(’alcttlation shows the S-endomorphisms of S,S

leaving

< (a,a2),(O,a) > invariant are givett

"’atric"s t’the frn

aft a+m’n’

’vhere

’n’’’’’ M’’

S" That

is’

Ends(A) is a

hmm"l)"ic

imgo of the ring of these natrices.

A

direct but msy calculation shows that if

+,nJ

[ +mJ

m’ then e=O, nd fl=,, =m’=O- Sinceidempotentscan belifted modulo nilpotent e,B S,m m’ M

o

Ends(A) has kernel ideal and the natural homomorphism from

e

+

m

+

,,m,m’ whichisnilpotentwesEnds(A) hasnoidempotentsother than 0and

so

A

isan indecomposable S-module.

In

thesame wayview Gasthe relation matrix of the factor ,odue

S"’/L;

where

La

isthe submoduleofS

generated by

the rowsofG. Then

S’"I/La

is

ismnorphic toa.direct sumofmodulesoftheform B, S S/<(a,,B,)>.

An

easycalculatonshows d,sh(M"A)

, dmsm(M.

B,)

a,

and dimsl(A/MA) 2 dimsl(B,/MB,)(1 m). IfA thenA/MA/ML B,/MB,soL MLsoby Nekaya’slemma L (0).

In

thisceA B, which is impossibleby thefirst dimensioncountabove. Thus

A

isan indecomposable S-modulewhich is not adirectsummand ofany B,.

By

the Krull-Schdt Threm

Somalia S(/

and F,G cannot be equivalent matricesoverS.

REMARK"

IfR is noetherian then dim R

supes,e()dimle(P/P)

so if R is noetherian, Proposition impliesthat if every matrixover R ishomotopic toadirersumof x2matrices then

dim R 1. This may not bethecasewhen Risnot noetherian the next result shows.

Let

K denoteafieldandvavMuationonK with

vMue

group

c

(R,+).

Let

R bethevaluation ring

corrponding

tov. SinceR isanelementarydivisorring

[9],

eachmxnmatrixoverR isequivalent to adiagonM matrix diag{d,...,d} with v(d,) v(+) when d,+ 0 and

d

0impli d 0 for

R

j. The

"elementary

divisors" d d uniquely determine the

equivMence

cls.

In

this case, following

[4],

we canexplicitly determine the monoid of homotopy classes.

Let

G+

l

and N(a+) {()

= n,’n,e

N,,e

a+},

where N is the set of nonnegative integers. The, N(G+)is anultiplicative monoid with multiplicationinduced from the equation

,

’*,.

For

(.),b(x) N(G+) say () b() if thereexists positive integers rand with r() sb().

It

isesy to check that is a congruenceon

N(a+). Let

PN(a+)

N(a+)/

Then PN(G+) is a monoid whichcanbe identified with the primitivepolynomialsin withexponents fromG

+. Let

I()be the congruencecls inPN(a+)

represented

by,()

N(a+). To

the

homotopy

clsin (R)

represented

(4)

668 F. DeMEYER AND H. KALHAIL

by the lxl matrix transformation (d)overRwecanassignthecongruence class

Izv(dl

in PN[s.].

Our

text resultisthatthis assignment extendstoan isomorphism.

PR.OPOSITION

2. IfR isavaluationring corresponding toavaluationvonafield K witlvalue groupGthen M(R) PN(G

+).

PROOF: Lemma

2and Proposition 3

(l)

of

[4]

imply any0 Ihl, M(R)containsan xrmatrix transformation diag{d,,...,d} whered, 0and v(d,) v(d,+) for all i.

Let

(R) PN(G+) by (Ihl)

=

x(a’).

We only

check is well defined, then the rest of the

argument

is routine. If Idiag{d, a}l Id-a{Y, Y,}Iin M(R) with

v(I)

v(I+) for alljthen

by

Proposition

3(9)

of

[,1],

diag{d d}@diag{1 ,0 0}isequivalenttodiag{y y,}@diag{1 1,0 0}.

Let

be the entries indiag{d dr}with pairwisedistinct valuations.

By

uniquenessofinvariant fa.ctors i,a,elementarydivisorring, theentries indiag{I I, withdistinct valuationsare

f’, y.

where

,,{f) ,,(d)

whe,, weorder

d,f

so

v(a’,)< o(d’,+)

and

o(f)

<

,,(/+)

foralli.

Let ,.,

#{djlv(dj)=

,,(d’,),

S

J

S "} and

.,, #{Yl"(/,) v(/;)l S J S

s}. Then @(Ihl)

IE=t ",zta")l

and @(Ihl)

I=

Moreover, by

uniqueness ofinvariant

factors,

pr, qs,

S

5 k so

I,= r,a;)l I,= s,x"l;l

i

PN[x’]. ’l’hus 4 is well defined.

The followingisneededto provea

generalized separated

divisorthrem. Undefined termiology

can betbund in

[7].

LEMMA

3.

Let

RdenoteaPrfifer domain of finite character whose valuation rings at ma.ximal ideals are pairwise independent. If0#L is an ideal in R then thereis afactorization L

=t

L,

where each L, is contained in

exactly

one maximal idealP, ofRandP,

# P

if

#

).

PROOF:

Since 0#L and R has finitecharacter, L is contained in

only

finitely many naxinml ideals P,,...,PofR.

Let

L, LRp,flR(1G k). Theorem4.10 of

[7]

impliesL

flptaasp,c(n)(LReR).

Since LRp Rp ifL P we have L

,I(LRp,

R)

O,IL, We

always have L, CP,

Let

v,,v be valuations corresponding to

P,,P respectively.

Sincethese valuationsare pairwise independent, Theorem 22.9

(2)

of

[7]

implies that for each 0

#

z L thereis an a Re,

Rp,

with v,(a) and u(a) 0. IfS R- P,O then anelementaryexercisegives, S

-

R Re,

Rp,

so after clearing the denominator wecan sumea R. Thusa LRp, R L, buta

P

soeach Li is contained exactlyonemaximal ideal of R.

Morver,

L,

+ L

Rwhenever

#

jsince L,

+ L#

iscontained in no maximal ideals ofR. Thus L

=L, H,

L,.

To

check uniquens sumeL

H= L

where

each

L’

iscontained in exactlyone maximalideal

P

ofRand

P # P’if

j

#

q.

From

theabovew

andafter relabeling we can sume

P Pj. Now

LRp,

(= Lj)Re, LjRp,

so

if,

Re, L,Rp,. Since

L

LRp,Ritfollows that

L’,

CL,

But

Rp,

L’,

Rp,L, and

RpL’,

RpL, Rpif P isamaximalideal of R notequalto so Rp

Li/L’

0forall maximalideals PofRso

L L’

i.

Following [13],

theideals L, in

Lemma

3 arecalled the

separated

divisorsof L. The

separated

divisors Div{d,}=t ofafinitesequence ofideals in R is the collection, counting multiplicity, ofallthe sepa.rated divisors ofthe individual ideals J,.

For

conveniencewelistsomedefinitions andaresult weneedfrom

[2]. A

Prfiferdomain R with quotient field K is saidtosatisfy theInvariant

Factor Threm

iffor any finitely

generated

submodule

M of%"} there exist simultaneousdecompositions of

R

")and M

R

" ’ ... -t Jz ...

M Elzq E,._Iz,._q)E,.z,.

where

,

K, theJ, areinvertiblefractionalideals of R, theF-,i are invertibleintegralidealsofRand E,C E,+ for 1,2 1.

A

Preferdomain R has the Steinitz property if for fractional ideals

(5)

GENERALIZED EQUIVALENCE OF MATRICES OVER PRUFER DOMAINS 669

and J, qJ

_

R 1.I.

A

Prfifer domain R has the

11/2

generator property in case fi)rany fi’aclioal itleal landany0#zt lthereisayt lwitl l=+Ry.

PROPOSITION

4. If R is a Priifer domain of finite character or Krull dinmnsion tlc gent’ral ll satisfies the Inva.riant

Factor

Theorem, R has the Steinitz property and R h the

15

property

PtOOF: []

SEPARATED DIVISOR THEOREM (Levy). Let

R beaPrfifer dotninand

A

anmx mtrix over R ofrankr.

Let Ma

be thesubmoduleof R(

" genera

by therowsof

A

and let

Sa

I. It’R

is of finite character or direr then there exist invertible ideals

E E

of R witl

E,CE,+(1 1)and invertible fractional ideals

, O+

0,, such that

=

R/E,

+ J+ +... +

d. if ,"<

,,

=,

R/E, if ,.=

,

wlered7 ==E,

ifr=,n,

=JRifr=Oand

J,Rif,’=n.

[[. I["/t isof finitecharacterwith pairwiseindependentvaluationringsat maxitnal idealsand

A,A’

are two,,x,matricesoverR of rank then

A

isequivalent to

A’

if andonlyifDv{E,}’"

Dio{k.’}’=

PROOF: I. Let

M be any finitely

genert

submodule ofRt". Since R satisfies the hva.riat

Factor

Threm thereexistsimultanusdecompositions ofR" and M,

M

EI

E__ E,.Jr,.

witha-, K quotient field ofR, theJ, invertible fractional ideals ofR and E, invertibleintegralitleals of R such that

, c

E,+ < <r-1. Since R satisfiesthe

1 generator

property, Proposition of

[2]

implies Jr/ErJr R/Er.

Hence

(where

d+

...

d,, do notoccurif n).

Here

ris the rankofM whichis therankof

A

if

m

is

generated bytherowsof

A.

This prov

SA

h the decomposition giveninI. If then

so

E ..

E,__

EJ

"’). TheSteinitz

Property

and ccellation imply

=

E,.I,. R

and

Jy =l

E,.

In

thesame way, ifr 0then R") d

...

J,, andJ...J, J,+...J, R.

If then R’)

R ...

R_ J,so

J

R.

I. Let

A,

A’

be two x nmatricesoverR.

Suppose

wehavethedecompositions given in for

Sa

and Sa, If

A

is

uivalent

to

A’

then

Sa Sa,

The uniqueness

part

oftheInvnriant

Factor

Threm

(s [12])

i,npli Div{E,}=,

Div{E},

and J,+...J,

J+ ...J’,. Conversely,

if Div{E,}=,

Oi,,{E}’,’=,

and

J+

...J,

d+ ...J

then

Sa

and Sa, havethesameinvariant factors and thus areisomorphic.

To

completethe proofof

II

weneed to checkthatif

Sa Sa,

then

A

isequivalentto

A’. Our

problem is to findisomorphisms

,

0,

a

making the commuting

diagram

Rm

Rm)

Sa

0

o A’ Snce

R 8

re,

beorem 1.8

o [9

mphe8

.

? 6ere?8

poectve

and

(6)

60 F. DeMEYER AND H. KAKAKHAIL

If SA P isthe projection let pbeasplittingmapso R(" p(P)+Qand ker ,1CQ. It thesame way Sa,

P’+T’,

R()

p’(P’)+Q’

and ker

1’

C

Q’.

Since

Sa Sa,

thereare

isomorl)hists

a: !’--

!"

aud 7’+

7". By

Proposition of

[12],

T isadirectsumof cyclicmodules R/Lfor ideals0 LC 1.

By

lean,ha3 wecan let {L,} bethe setof separateddivisorsofL. The ChineseRetnain(h’r

implies R/L +R/L,. SinceL, is contained in

only

onemaximalideal of R, R/L, is local tbr all i.

Since Q/kero 7",Theorem 1.6of

[11]

impliesthere isasimultanusdepositiouof Qan(I ker 0

In

thesamewaythereisasimultanusdecompositionof

Q’

andker

0’.

Thus the given isomorpltis

T

7"

extends to an isomorphism :Q

Q’

such that +(ker0) ker

0’.

Thisgives isom<>rpliss

+

and

+ p’a +

makingthe commutativediagram

R(M) R(M)

SA

0

Sillc(’07(ker 0) ker,l’ wehavetheexact

R(

’A’

IrnageA’---.O

ald

:,

is the lt-honomorphism given since R

’’

is free. Thiscompletesthe proofof the

Separated

Divisor Theorem.

3.

THE MONOID OF HOMOTOPY CLASSES

LEMMA

5.

Let

RbeaPr/iferdomainoffinitecharacterorKrulldimension <1.

(1)

If0

# Ill

.(R) then thereexists g such that

Ill

Igl in sM(R) and coker(g) is a torsio R-module.

(2) Let

Istl, lgle ./I(R} with

coker(g)

and

coker(st)

torsion R-modules. If M g g P Q

theu I/I Igl in .X4(R) if andonly ifthereexists

R-progenerators

K,L such that g(R)If Q63L and

coker(f)

(R)K coke,’(g)(R)L.

PROOF:

The proofnowfollows "mutatis mundantis" asthe proofof lemma 6of

[4].

Following page393 of

[4]

adescriptionof Yt4(R) in terms of ideals of Rcanbe givennow. Consider the set

A

{(M,R(’))Im

>1,M isafinitely

generated

Rsubmodule of R(m)such that

R("*)/M

is torsion Rmodule.

Definemultiplicationin

A

asfollows: (M,

R(m))(N,

R

(n))

(M(R)N,R(m) (R)

R(n)),

then

A

isacommutativc semigroup. Defineasemigroup homomorphism

p:A--..’(R) A4(R)-{0}

byp(M,R

(’)) Jil

where M R(’) is the inclusion map.

Note

that by

Lemma

5 p(N,

R("))

if and

only

ifthereexist

R-progenerators

Kand Lsuch that

R(’) (R)K

R(’)(R)

L and

(R(m)IM)(R)

K

- (R(")IN)(R)

L.

p(M,R("))

Thust)inducesanequivalence relation on Aasfollows:

(M,R(m))~(N,R(n))

ifand onlyif

p(M,R(m))=

p(N,R

(n))

(7)

GENERALIZED EQUIVALENCE OF MATRICES OVER PRUFER DOMAINS 671

Let

.4bethesetofequivalenceclasses of

A.

Then the

product

on

A

induces themultiplicationo,.4, turningitintoacommutative monoid with identitythe class containing(R,R).

LEMMA

6.

Let

R beaPrfifer domain of finitecharacteror Krulldimension < 1, then tlw nlap

"

.4

.

(R) induced by pis anisomorphism.

PROOF: Same

as Proposition 7 of

[4].

THEOREM

7. IfR is a Priifer domain offinite character or Krull dimension < then every homomorphism of

R-progenerators

ishomotopictoamatrix transformation of theform

0 a: b:

am bm m 2m where

if/

=(a,b) then

I

3l+(l<j<n-1).

PROOF:

The proof of Theorem7 now followsexactly ason pages 391-394of

[4].

LEMMA

8. Let Rdenotean integraldomain. Then thefollowingareequivalent.

Eachnonzeroelementof R is contained inonly finitelymany maximal ideals ofR.

2)

IfN is a finitely

generated

submodule ofR(") then

Re

(R)N is a direct

summand

of

R

") for

almost all PeMaxSpec(R).

(3) For

eachexact sequence 0 N Q

m

0of finitely

generated

R-moduleswith Qprojective’

the associated sequence 0 Rt,(R)N R,(R)Q Rp(R)M 0is

split

exactfor almost allPeMa:cSpec( /i’)

PROOF:

The equivalenceof2and3 followseasilysince everyfinitely

generated

projective naodule isadirectsummand ofafreemodule offiniterank. Tosee 2 let0 a R. Then

(a)

isasubmodule

ofRso (R) iseither0or aunit in

Re

foralmost all PMaxSpec(R). IfRisan integraldomain then

(R)a 0 in

Re

so a P for almost all PeMa:Spee(R).

To

show 2 let R(")

Rx ...

$R,,, and

let N

Rn

+...

+

Rn, with0 n,

R(n}(1 _< _<

k).

Let

n a+...

+

a,:, wherewe can assume a 0.

Over a’(R

wehave

a[n,x,.

xnis abasisfor

(a-(R)

") and

a-{n,n:,..,n: generates a’N

so

replace x

and

n

by

a{n

over

ajaR.

Then a,,...,,, is a basis for

(a’R)

n and z,n.

generates

a?

N. Writen. bz

+

b.z

+ +

b,x,. Sinceze

a[N

wecan

replace

n by b.z+...

+

where be 0 or ,e

a’[Rn a’{R:.

If b. 0then over

ba-(R

wehave

,bn.,:z ,,

isabasis for

(b.a-[R)

") and

z,btn2,nz

nt, generates

baN

so

replace

and n by

bn

over

b.a’R.

Then N

(ba-{

R)x

+

(b. a’{ R)x

+

(b a-{ R)nz +...

+

(b a-{ R)n.

Afler

finitelymanystepswe can find anelementc R withc-N adirect summandof/i5"). Sincecis inonly finitelymany naaxima.l idealsofRby 1, Re(R)N isadirectsummand of

R

’*)for almost allPMa:Spec(R).

IfR is an

integral

domainof finite character then a nonzero homomorphism

m

N of R- progeneratorsinducesan

image

split

homomorphism R,(R)MX-,IRt,(R)N

for almost all

PeMaxSpec(R)by

3 of

Lemma

8.

In

thiscasetheinclusions R

Re

induceanaturalmap :M(R)

LEMMA

9. If R isa Prfifer domain of finitecharacter

and

M(R)

,,M,,s,n)M(R,)

is inducedfrom the inclusions R Rp then isamonomorphism.

PROOF: Let

0

Ill

0 Igl bein M(R)andassume(I/’1) (Igl). Then [1(R)

II l1

(R)

for all PMaxSpec(R).

By

Theorem 7, 0

Is’l

determines adescendingsequence 1

I.

::)1,, of ideals in R with 0

I

(a,b) and 0 Igl determines

J or,,,

with 0

or,

(c,,d,). Under

j R,,)

the isomorphism#of

Lemma

6,

(’]=I

R")corresponds to

Ill

and (9=

,

correspondstoI.ql-

We

show

(’]=I,R

’)

(or,R)

in A

Let

P be a maximal ideal for which

II

(R)

(8)

672 F. DeMEYER AND H. KAKAKHAIL

is split.

In

A(Rp) the split class is represented by (Re,Re) and projective modules are free over

Re

so

(R(’)/=t

I)tt)(R)

Re

(0) which implies

I

c Pfor j 5 n.

In

thesame way

y c

P so

lRp Rp

JRe

for all i,j whenever

11 fJ Jl l

issplit over

Re.

Consider thefinite set

(si,ce

R h finite

character)

S {PMazSpec(R)II1

fl

Ila.I}

{PMazSc(R)II1

@

l #

Iln.I}. Since

17= e,,ng’l IZ neJ,nl

in A(Rp) for eh P S there are positive integers se,te with

spn tpmand

[Rp/Rpb]t’"

[Rp/RpJ]).

Let

He,ssp and p,stp. Then for each P S,

l[Rp/Rpl2]{,) [,=lRe/RpI,](,)n _ [t=lRP/Rejt](,)m

,=1

[RP/Jk]

(t)" The ideals Rt,

I

and

ReJk in thisdecompositionare

uniquely

determinas sets

(ii

pg 260 of

[12]).

This means thelist

ofideals RpIt

Re I

and

RJ RFJ...

where

It I

and

J J

are thesa,c for

each p S. Therefore,foreachj,t,

P,Ms,,n)(J Re

R) forcorresponding k,w

We have shown

[t(R/I)] ’ [=(R/J)]

so

(R),=6) (R",=J)

in A and is a ,onomorphism.

We

need two ey lemm about valuationson Preferdomains.

IEMMA

10.

Let

R be aPrefer domain of finitecharacter whose valuations at ma.ximal ideals re pairwise independent.

Let

p p, be a finite set of maximal idealsofR, let G, be the value group ofRe, and let0<a,G,(1 n). Then there existsafinitely

generated

ideal of R such that IRe, {o Re, lve,() 9i} and the

only

maximalidealsofRcontaining are

PROOF:

Since the valuations atthe maximalidealsofRarepirwise independent, Theorem22.9 of 7 impliesthere exists anx Re,

... Re.

such that ve,(x) ,(1 n).

As

weobserved in the

proof

of

Lemma

3, we canchoosez R. SinRhfinitecharterwe can let

O

Q,, beall the

maximal ideals of R distinctfrom {Pi},% such that r

O(1 S

j m). Again, Theorem 22.9 of

[7]

implies thereare u, R withve,(,)

,

and

v,(,)

0forallj i,k.

Let

I (x, u). Then is finitely

generated, IRe,

zRe, { Re, lve,(a)

E

9,} and the onlymaximal ideals of R containing re P,

., Pn.

LEMMA

11. Let v,v be valuationson field K with value groupsGt,G2 and valuation rings V,

V

respectively. Ifforchpair(9,)

G

x

G

with0S and0S9

there

isan

andv(r)=9then thevaluationrings

v, v

areindependent.

PROOF: (S

9, pg 289 of

[7]).

THEOREM

12.

Let

RbeaPrfferdomain. Theinclusionmaps R

Re

induce n

isomorphism

’(R)

e,,s,,n)(Re)

ifand

only

if R is of finite character and the valuation rings at the maximalideals ofRarepairwiseindependent.

PROOF: Assume

RisaPrffer domain of finitecharacter.

Lemma

9giv isamonomorphism.

We

check that if in ddition the valuationringsatthe maximal ideals of Rarepairwiseindependent then is anepimorphism.

Let

(lgel)e,,s,,<n) beanelement of

e,,s,,n)(Re).

Then ae isan inaage splitmap for all butfinitelymany maximal ideMs

p

pof R. Each

I.1

canbe

reprented

by

adiagonalmatrix

(Proposition 2). By

tensoring thesematric withidentitymatrices ofappropriate sizeswecan sumeeach 19,1is

reprented by

a

diagonM

mxnmatrix.

Let

19e, be

represented by

diag(a, a,m)(1 k).

Let

v, beavaluationdeterminedby thevaluationring

Re,

withvalue group G, and let 9, vi(a,)(1 k, j m).

Lena

10 gives finitely

generated

ideals

I

contained

(9)

GENERALIZED EQUIVALENCE OF MATRICES OVER PRUFER DOMAINS 673

in exactly the maximal ideals

P P

and

IRe,

{a RP,Iv,(a)> g,} for <j<m. I,et

I11

sucl that l/I corresponds tothe element

(

l,

R(’))

ofA u {0} under theisomorphisn 6. Then (Ifl) (lgPl)e,MaSec(R)SOb isanepimorphism.

Conversely, assume the inclusion maps R- Rp for PMaxSpec(R) induce the

isomorl)lis Let

0 R and let

e,

R R by the honomorphism given by left multiplication by a. ’l’lc

levi lnl

in M(R) if and only ifa is aunit in

n (Proposition 3(5)

of

[4]).

The inmgc

o1"4

will lie

in t,,MaSpee(a).A(Rp)onlyif the image of

levi

in M(Rp) is

Ilm.I

foralmost all PM,zSpcc(R). This means a

it

PforalmostallPMaxSpec(R)soR must havefinitecharacter.

Let

P,Qbetnaximal ideals of R and let S Rpf

RQ.

Since isanepimorphism, theinduced map ,"A/I(S) AI(Rt,)

isanepimorphism.

To

seethe valuation ringsR, and

P

areindependentwecheck the condition of Iemna 11.

Let

vp and 0 be thevaluationscorrespondingto valuation rings Rpand

groul)s(;p,Gq. I,(’t 0<.q Gt,and0<_h

GO

and let Rp, RQwith vp(a) g,vQtb) h.

oll(), liere isa

Ihl

3A(S)withII(R)hl

levi

in

M(Re)and

Ill,

hi Itl

inM(RQ). SinceSisas(’ilo(’al Bczou! donai,

s

isa elementarydivisor domain

[9]

so wecan representh bythediagonal nalrix

(liag(c, ,c,.).

As

we sawin theproof of Proposition 2,we canfind units

ue Re

and

w

RQSll(’htlat

cju in

Re

and

c

u,bin RQ(I <_j<_k). Thus vt,(c) vt,(ua) v,(a)and ve(c) VQ(Wlb) vQ(b) which shows theva.luation rings Re and /Q arepairwise independent.

REFERENCES

I. H.

Bass,

K-Theory and Stable

Algebra,

Publ.

I.H.E.S. 22(1964),

pp. 5-60.

"2..I. Brewerand L. Klingler, Pole Assignabilityand theInvariant FactorTheoren for Priit’cr l)o- na.insand Dedekind Domains,

J. Algebra.

Vol.

1.112 #2.

1987,pp. 536-545.

3. WillyBrandal,TheCommutative

Rings

Whose

Finitely

Generated ModulesI)ecoml)OSe,

Notes

in

Mathenaics, Springer-v’erlag,

Vol.

723(1979).

1.

F. R. DeMeyer

and

T. J. Ford,

Homomorphisms of

Progenerator Modules, J.

of Algebra,

!1:,

Number 2, March 1988pp. 379-398.

5.

F. R. DeMeyer

and

E. Ingraham,Separable

Algebras

Over

Commutative

R.ings.,

Lecture

in Mathematics,

No.

181,Springer-Verlag, Heidelberg, 1971.

6. ’l’.3. Ford, Homomorphismsofprogenerator modulesundera

change

of base ring, (_’,omnmi(’a- tions

i.n Al;ebra 1_6(3)(1988)457-482.

7.

R.

Gihner.

Multil)licative

IdealTheory,

M.

Dekker,

New York,

1972.

8.

W.

Heinzer, Quotient

Overrings

of

Integral

Domains, Mathematikh 17

(1970)p.

139-148.

9.

I. Kaplansky,

Modulesover DedekindRings and Valuation Rings,

Trans. Amer.

Math.

Soc.

72

(1952),

327-340.

10. M. D.

Larsen, W. J.

Lewisand T.

S. Shores,

Elementary Divisor Ringsand Finitely Presented

Modules, T.rans. Amer.

Math.

Soc.

Volume

187, Issue

1, 1974 pp. 231-248.

11.

L.

l,evy, DecomposingPairsof

Modules, Trans. Amer.

Math.

Soc.

122

(1966),

pp. 64-80.

12. ----, lnvariantfactor Theorem for Priifer Domains of Finite

Character, J.

of

Algebra,

Vol. 106, 1987,pp. 259-264.

13. ----,Almost

Diagonal

Matrices

Over

Dedekind Domains, Math

Z.

124

(1972),

pp. 89-99.

Frank

DeMeyer

Department

ofMathematics Colorado

State

University Fort Collins,

CO U S A

80523

Hainya Kakakhail 21A VictoriaPark The

Mall,

Lahore Pakistan

(10)

Mathematical Problems in Engineering

Special Issue on

Modeling Experimental Nonlinear Dynamics and Chaotic Scenarios

Call for Papers

Thinking about nonlinearity in engineering areas, up to the 70s, was focused on intentionally built nonlinear parts in order to improve the operational characteristics of a device or system. Keying, saturation, hysteretic phenomena, and dead zones were added to existing devices increasing their behavior diversity and precision. In this context, an intrinsic nonlinearity was treated just as a linear approximation, around equilibrium points.

Inspired on the rediscovering of the richness of nonlinear and chaotic phenomena, engineers started using analytical tools from “Qualitative Theory of Differential Equations,”

allowing more precise analysis and synthesis, in order to produce new vital products and services. Bifurcation theory, dynamical systems and chaos started to be part of the mandatory set of tools for design engineers.

This proposed special edition of the Mathematical Prob- lems in Engineering aims to provide a picture of the impor- tance of the bifurcation theory, relating it with nonlinear and chaotic dynamics for natural and engineered systems.

Ideas of how this dynamics can be captured through precisely tailored real and numerical experiments and understanding by the combination of specific tools that associate dynamical system theory and geometric tools in a very clever, sophis- ticated, and at the same time simple and unique analytical environment are the subject of this issue, allowing new methods to design high-precision devices and equipment.

Authors should follow the Mathematical Problems in Engineering manuscript format described at http://www .hindawi.com/journals/mpe/. Prospective authors should submit an electronic copy of their complete manuscript through the journal Manuscript Tracking System athttp://

mts.hindawi.com/according to the following timetable:

Manuscript Due December 1, 2008 First Round of Reviews March 1, 2009 Publication Date June 1, 2009

Guest Editors

José Roberto Castilho Piqueira,Telecommunication and Control Engineering Department, Polytechnic School, The University of São Paulo, 05508-970 São Paulo, Brazil;

[email protected]

Elbert E. Neher Macau,Laboratório Associado de Matemática Aplicada e Computação (LAC), Instituto Nacional de Pesquisas Espaciais (INPE), São Josè dos Campos, 12227-010 São Paulo, Brazil ; [email protected] Celso Grebogi,Center for Applied Dynamics Research, King’s College, University of Aberdeen, Aberdeen AB24 3UE, UK; [email protected]

Hindawi Publishing Corporation http://www.hindawi.com

参照

関連したドキュメント

The elements of a finite nonempty partially ordered set are exposed at independent uniform times in [ 0, 1 ] to a selector who, at any given time, can see the structure of the

This is a consequence of a more general result on interacting particle systems that shows that a stationary measure is ergodic if and only if the sigma algebra of sets invariant

EQUIVALENCE AND CONGRUENCE OF MATRICES UNDER THE ACTION OF STANDARD PARABOLIC SUBGROUPS ∗.. FERNANDO

It is well known that an elliptic curve over a finite field has a group structure which is the product of at most two cyclic groups.. Here L k is the kth Lucas number and F k is the

The equivalence between some conditions concerning elliptic matrices is shown, namely, the Cordes condition, a generalized form of Campanato’s condition, and a generalized form of

Let F be a simple smooth closed curve and denote its exterior by Aco.. From here our plan is to approximate the solution of the problem P using the finite element method. The

Next we shall prove Lemma 3.. Then G=F' follows from Proposition 1. This completes the proof of Lemma 3. Let us consider the exact sequence.. r\

Using the finite element method we calculated the approximate values of λ 1 for various values of p for different shape of domain Ω in twodimension.. These domains are the unit