ro duing xew fijetions from yld
@to pp er in edvnes in wthemtisA
epril PPD IWWR
hvid peldmn
niversity of xew rmpshire
tmes ropp
wsshusetts snstitute of ehnology
efeg
por ny purely omintoril onstrution tht pro dues new nite sets from
given onesD ijetion etween two given sets nturlly determines ijetion
etween the orresp onding onstruted setsF e investigte the p ossiility
of going in reverseX dening ijetion etween the originl o jets from
given ijetion etween the onstruted o jetsD in eet neling4 the
onstrutionF e present preise formultion of this question nd then give
onrete riterion for determining whether suh nelltion is p ossileF sf
the riterion is stisedD then nelltion pro edure existsD nd indeedD for
mny of the onstrutions tht we hve studied p olynomilEtime nellE
tion pro edure hs een foundY on the other hndD when the riterion is not
stisedD no pro edureD however omplitedD n p erform the nelltionF
he elerted involution priniple of qrsi nd wilne ts into our ruri
one it is reognized s thinly disguised form of neling disjoint union
with xed setD itself well{known tehnique tht ts into our theoretil
frmeworkF e show tht the onstrution tht forms the grtesin pro dut
of given set with some otherD xed set nnot in generl e neledD ut
tht it n e neled in the se where the xed set rries distinguished
elementF e lso show tht the onstrution tht tkes the mth grtesin
p ower of given set n e neledD ut tht the onstrution tht forms
the p ower set of given set nnotF
uey wordsX fijetionD ijetive pro ofD qEsetD grtesin p owerD p ower set
IF sntro dution
emong the vrious mens of estlishing omintoril identitiesD mny omE
intorilists esp eilly prize ijetive pro ofs4X diret demonstrtions y
mens of mnifest ijetionsF he present sttus of the siene of ijetive
pro of my e likened to tht of rulerEndEompss onstrutions prior to
quss nd ntzelF iven with mny ingenious exmplesD ijetive pro of 4
remins n informl notionF he lk of suitle formliztion prevents exE
plortion of the limittions of ijetive pro ofD4 nd so no results we know of
my e interpreted to syD in ny senseD tht there exist nlytilly verile
omintoril identities dmitting no ijetive pro ofF sndeed the sitution
my well e sensitive to how ijetive pro of 4 is mde preiseD muh s nE
gles my e triseted with mrked strightedgesD ut not with unmrked
strightedgesF
sn this pp erD we show tht just s lgeri metho ds @s emo died in qlois
theoryA n e rought to er on questions of onstrutiility of geometriE
l o jets y use of geometril to olsD so to o n metho ds from lger e
pplied to questions of onstrutiility of omintoril o jets vi omiE
ntoril to ols suh s ijetionsF
e do not oer sp ei formliztion of the notion of ijetive pro of 4 in
this pp erF ther we oerD nd exploreD distintion etween wek ijeE
tive pro ofs nd strong ijetive pro ofs tht should e relevnt in virtully
ny prtiulr forml ontextF vet us now explin this distintionF e omiE
ntoril identity onerns two indexed fmilies of nite setsD sy fe g nd
i iPs
ff g D nd sserts tht for ll iD je j a jf jF por our purp oses nothing will
i iPs i i
e lost if we sp ek s though i were xedD so herefter we drop the susriptF
xowD let @A e n ritrry onstrution tht pro dues new nite sets from
old suh tht the rdinlity of determines the rdinlity of for ll F
hen mnifest ijetion etween e nd f lso proves tht jej a jf jY
we onsider this wek ijetive proof tht jej a jf jD insmuh s the ft
tht the rdinlity of determines the rdinlity of is not itself proved
ijetivelyF e strong ijetive proof tht jej a jf j must exhiit mnifest
ijetion etween the sets e nd f themselvesF
es we shll seeD for some onstrutions @A it is lwys p ossile to dene
ijetion f X e 3 f just in terms of ny given ijetion p X e 3 f D nd I
not tking into ount of the sp eil nture of the elements of the sets e nd
f F rovided tht the denition of f from p is within one9s forml mensD
one is then gurnteed tht wek ijetive proof using @A determines
strong ijetive proofF por our purp osesD to sy tht the denition of f from
p is within one9s forml mens4 is to sy tht f n e omputed from p D
in senrio tht we will mke preise for the ske of deniteness @though
the theorems tht we prove re not overly sensitive to the preise detils
of the mo delAF hen the nture of @A p ermits us to ompute f from p D
we sy tht suh @A n e eetively neledF e give exmples of
onstrutions @A tht n e eetively neledD long with others tht
nnotF
wore generllyD supp ose tht @Y A is onstrution tht tkes two rguE
ments nd tht g is some sort of strutured setD notion tht we shll mke
preise lterF upp ose ijetion p X @eY g A 3 @f Y g A is givenF e
derive riterion for deiding whether ijetion f X e 3 f n e dened
solely in terms of p nd the given struture on g F
his riterion is lgeri in ntureD nd in sp ei pplitions it devolves
into questions out p ermuttion representtions of nite groupsF hen
the riterion is stisedD nelltion priniple is gurnteed to existD nd
we my pro eed to seek it outF sndeedD we hve typilly found tht when
the nelility riterion is stisedD nelltion pro edure exists whose
runningEtime is ounded y p olynomil funtion of the sizes of the sets in
questionF
sf the riterion is not stisedD then there is no p oint in serhing for nelE
ltion pro edureF roweverD even suh negtive results hve p ositive imp ort
of ertin kind for the working omintorilistF por instneD knowing
tht the p ower set onstrution nnot generlly e neled suggests the
e f
p ossiility of lo oking for interesting ijetions etween P nd P even in
situtions where no interesting ijetion etween e nd f is villeF sn
ny seD questions out nelilityD when run through the mill of our
generlEpurp ose mhineryD give rise to interesting questions out p ermutE
tion representtions of nite groupsD nd thus n e resoure for lgerists
in serh of novel questions to p onderF
yur metho ds ome from group theory nd re purely mthemtilD s opE
p osed to metmthemtilF xevertheless it my e fruitful to interpret our P
results s sttements out omintoris in vrious top osesD though we do
not pursue thisF
yur work n e seen s following in trdition initited y edrino qrsi
nd tephen wilne R in their study of the involution priniple nd ontinued
y rer ert ilf IP nd fsil qordon S in their work on the priniple of
inlusion nd exlusionF
elso of some relevne is work of vszlo vovsz nd o ert eppleson on
nite sets with reltionl strutures @see I nd the referenes ited thereinAF
heir work diers from ours in tht we regrd two nite strutures s eing
equivlent if they re denle from one notherD wheres those uthors
require tht the strutures tully e isomorphiF xeverthelessD there re
mny p oints of similrity etween their pp ers nd our pp erD in terms of
oth metho ds used nd results otinedD nd it would e go o d to understnd
etter the reltionship etween the two theoriesF
yur pp er is orgnized s followsF sn etion P we give preliminries onE
erning group tions nd x some nottionF xextD in etion Q we intro due
proto ol to mke denite wht we men y eetive nelltionF hen
we prove heorem ID the fsi iquivrine griterionD to formlize the intuE
itive notion eetively nelle omintoril onstrution4 @etion RAD
nd we give severl vritions thereof @etion SAF heorem P rests the
fsi iquivrine griterion in more useful nd more elegnt form @eE
tion TAF etion U digresses to tret some issues relted to vrint form
of nelltionD for the ske of ompletenessF etion V shows tht issues of
uniformity4 of nelltion pro eduresD whih re not ddressed elsewhere
in the rtileD n e given preise formultion if one mo dies the frmeE
work so tht the role plyed y tions of groups is tken over y tions
of monoidsF he ve setions tht follow pply heorem P to otin onE
rete resultsD some p ositive nd some negtiveD some new nd some oldX
p ositive result for nelltion of disjoint union with xed set @oldD ut
we oserve how it enompsses the elerted involution priniple of qrsi
nd wilneD therey tting tht tehnique into our ruriAY negtive result
for grtesin pro duts @oldD though not previously stted in this form in the
litertureD t lest to our knowledgeAY p ositive result for grtesin pro dE
uts with p ointed sets @newAY p ositive result for grtesin p owers @newAY
nd generlly negtive result for p ower sets @newA whih is lried y the Q
solution of n interesting groupEtheoretil prolemF etion IR gives slik
pro of of very generl equivrine riterionF etion IS situtes our work
in more strtD tegoryEtheoreti settingF pinllyD etion IT trets the
relevne of our pp er to issues in the foundtions of mthemtisF yp en
prolems re intersp ersed throughout the pp erF
st is not neessry to understnd the foundtionl mteril in the rst hlf of
the pp er to p eruse the vrious expliit exmples in etions W through IQY
indeedD these setions ontin the min fruits of our workD nd the reder
my wish to smple them efore studying the ro otsD trunkD nd rnhesF es
middle ourseD the reder might strt with etions PD QD RD nd T efore
pro eeding to etions W through IQF
ome of our susidiry prop ositions hve trivil pro ofs tht my not e
trivil for the reder to nd unidedY we hve inluded ompt versions
of these pro ofsF roweverD rther thn try to follow these veritionsD the
reder might nd it more helpful to drw the pproprite digrms nd hse
rrows in the usul wyD onsulting our pro ofs only s lst resortF
PF reliminries
e denote the integers y D the nturl numers @inluding HA y xF ell
other sets onsidered re ssumed niteF
e qEset is set with n tion y group q @on the leftAF @por generl
kground on group tions on nite setsD see ghpter I of TF sf r is
sugroup of qD the osetEspe qar is lso qEsetD under the denition
H H
g @g r A a @g g Ar F e ijetion f X e 3 f etween qEsets is qEset isomorE
phism if g f a f g for ll g P qF e qEorit is trnsitive qEsetF ivery qEset
is disjoint union of qEoritsF he stilizerD t xD of x in qEset is the
group fg j g x a xgF sf r is sugroup of t xD we sy r stilizes xF
vet y e qEoritF por x P y D the mp from y to qa@t xA sending
g x to g @t xA for ll g P q is wellEdened qEset isomorphismF sf y
ontins n element whose stilizer is r D we sy tht y is qEorit of