Contents lists available atScienceDirect
Journal
of
Algebra
www.elsevier.com/locate/jalgebra
Monomial
G-posets
and
their
Lefschetz
invariants
Serge Bouca, Hatice Mutlub,∗
aCNRS-LAMFA,UniversitédePicardie- JulesVerne,33rueStLeu, 80039 - Amiens, France
b
DepartmentofMathematics,BilkentUniversity,06800- Bilkent,Ankara, Turkey
a r t i c l e i n f o a b s t r a c t
Articlehistory:
Received15August2018 Availableonline14March2019 CommunicatedbyMarkus Linckelmann MSC: 06A11 19A22 20J15 Keywords: Burnsidering Monomial Tensorinduction Lefschetzinvariant
Let G be a finite group, and C be an abelian group. We introducethenotionsofC-monomialG-setsandC-monomial G-posets,andstatesomeoftheircategoricalproperties.This gives in particular a new description of the C-monomial BurnsideringBC(G).WealsointroduceLefschetzinvariants
of C-monomial G-posets, which are elements of BC(G).
Theseinvariantsallowforadefinitionofageneralizedtensor induction multiplicative map TU,λ : BC(G) → BC(H)
associated to any C-monomial (G,H)-biset (U,λ), which inturn givesa group homomorphismBC(G)× → BC(H)×
betweentheunitgroupsofC-monomialBurnsiderings. ©2019ElsevierInc.Allrightsreserved.
1. Introduction
LetG beafinitegroup,andC beanabeliangroup.Inthiswork,wefirstintroducethe notionofC-monomialG-set:thisisapair(X,l) consistingofafiniteG-setX,together withafunctorfromthetransportercategoryX of X,tothe groupoid•Cwithoneobject
andautomorphismgroupC.TheC-monomialG-setsformacategoryCM G-set,andwe
show that it is equivalent to the category CF G-set of C-fibred G-sets considered by
* Correspondingauthor.
E-mailaddresses:serge.bouc@u-picardie.fr(S. Bouc),hatice.mutlu@bilkent.edu.tr(H. Mutlu). https://doi.org/10.1016/j.jalgebra.2019.02.036
Barker ([1]). Inparticular, the C-monomial Burnside ring BC(G) introduced by Dress
([5])isisomorphictotheGrothendieckringofthecategoryCM G-set.
WeextendthesedefinitionstothenotionofC-monomialG-poset:thisisapair(X,l) consisting ofafiniteG-posetX,andafunctorl fromthetransportercategoryX to •C.
We associate to each such pair (X,l) aLefschetz invariant Λ(X,l) lying inBC(G). We
showthatanyelementofBC(G) isequaltotheLefschetzinvariantofsome(nonunique)
C-monomialG-poset.
Wealso introducethecategoryCM G-poset ofC-monomialG-posets,and showthat
therearenaturalfunctorsofinduction IndHG :CM H-poset→CM G-poset andof
restric-tion ResGH:CM G-poset→CM H-poset,wheneverH isasubgroupofG.Thesefunctors
are compatiblewiththeconstructionofLefschetz invariants.
We extendseveralclassicalpropertiesoftheLefschetz invariantsof G-posetsto Lef-schetz invariantsofC-monomial G-posets (theclassicalcasebeing thecasewhere C is
trivial).
Wenextturnto theconstructionofgeneralizedtensor induction functors
TU,λ:CM G-poset→CM H-poset
associated,forarbitraryfinitegroupsG andH,toanyC-monomial(G,H)-biset(U,λ).
Weshow thatthesefunctors inducewelldefinedtensorinductionmaps
TU,λ: BC(G)→ BC(H),
which arenotadditiveingeneral,butmultiplicativeandpreserveidentity elements.In particular,wegetinducedgrouphomomorphismsbetweenthecorrespondingunitgroups of monomial Burnside rings,similar to those obtainedby Carman([4]) for otherusual representation rings.
Weshowmoreoverthatunderanadditionalassumption,thesetensorinduction func-tors andtheirassociated tensorinductionmaps arewellbehaved forcomposition.This yieldstoa(partial)fibredbisetfunctorstructureonthegroupofunitsofthemonomial Burnside ring.
2. ThemonomialBurnside ring
LetG beafinitegroupandC beanabeliangroupwhichisnotedmultiplicatively.We denote byG-set thecategory of finiteG-sets (withG-equivariantmaps asmorphisms), and B(G) theusualBurnsideringofG,i.e.theGrothendieckringofG-set forrelations given bydisjointuniondecompositionsoffinite G-sets.
2.1. Thecategory of C-fibredG-sets
AC-fibredG-set isdefinedtobeaC-free(C×G)-setwithfinitelymanyC-orbits.Let
maps.ThecoproductofC-fibredG-setsX,Y istheircoproductX Y assets,withthe obvious(C× G)-action.IfX andY are C-fibred G-sets,there isaC-actiononX× Y
defined by c(x,y) = (cx,c−1y) for any c ∈ C and (x,y) ∈ X × Y . The C-orbit of an element(x,y) ofX× Y isdenotedbyx⊗ y andthesetofC-orbitsisdenotedbyX⊗ Y .
MoreoverC× G actsonX⊗ Y by
(c, g)(x⊗ y) = cgx ⊗ gy
for any (c,g) ∈ C × G andx⊗ y ∈ X ⊗ Y . One checks easily thatX ⊗ Y is again a
C-fibredG-set, calledthetensorproductofX andY .
We denote the isomorphism class of a C-fibred G-set X by [X]. The C-monomial Burnside ring BC(G),introduced by Dress ([5]), is definedas the Grothendieckgroup
ofthecategory ofC-fibredG-sets, forrelationsgivenby[X]+ [Y ]= [X Y ].Thering structureofBC(G) isinducedby[X]· [Y ]= [X⊗ Y ].Theidentity elementisthesetC
withtrivialG-actionandthezeroelementistheemptyset.IfC istrivialwerecoverthe ordinaryBurnside ringofthegroupG.
GivenaC-fibredG-setX,wedenotethesetofC-orbitsonX byC\X.Thegroup G acts onC\X, and X is (C× G)-transitive ifand only if C\X isG-transitive. If C\X
is transitiveas aG-set it isisomorphic to G/U forsomeU ≤ G.There exists agroup homomorphism μ: U → C suchthat ifU isthe stabilizer of the orbitCx,then ax=
μ(a)x foralla∈ U.Sincethestabilizer(C× G)x ofx inC× G isequalto
(C× G)x={
μ(a)−1, a| a ∈ U},
theC-fibredG-setX isdetermineduptoisomorphismbythesubgroupU andμ.
Conversely,letU beasubgroupofG,andμ: U→ C beagrouphomomorphism.Then wesetUμ={
μ(a)−1,a| a∈ U},anddenoteby[U,μ]GtheC-fibredG-set(C×G)/Uμ.
Thepair (U,μ) iscalled aC-subcharacter of G. Wedenote the set of C-subcharacters
by ch(G).ThegroupG actson ch(G) byconjugation.TheG-set ch(G) isaposetwith therelation≤ definedby
(U, μ)≤ (V, ν) ⇔ U ≤ V and ResVUν = μ
forany(U,μ) and(V,ν) in ch(G). Asanabeliangroupwe have
BC(G) =
(U,μ)∈Gch(G)
Z[U, μ]G
where(V,ν) runsoverG-representativesoftheC-subcharactersofG,detailscanbeseen in [1].
2.2. Thecategory of C-monomialG-sets
LetG beafinite groupandC beanabeliangroup.GivenaG-setX,weconsider its transportercategoryX whose objectsaretheelementsofX andgivenx,y inX theset of morphismsfromx toy is
HomX(x, y) ={g ∈ G | gx = y}.
Let•C denotethecategorywithoneobjectwheremorphismsaretheelementsofC and
compositionismultiplication inC.Nowwedefine C-monomialG-setsasfollows.
Definition 1.A C-monomial G-set isa pair(X,l) consistingof a finite G-set X and a functor l: X → •C.
In other words, foreach x,y ∈ X andg ∈ G such thatgx= y, wehave anelement l(g,x,y) ofC,withthepropertythatl(h,y,z)l(g,x,y)= l(hg,x,z) ifh∈ G andhy = z,
and l(1,x,x)= 1 foranyx∈ X.
Let (X,l) and (Y,m) be C-monomial G-sets. If f : X → Y is amap of G-sets, we slightly abuse notation and also denote by f : X → Y the obvious functor induced by f . Now amap(f,λ): (X,l)→ (Y,m) of C-monomialG-sets is apairconsisting of amap f : X → Y of G-sets andanaturaltransformation λ: l→ m◦ f. Wedenote by
CM G-set thecategory whoseobjects areC-monomial G-sets,morphisms arethemaps
of C-monomialG-sets,andcompositionistheobviousone.
Let (X,l) and (X,l) be C-monomial G-sets. We define the disjoint union of
C-monomial G-sets as (X,l) (X,l) = (X X,l l) where X X is the disjoint unionofG-setsand
l l: X X→ •C
is thefunctor suchthat
(l l)(g, z1, z2) =
l(g, z1, z2) z1, z2∈ X
l(g, z1, z2) z1, z2∈ X
forany z1,z2∈ X X suchthatgz1= z2 forsomeg∈ G.
The product of C-monomial G-sets (X,l), (X,l) is defined to be (X × X,l× l) where X× XistheproductofG-setsandl× l: X× Y → •Cisthefunctordefinedby
(l× l)g, (x, x), (y, y)= l(g, x, y)l(g, x, y) forg∈ G and(x,x),(y,y)∈ X × X suchthatg(x,x)= (y,y).
Our goal is to show that the categories CM G-set and CF G-set are equivalent. For
this, wedefineafunctorF :CM G-set→CF G-set asfollows:givenaC-monomialG-set
F (X, l) = C×lX,
which is the direct product C × X endowed with the (C × G)-action defined by (k,g)(c,x)=kcl(g,x,gx),gxforany(k,g)∈ C × G and (c,x)∈ C × X.
Givenamap(f,λ): (X,l)→ (Y,m) of C-monomialG-sets,wedefine
F (f, λ) : C×lX → C ×mY
byF (f,λ)(c,x)= (cλx,f (x)) for any (c,x)∈ C ×lX.ThenF (f,λ) isa(C× G)-map:
indeed,given(k,g)∈ C × G and(c,x)∈ C × X,wehave (k, g)F (f, λ)(c, x) = (k, g)cλx, f (x) =kcλxm(g, f (x), f (gx)), f (gx) =kcλgxl(g, x, gx), f (gx) = F (f, λ)kcl(g, x, gx), gx = F (f, λ)(k, g)(c, x).
ItisclearthatF :CM G-set→CF G-set is afunctor.
Lemma 2.Let C be an abelian group and G be a finite group. Then the above functor F :CM G-set→CF G-set isan equivalenceofcategories.
Proof. We provethat F is fully faithfuland essentially surjective. Firstwe show that
F is essentially surjective. Given aC-fibred G-set X, letC\X be the set of C-orbits.
Clearly C\X is aG-set. Wedefine a functorl : C\X → •C. LetCx, Cy ∈ C\X such
thatCgx= Cy for some g ∈ G. Then there exists auniquec ∈ C suchthat gx= cy. Wesetl(g,Cx,Cy)= c.WehaveF (C\X,l)= C×l(C\X).Now chooseaset[C\X] of
G-representatives ofthe G-actionon C\X. Then forany x∈ X, there exists a unique
Cσx ∈ [C\X] suchthatx∈ Cσx. Since X isC-free,there exists auniquecx ∈ C such
thatx= cxσx.Wedefinea(C×G)-mapf : X→ C×l(C\X) suchthatf (x)= (cx,Cσx).
Then
(c, g)f (x) = (c, g)(cx, Cσx) = (cxcl(g, Cσx, Cgσx), Cgσx) = (cxc, Cgσx)
= (ccgx, Cgσx) = f
(c, g)x.
Sof isa(C× G)-mapandclearlyanisomorphism.Thus,F isessentiallysurjective. Let(X,l) and(Y,m) beC-monomialG-sets.Weneedtoshow thatthemap
F : Hom(X, l), (Y, m)→ HomF (X, l), F (Y, m)
inducedbyF issurjectiveand injective.Letϕ: C×lX → C ×mY bea(C× G)-map.
Given(1,x)∈ C×lX,letϕ(1,x)= (cx,zx) for(cx,zx)∈ C×Y .Sinceϕ isa(C×G)-map,
ϕ(c, x) = (ccx, zx)
and
ϕ(1, gx) =cxm(g, zx, gzx)l−1(g, x, gx), gzx
forany c∈ C and g∈ G. Wedefine amap
(f, λ) : (X, l)→ (Y, m)
suchthatf : X→ Y isdefinedbyf (x)= zxandλ: l→ m◦ f isdefinedbyλx= cxfor
any x∈ X.Clearly, f isaG-setmap.Letx∈ X andg∈ G.Then mg, f (x), f (gx)λx= m g, f (x), f (gx)cx= cxm g, f (x), f (gx)l−1(g, x, gx)l(g, x, gx) = l(g, x, gx)cgx= l(g, x, gx)λgx.
So λ: l→ m◦ f is anaturaltransformation and(f,λ) isamap ofC-monomialG-sets.
Thus, F (f,λ)= ϕ andF issurjective.Theinjectivityisclear,so F isfullyfaithful. 2 Proposition 3.Let G bea finite group. Then BC(G) is isomorphic to theGrothendieck
ring of thecategory CM G-set,forrelationsgiven bydecomposition intodisjointunions
of C-monomialG-sets andmultiplication inducedbyproductof C-monomialG-sets.
Proof. WeletB1
C(G) denotetheGrothendieckringofthecategoryCM G-set.The
equiv-alence F :CM G-set→CF G-set induces abijection F : BC1(G)→ BC(G) suchthat F (X, l)= [C×lX]
for any C-monomial G-set (X,l). Now we show that F is a ring homomorphism. Let (X1,l1) and(X2,l2) beC-monomialG-sets.Then
F (X1, l1) + (X1, l1) = F (X1, l1) (X1, l1) = F (X1 X2, l1 l2) = C×l1l2(X1 X2) = (X1, l1) (X2, l2) = [C×l1X1] + [C×l2X2].
FormultiplicativityofF we define amap f : C×l1×l2(X1× X2)→ (C ×l1X1)×C(C×l2X2) such that fc,(x1,x2) = (c,x1)×C (1,x2). Let (k,g) ∈ C × G and c,(x1,x2) ∈ C×l1×l2(X1× X2).Then (k, g)fc, (x1, x2) = (k, g)(c, x1)×C(1, x2) =(k, g)(c, x1)×C(1, g)(1, x2) =kcl1(g, x1, gx1), gx1 ×C l2(g, x2, gx2), gx2 = (kcl1(g, x1, gx1)l2(g, x2, gx2), gx1)×C(1, gx2) = fkcl1(g, x1, gx1)l2(g, x2, gx2), g(x1, x2) = f(k, g)(c, (x1, x2)) .
Sof isa(C× G)-mapandobviously, f isa(C× G)-isomorphism.Usingf weget F[X1, l1]· [X2, l2] = F[X1× X2, l1× l2] = C×l2×l2(X1× X2) = (C×l1X1)×C(C×l2X2) .
Thus,thedesiredresultfollows. 2
Remark 4.Let (X,l) be a C-monomial G-set. For all x ∈ X, we get a character lx :
Gx→ C defined bylx(g)= l(g,x,x) for g ∈ Gx. On theother hand givena subgroup
U of G anda grouphomomorphism μ : U → C we get aC-monomial G-set (G/U,μ) whereand μ : G/U → •C isthefunctorsuchthatgivengU,kU∈ G/U ifhgU = kU for
someg ∈ G then μ(h,gU,kU )= μ(k−1hg). Moreover, [U,μ]G and [G/U,μ] represents
thesameelement inBC(G).
2.3. The Lefschetzinvariantattachedtoamonomial G-poset
A G-poset X is a partially ordered set (X,≤) with a compatible G-action (that is
gx≤ gy wheneverg∈ G andx≤ y in X).A mapofG-posetsisaG-equivariantmapof posets.Wedenote byG-poset thecategoryoffinite G-posetsobtainedinthisway.
ThereisanobviousfunctorιG: G-set→ G-poset sendingeachfiniteG-settotheset
X orderedbytheequalityrelation, andeachG-equivariantmap toitself.
TheLefschetzinvariantattachedtoafiniteG-poset,whichisanelementofthe Burn-sideringof G hasbeenintroducedin[7] byThévenaz.WewilldefinesimilarlyaLefschetz invariantattachedtoaC-monomialG-posetasanelementoftheC-monomialBurnside ringofG.
2.3.1. The categoryof C-monomialG-posets
Given aG-posetX,weconsiderthecategoryX whose objectsaretheelementsofX
and givenx,y inX theset ofmorphismsfromx toy is
HomX(x, y) ={g ∈ G | gx ≤ y}. Now wedefineaC-monomialG-posetasfollows.
Definition 5.A C-monomial G-poset is a pair (X,l) consisting of a G-poset X and a functor l: X → •C.
In other words, foreach x,y ∈ X andg ∈ G such thatgx≤ y, wehave anelement l(g,x,y) ofC,withthepropertythatl(h,y,z)l(g,x,y)= l(hg,x,z) ifh∈ G andhy≤ z,
and l(1,x,x)= 1 foranyx∈ X.
Let(X,l) and(Y,m) beC-monomialG-posets.AmapofC-monomialG-posets from
(X,l) to(Y,m) is apair(f,λ): (X,l)→ (Y,m),wheref : X → Y isamap ofG-posets
and λ: l→ m◦ f is anaturaltransformation.Wedenote the category of C-monomial G-posetsbyCM G-poset.ProductanddisjointunionofC-monomialG-posetsaredefined
as for C-monomial G-sets. When C is the trivial group, we will identify the category
CM G-poset withG-poset.
Remark 6.If (X,l) is a C-monomial G-poset, then for any x∈ X we get a character lx: Gx→ C definedbylx(g)= l(g,x,x).Moreover,ifx≤ y,then
resGx
Gx∩Gylx= res
Gy
Gx∩Gyly
becausewehavethefollowingcommutativediagram:
l(x) l(1,x,y) l(g,x,x) l(y) l(g,y,y) l(x) l(1,x,y) l(y).
LetH be asubgroup ofG and(X,l) beaC-monomialH-set.WeletG×HX tobe
the quotient of G× X by the action of H. The set G×HX isa G-set via theaction
g(u,Hx)= (gu,Hx),forany g∈ G, and(u,Hx)∈ G×HX.Wedefine anorderrelation
≤ onG×HX as
∀(u,Hx), (v,Hy)∈ G ×HX, (u,Hx)≤ (v,Hy)⇔ ∃h ∈ H, u = vh, x ≤ h−1y.
G×HX =
g∈G/H
g×HX,
it’senoughtoconsider thechainsoftype(u,Hx0)< ...< (u,Hxn) in G×HX forsome
u∈ G andachainx0< ...< xn inX forsomen∈ N.
Let(u,Hx),(u,Hy) ∈ G×HX and g ∈ G suchthat g(u,Hx) ≤ (u,Hy).Then there
existsh∈ H suchthatgu= uh andhx≤ y.WedefinetheinducedC-monomialG-poset
IndGH(X,l) of(X,l) asthepair(G×HX,G×Hl) whereG×Hl:G×HX → •Cisdefined
by (G×Hl) g, (u,Hx), (u,Hy) = l(h, x, y). Nowshowthat(G×HX,G×Hl) isaC-monomialG-poset.
Let(u,Hx),(u,Hy),(u,Hz)∈ G×HX suchthat
g(u,Hx)≤ (u,Hy)
and
g(u,Hy)≤ (u,Hz)
forsomeg,g∈ G.Then thereexistsomeh,h ∈ H suchthat
gu = uh, gu = uh, hx≤ y, hy≤ z.
Thent= hh∈ H.Moreoverggu= uhh= ut andtx= hhx≤ z.Nowwe get (G×Hl) gg, (u,Hx), (u,Hz) = l(t, x, z) = l(hh, x, z) = l(h, y, z)l(h, x, y) = (G×Hl) g, (u,Hx), (u,Hy) (G×Hl)
g, (u,Hy), (u,Hz)
. Wealsohave(G×Hl) 1,(u,Hx),(u,Hx)
= 1 forany (u,Hx)∈ G×HX.Thus G×Hl
isafunctor. So IndGH(X,l) isaC-monomialG-poset.
GivenaC-monomialG-poset (Y,m),therestriction ResGH(Y,m) of(Y,m) is thepair (ResGHY,resG
Hm) where Res G
HY istherestrictionoftheG-posetY toH-posetand resGHm
istherestrictionofthefunctorm fromY to ResGHY .
Proposition7.LetG be afinitegroup.
1. If Y is a finite G-poset, denote by 1Y : Y → •C the trivial functor defined by
1Y(g,x,y)= 1 for any g ∈ G and x,y ∈ Y such that gx≤ y.Then theassignment
Y → (Y,1Y) isafunctorτG from G-poset to CM G-poset.
2. LetH be asubgroup of G.The assignment (X,l)→ IndGH(X,l) is afunctor IndGH : CM H-poset→CM G-poset, and theassignment (Y,m) → ResGH(Y,m) is a functor
3. Moreoverthediagrams H-poset Ind G H τH G-poset τG CM H-poset IndGH CM G-poset
and G-poset Res
G H τG H-poset τH CM G-poset ResGH CM H-poset
of categoriesandfunctorsare commutative.
Proof. 1. Letf : X→ Y beamap ofG-posets.Weset
τG(f ) = (f, 1f) : (X, 1X)→ (Y, 1Y),
where 1f : 1X → 1Y ◦ f is definedby1f x = 1 foranyx∈ X.Obviously(f,1f) isa
mapofC-monomialG-posets andτG is afunctor.
2. Let(f,λ): (X,l)→ (Y,m) beamap ofC-monomialH-posets.Weset thepair IndGH(f, λ) = (G×Hf, G×Hλ) : (G×HX, G×Hl)→ (G ×HY, G×Hm) where G×Hf : G×HX → G ×HY isdefinedby(G×Hf )(u,Hx)= u,Hf (x) and G×Hλ : G×Hl→ (G ×Hm)◦ (G ×Hf )
isdefinedby(G×Hλ)(u,Hx) = λxforany(u,Hx)∈ G×HX.It’sclearthatG×Hf isa
mapofC-monomialG-posets.NowweshowthatG×Hλ isanaturaltransformation.
Let (u,Hx),(u,Hy)∈ G×HX such that g(u,Hx) ≤ (u,Hy) for someg ∈ G. Then
gu= uh andhx≤ y forsomeh∈ H.Sinceλ: l→ m◦ f isanaturaltransformation, weget (G×Hm) g,u,Hf (x) ,u,Hf (y) (G×Hλ)(u,Hx) = m h, f (x), f (y)λx = λyl(h, x, y) = (G×Hλ)(u,Hy)(G×Hl) g, (u,Hx), (u,Hy) .
Nowconsider(idX,idl): (X,l)→ (X,l) where idX : X→ X istheidentity mapon
theH-setX and idl: l→ l◦ idX istheidentitytransformation.Thenweget
IndGH( idX, idl) = ( idG×HX, idG×Hl).
Now let (f,λ) : (X,l) → (Y,m) and (t,β) : (Y,m) → (Z,r) be the maps of
(G×Ht)◦ (G ×Hf ) = G×H(t◦ f)
and
(G×Hβ)◦ (G ×Hλ) = G×H(β◦ λ).
Thus,
IndGH(t, β)◦ IndGH(f, λ) = IndGH(t, β)◦ (f, λ).
So IndGH:CM H-poset→CM G-poset isafunctor.
Nowlet(f,λ): (X,l)→ (Y,m) beamapofC-monomialG-posets.Wesetthepair ResGH(f, λ) = (f|H, λ|H) : ( ResHGX, resGHl)→ ( ResGHY, resGHm)
where f|H: ResGHX → Res G
HY isdefinedastherestrictionofmapofG-posets f to
map of H-posetsand λ|H : resGHl→ resGHm◦ f|H is definedas therestrictionof λ.
Clearly,wegetthat ResGH :CM G-poset→CM H-poset isafunctor.
3. LetX beanH-poset.Commutativity ofthefirstdiagram followsfrom
τG◦ IndGH(X) = τG(G×HX) = (G×HX, 1G×HX) = Ind
G
H(X, 1X) = IndGH◦ τH(X).
NowletY be aG-poset.Commutativityoftheseconddiagramfollows from
τH◦ ResGH(Y ) = τH( ResGHY ) = ( ResGHY, 1ResG HY)
= ( ResGHY, resGH1Y) = ResHG(Y, 1Y) = ResGH◦ τH(Y ). 2
Proposition8.LetG beafinitegroupandH besubgroupofG.Thenthefunctor IndGH :
CM H-poset→CM G-poset isleftadjointtothefunctor (Y,m)→ ResGH(Y,m).
Proof. WeprovethatforanyC-monomialH-poset(X,l) andanyC-monomialG-poset
(Y,m) wehaveabijection HomCM G
IndGH(X, l), (Y, m) ∼= HomCM H
(X, l), ResGH(Y, m) naturalin(X,l) and(Y,m).
Wedefine
ϕ : HomCM G
IndGH(X, l), (Y, m)→ HomCM H
(X, l), ResGH(Y, m) where
suchthat
ϕ(f ) : X → ResGH(Y ) definedbyϕ(f )(x)= f (1,Hx) and
ϕ(λ) : l→ resm ◦ ϕ(f)
defined by ϕ(λ)x = λ(1,Hx) for any x∈ X. Obviously, ϕ(f ) is amap of H-posets. We
need toshowthat
ϕ(λ) : l→ resm ◦ ϕ(f)
is anaturaltransformation.Letx,y∈ X suchthatgx≤ y forsomeg∈ G. Then mh,ϕ(f )(x), ϕ(f )(y)ϕ(λ)x= m h, f (1,Hx), f (1,Hy) λ(1,Hx) = λ(1,Hy)l(h, x, y) = ϕ(λ)y(G×Hl) h, (1,Hx), (1,Hy) .
Wedefine aninversemap toϕ as
θ : HomCM H
(X, l), ResGH(Y, m)→ HomCM G
IndGH(X, l), (Y, m) where θ : (ψ, β)→θ(ψ), θ(β) suchthat θ(ψ) : G×HX→ Y
definedas θ(ψ)(u,Hx)= uψ(x) and
θ(β) : G×Hl→ m ◦ θ(ψ)
definedas
θ(β)(u,Hx)= m
u, ψ(x), uψ(x)βx
for any (u,Hx) ∈ G×HX.Obviously, themap θ(ψ) is amap ofG-posets.We needto
show that θ(β) is a naturaltransformation. Let (u,Hx),(u,Hy) ∈ G×H X such that
g(u,Hx)≤ (u,Hy) forsomeg∈ G.Thenthereexistssomeh∈ H suchthatgu= uh and
mg, θ(ψ)(u,Hx), θ(ψ)(u,Hy)
θ(β)(u,Hx) = m
g, uψ(x), uψ(y)mu, ψ(x), uψ(x)βx
= mu, ψ(y), uψ(y)mh, ψ(x), ψ(y)βx= m
u, ψ(y), uψ(y)βyl(h, x, y) = θ(β)(u,Hy)(G×Hl) h, (uHx), (u,Hy) .
Clearly,ϕ andθ aremutual inversemaps, andnaturalin(X,l) and(Y,m). 2
2.3.2. The LefschetzinvariantattachedtoaC-monomialG-poset
Let (X,l) be a C-monomial G-poset.The Lefschetz invariant Λ(X,l) of (X,l) is the
elementofBC(G) definedby Λ(X,l)= x0<...<xn∈GX (−1)n Gx0,...,xn, Res Gx0 Gx0,...,xn(lx0) G
wherex0< ...< xnrunsoverG-representativesofthechainsinX.ThegroupGx0,...,xnis
thestabilizerof theset {x0,...,xn},thatisGx0,...,xn =∩
n
i=0Gxi.Here Res
Gx0
Gx0,...,xn(lx0)
denotes the restriction of the character lx0 introduced in Remark 4. Observe that if
x0< ...< xn isachaininX forsomen∈ N, byRemark6wehave
ResGx0
Gx0,...,xnlx0= Res Gxi Gx0,...,xnlxi
forany0≤ i≤ n.
Let (X,l) be a C-monomial G-poset. Given n ∈ N, let Sdn(X) denote the set of
chainsinX withordern+ 1.Obviously,theset Sdn(X) isaG-set.Then(Sdn(X),ln)
isaC-monomialG-set whereln: Sdn(X)→ •C isthefunctordefinedby
ln(g, x0< ... < xn, y0< ... < yn) = l(g, x0, y0)
foranyx0< ...< xn,andy0< ...< yn in Sdn(X) suchthat
g(x0< ... < xn) = y0< ... < yn
forsomeg∈ G.
Remark9.Given a C-monomial G-poset (X,l), we have the following isomorphism of monomialG-sets: Sdn(X), ln ∼=
x0<...<xn∈GSdn(X) G/Gx0,...,xn, ResGx0 Gx0,...,xn(lx0) foranyn∈ N.
Proof. Let[G/Sdn(X)] be asetof representative oftheG-actionon Sdn(X).Let x=
x0 < ... < xn be a chain in Sdn(X) then there exist some gx ∈ G and a unique
(f, λ) :Sdn(X), ln →
x0<...<xn∈GSdn(X) G/Gx0,...,xn, ResGx0 Gx0,...,xn(lx0)
where f (x)= gxGσx∈ G/Gσx andλx= l(gx−1,gxσx0,σx0).Obviously,
f : Sdn(X)→
x0<...<xn∈GSdn(X)
G/Gx0,...,xn
is anisomorphismofG-sets.Weshow that
λ : ln→
x0<...<xn∈GSdn(X)
ResGx0
Gx0,...,xn(lx0)◦ f
is anaturaltransformation.Letx= x0 < ...< xn, and y = y0< ...< yn be sequences
in Sdn(X) suchthatgx= y forsomeg∈ G.Thereexistauniqueσx,σy∈ [G/Sdn(X)]
such that x = gxσx and y = gyσy for some gx and gy in G. Then x0 = gxσx0 and
y0= gyσy0 soy0= gx0= ggxσx0.Thus,byuniquenessσx0 = σy0 andsog−1y ggx∈ Gσx0.
Then settingr = ResGx0
Gx (lx0)
g,f (x),f (y)λx,wehavethat
r = lx0(g, gxGσx, gyGσy)l(g −1 x , gxσx0, σx0) = lx0(g −1 y ggx)l(gx−1, gxσx0, σx0) = l(gy−1ggx, x0, x0)l(gx−1, gxσx0, σx0) = l(gx, σx0, gxσx0)l(g, gxσx0, ggxσx0)l(g−1y , ggxσx0, σx0)l(gx−1, gxσx0, σx0) = l(g, x0, y0)l(g−1y , gyσy0, σy0) = ln(g, x, y)λy. 2
ByRemark9,theLefschetzinvariantofaC-monomialG-set(X,l) canbewritten as Λ(X,l)= x0<...<xn∈GX (−1)n Gx0,...xn, Res Gx0 Gx0,...,xn(lx0) G= n∈N (−1)nSdn(X), ln .
It followsthatΛX = ΛτG(X),where ΛX theLefschetzinvariantoftheG-posetX
intro-duced in [2].
Wedefine similarlythereduced Lefschetzinvariant of(X,l)
Λ(X,l)= Λ(X,l)− [G, 1G]G
where 1G is thetrivialcharacterofG.
Lemma 10.LetG be afinite groupandC bean abelian group.
1. Let(X,l) bea C-monomialG-set,viewed as a C-monomial G-posetordered by the equalityrelationonX. ThenΛ(X,l)= [C×lX] inBC(G).
2. Let (X,l) and(Y,m) be C-monomialG-posets.Then Λ(XY,r) = Λ(X,l)+ Λ(Y,m) in
BC(G).
3. Given C-monomialG-posets (X,l) and(Y,m), we have Λ(X×Y,l×m) = Λ(X,l)Λ(Y,m)
in BC(G).
Proof. 1.and2.areclear.
3.Inthefollowingproof usingtheinclusion
BC(G) → Q ⊗ZBC(G)
we identify the elements of BC(G) with their image in Q⊗ZBC(G). We start with
rearrangingthechainsinX×Y asintheproofofLemma 11.2.9in [2].Letn∈ N.Given achain z = z0 < ... < zn inX × Y projection of z on X is denotedby zX and on Y
is denotedby zY. Then zX is achain inX with order i+ 1 for somei ≤ n and zY is
achaininY withorder j + 1 for somej ≤ n such thati+ j = n. Letsi be thechain
s0< ...< si andtj bethechaint0< ...< tj.Now
Λ(X×Y,l×m)= n∈N, z∈GSdn(X×Y ) (−1)n Gz, Res Gz0 Gz (lz0) G = n∈N, z∈ Sdn(X×Y ) (−1)n|Gz| |G|[Gz, Res Gz0 Gz (lz0)]G = i,j∈N si∈X tj∈Y Γsi,tj where Γsi,tj = n∈N z∈ Sdn(X×Y ):zX=si, zY=tj (−1)n|Gsi∩ Gtj| |G| Gsi∩ Gtj, Res Gs0 Gsi(ls0) Res Gt0 Gtj(mt0) G = |Gsi∩ Gtj| |G| Gsi∩ Gtj, Res Gs0 Gsi(ls0) Res Gt0 Gtj(mt0) G n∈N z∈ Sdn(X×Y ):zX=si, zY=tj (−1)n = |Gsi∩ Gtj| |G| Gsi∩ Gtj, Res Gs0 Gsi(ls0) Res Gt0 Gtj(mt0) G(−1) i+j. Now, Λ(X×Y,l×m)= i,j∈N si∈X tj∈Y (−1)i+j|Gsi∩ Gtj| |G| Gsi∩ Gtj, Res Gs0 Gsi(ls0) Res Gt0 Gtj(mt0) G.
Ontheother hand Λ(X,l)Λ(Y,m)= i∈N si∈GX (−1) Gsi, Res Gs0 Gsi(ls0) j∈N tj∈GY (−1)j Gtj, Res Gt0 Gtj(mt0) G = i,j∈N si∈X tj∈Y GigGtj⊆G (−1)i+j|Gsi||Gtj| |G|2 Gsi∩ gG tj, Res Gs0 Gsi(ls0) Res gG t0 gG tj( gm t0) G = i,j∈N si∈X tj∈Y g∈G (−1)i+j |Gsi||Gtj| |G|2|G sigGtj| Gsi∩ gG tj, Res Gs0 Gsi(ls0) Res gG t0 gG tj( gm t0) G = i,j∈N si∈X tj∈Y g∈G (−1)i+j|Gsi∩ gG tj| |G|2 Gsi∩ gG tj, Res Gs0 Gsi(ls0) Res gG t0 gG tj( gm t0) G = i,j∈N si∈X tj∈Y g∈G (−1)i+j|Gsi∩ Ggtj| |G|2 Gsi∩ Ggtj, Res Gs0 Gsi(ls0) Res Ggt0 Ggtj(mgt0) G = i,j∈N si∈X tj∈Y (−1)i+j|Gsi∩ Gtj| |G| Gsi∩ Gtj, Res Gs0 Gsi(ls0) Res Gt0 Gtj(mt0) G.
Thus, Λ(X×Y,l×m)= Λ(X,l)Λ(Y,m). 2
The first assertionof Lemma 10tells us thatevery positiveelement of BC(G) is in
of the form Λ(X,l) for some C-monomial G-poset (X,l). Now consider the poset X =
{a,b,c,d,e} with theordering{a≤ c,a≤ d,a≤ e,b≤ c,b≤ d,b≤ e}.Considertrivial
G-actionon X.Then ΛτG(X)=−1BC(G). SoasaconsequenceofLemma10weget the
following corollary.
Corollary 11. Any element of themonomial Burnside ring can be expressed as the Lef-schetz invariantofsome (nonunique)monomial G-poset.
Proposition12. LetH beasubgroupof G.GivenaC-monomialH-poset(X,l),wehave
IndGH(Λ(X,l)) = ΛIndG H(X,l).
Proof. Since IndGH(Λ(X,l)) = n∈N (−1)nIndGHSdn(X), ln ,
weneedtoshowthatthereexists aC-monomialG-setisomorphismbetween G×H Sdn(X), G×Hln and Sdn(G×HX), (G×Hl)n foranyn∈ N. Wedefine (fn, id) : G×H Sdn(X), G×Hln →Sdn(G×HX), (G×Hl)n where fn : G×H Sdn(X)→ Sdn(G×HX) suchthat fn(u,Hx0< ... < xn) = (u,Hx0) < ... < (u,Hxn) foranychain(u,Hx0< ...< xn) inG×H Sdn(X).
Let(u0,Hx0)< ...< (un,Hxn) beachainin Sdn(G×HX).There existsomehi∈ H
suchthatuihi = ui+1 andhi−1xi< xi+1 forall0≤ i≤ n− 1.Then
fn
u0,Hx0< h0x1< ... < h0...hn−1xn
= (u0,Hx0) < ... < (un,Hxn).
Obviously,fn isamap ofG-setsandinjective.
Now, we show thatG×Hln = (G×Hl)n◦ fn. We consider an element k∈ G, and
chains(u,Hx0< ...< xn) inG×H Sdn(X) suchthat
k(u,Hx0< ... < xn) = (v,Hy0< ... < yn).
Thereexists someh∈ H suchthatku= vh and hxi= yi forall0≤ i≤ n.Then
(G×Hl)n k, fn(u,Hx0< ... < xn), fn(v,Hy0< ... < yn) = (G×Hl)n k, (u,Hx0) < ... < (u,Hxn), (v,Hy0) < ... < (v,Hyn) = ln(h, x0< ... < xn, y0< ... < yn) = (G×Hln) k, (u,Hx0< ... < xn), (v,Hy0< ... < yn) . 2
Let (X,l) beaG-poset andletx∈ X.Then thepairs(]x,·[X,l>x) and (]·,x[X,l<x)
are C-monomialGx-posetswhere
]x,·[X={y ∈ X | x < y}, ]·, x[X={y ∈ X | y < x}
whichareGx-posetsandl>x: ]x,·[X→ •C andl<x: ]·, x[X → •C aretherestrictionsof
thefunctorl.
Lemma 13.Let(X,l) be amonomial G-poset. Wehave
Λ(X,l)=− x∈[G/X] IndGGx[Gx, lx]Gx· Λ]x,·[X . Proof. Λ(X,l)= x0<...<xn∈GX (−1)n Gx0,...,xn, Res Gx0 Gx0,...,xn(lx0) G = x0∈GX x1<...<xn∈GX:x0<x1 (−1)n Gx0,...,xn, Res Gx0 Gx0,...,xn(lx0) G = x0∈GX IndGGx0 x1<...<xn∈Gx0]x0,·[X (−1)n Gx0,...,xn, Res Gx0 Gx0,...,xn(lx0) Gx0 = x0∈GX IndGGx0 Gx0, lx0 Gx0 x1<...<xn∈Gx0 ]x0,·[X (−1)n Gx0,...,xn, 1Gx0,..,xn Gx0 =− x∈GX IndGGx[Gx, lx]Gx· Λ]x,·[X . 2
Remark 14.Wecandefine theopposite ofaC-monomial G-poset (X,l) as follows. We consider the pair (Xop,lop) where Xop is the opposite G-poset with the order ≤op
definedby
∀x, y ∈ X, g ∈ G, gx ≤opy⇔ y ≤ gx
and lop: Xop→ •
C isdefinedby
lop(g, x, y) = l−1(g−1, y, x)
for any x,y ∈ Xop and g ∈ G such that gx ≤op y. Obviously, the pair (Xop,lop) is a C-monomial G-poset. Moreover the assignment (X,l) → (Xop,lop) is a functor
CM G-poset→CM G-poset:if(f,λ): (X,l)→ (Y,m) isamapofC-monomialG-posets,
thenf : Xop→ Y opisamapofG-posetsandforanygx≤opx,wegetthecommutative
l(x) λx lop(g,x,x) m◦ f(x) mopg,f (x),f (x) l(x) λx m◦ f(x). Observethat(lop)
x(g)= l−1(g−1,x,x)= l(g,x,x)= lx(g),foranyx∈ X andg∈ Gx.
ItfollowsthatΛ(X,l)= Λ(Xop,lop).
Let(f,λ): (X,l)→ (Y,m) be amap ofC-monomialG-posets.Given y ∈ Y , follow-ing [3] we set
fy ={x ∈ X | f(x) ≤ y}, fy={x ∈ X | f(x) ≥ y}
which are both Gy-posets. We denote by (fy,l|fy) the C-monomial Gy-poset where
l|fy : fy → •C istherestrictionofthe functorl. Similarly,wedenote by(fy,l|fy) tobe
C-monomialGy-posetwhere l|fy : fy→ •C istherestrictionofthefunctor l.
Example15.Let(f,λ): (X,l)→ (Y,m) beamap ofC-monomialG-posets.Wedefinea
G-posetX∗f,λY withunderlyingG-setX Y asfollows:forz,z∈ X Y ,weset
z≤ z⇔ ⎧ ⎪ ⎪ ⎨ ⎪ ⎪ ⎩ z, z∈ X and z≤ z ∈ X z, z∈ Y and z≤ z ∈ Y z∈ X, z∈ Y and f(z) ≤ z∈ Y .
Wedefinethefunctorl∗f,λm: X Y → •C by
(l∗f,λm)(g, z, z) = ⎧ ⎪ ⎪ ⎨ ⎪ ⎪ ⎩ l(g, z, z) if z, z∈ X m(g, z, z) if z, z∈ Y m(g, f (z), z)λz if z∈ X, z∈ Y ,
foranyz,z∈ X ∗f,λY andg∈ G suchthatgz≤ z.
Nowletz1,z2,z3∈ X ∗f,λY andg,g∈ G suchthatgz1≤ z2 andgz2≤ z3.Weaim
toshowthat
(l∗f,λm)(gg, z1, z3) = (l∗f,λm)(g, z2, z3)(l∗f,λm)(g, z1, z2).
Wehavefour casesto consider: • z1,z2,z3∈ X,
• z1∈ X andz2, z3∈ Y ,
• z1,z2,z3∈ Y .
Inthefirstcaseweget
(l∗f,λm)(gg, z1, z3) = l(gg, z1, z3) = l(g, z2, z3)l(g, z1, z2)
= (l∗f,λm)(g, z2, z3)(l∗f,λm)(g, z1, z2).
Inthesecond case,using thenaturalityofλ weget
(l∗f,λm)(gg, z1, z3) = m gg, f (z1), z3 λz1 = m g, f (z2), z3 mg, f (z1), f (z2) λz1 = l(g, z1, z2)m g, f (z2), z3 λz2 = (l∗f,λm)(g , z 2, z3)(l∗f,λm)(g, z1, z2).
Inthethirdcase,weget (l∗f,λm)(gg, z1, z3) = m gg, f (z1), z3 λz1 = m g, f (z2), z3 mg, f (z1), f (z2) λz1 = (l∗f,λm)(g, z1, z2)(l∗f,λm)(g, z2, z3).
Inthefourthcase
(l∗f,λm)(gg, z1, z3) = m gg, z1, z3 = m(g, z2, z3)m(g, z1, z2) = (l∗f,λm)(g, z2, z3)(l∗f,λm)(g, z1, z2).
Letz∈ X ∗f,λY thenobviously wehave(l∗f,λm)(1,z,z)= 1.Thus,(X∗f,λY,l∗f,λm)
is aC-monomialG-poset.
Lemma 16. Let (f,λ) : (X,l) → (Y,m) be a map of C-monomial G-posets. Then
Λ(X∗f,λY,l∗f,λm)= Λ(Y,m).
Proof. 1. Letz∈ Z = X ∗f,λY .Ifz∈ X considerthemapg : ]z,·[Z→ [f(z),·[Y defined
by
g(t) =
f (t) if t∈ X
t if t∈ Y .
Letg: f (z),· →z,· definedbyg(s)= s.Then g andg aremapsofGz-posets
such that g◦ g = Id and Id ≤ g ◦ g. So if z ∈ X using [[3], Lemma 4.2.4 and Proposition 4.2.5],weget Λ]z,·[= Λ[f (z),·[= 0.Thus,
Λ(X∗f,λY,l∗f,λm)=− z∈[G\X∗f,λY ] IndGGz([Gz, lz]Gz· Λ]z,·[) =− y∈[G\Y ]
IndGGy([Gy, ly]Gy · Λ]y,·[) = Λ(Y,m). 2
Asaconsequence,wegiveananalogueofProposition 4.2.7 in [3], whichinturn was inspiredbyamuchdeepertheoremof Quillenin [6].
Proposition17. Let(f,λ): (X,l)→ (Y,m) be amapof C-monomialG-posets.Then in BC(G) Λ(Y,m)= Λ(X,l)+ y∈G\Y IndGGy( Λfy Λ(]y,·[ Y,m>y)). Λ(Y,m)= Λ(X,l)+ y∈G\Y
IndGGy( Λfy Λ(]·,y[Y,m<y)).
Proof. We follow the proof of Proposition 4.2.7 in [3]. For any n ∈ N, any chain z = z0< ...< zn∈ Sdn(X∗f,λY ) canbeoftwotypes,dependingonzn∈ X orzn ∈ Y .For
asequence z ofthefirsttypeweget
ResGzn
Gz0,...,zn(l∗f,λm)nzn= Res
Gzn
Gz0,...,znln.
Nowasequence z of thesecond typehas asmallest element y = zi inY , thus, wecan
writethesequenceas
x0< ... < xi−1< y < y0< ... < yn−i−1
suchthatx0< ...< xi−1 isin Sdi−1(fy),and y0< ...< yn−i−1 isin Sdn−i−1(]y,·[Y).
Weget
ResGzn
Gz0,...,zn(l∗f,λm)nzn= Res
Gy
Gz0,...,zn(m).
Letxi−1denotethechainx0< ...< xi−1andyn−i−1denotethechainy0< ...< yn−i−1.
Then,byLemma10andLemma16weget Λ(Y,m)= Λ(X∗f,λY,l∗f,λm)= n∈N (−1)nSdn(X∗f,λY ), (l∗f,λm)n = n∈N z0<...<zn∈ Sdn(l∗f,λm) (−1)n Gz0,...,zn, Res Gzn Gz0,...,zn(l∗f,λm)n G = n∈N (−1)nSdn(X), ln
+ y∈[G\Y ] IndGGy n i=0 xi−1∈ Sdi−1(fy) yn−i−1∈ Sdn−i−1(]y,·[Gy)
Gxi−1,y,yn−i−1, Res
Gy Gxi−1,y,yn−i−1my Gy = Λ(X,l)+ y∈[G\Y ] IndGGy Λ(fy,1 f y) Λ(]y,·[Y,m>y) .
Forthesecond assertionwe considertheoppositemap (f, λ) : (Xop, lop)→ (Yop, mop)
Since wehaveΛ(X,l)= Λ(Xop,lop) byRemark14,theresultfollows. 2
Corollary 18.Let(f,λ): (X,l)→ (Y,m) beamapof C-monomialG-posets.If Λfy = 0
forally∈ Y (resp.if Λfy = 0 forally∈ Y ),then ΛX,l= ΛY,m.
Remark 19.The assumption of this corollary is fulfilled in particular if f : X → Y
admitsarightadjointg,inotherwordsifthere existsamap ofposetsg : Y → X such
thatf (x)≤ y ⇔ x≤ g(y) foranyx∈ X andy∈ Y , i.e.equivalentlyiff◦ g(y)≤ y and g◦ f(x)≤ x foranyx∈ X andanyy∈ Y .
Now we set some notation. Given a C-monomial G-set (X,l), we can rewrite its Lefschetz invariantas Λ(X,l)= x0<...<xn∈GX (−1)n Gx0,...xn, Res Gx0 Gx0,...,xn(lx0) G = (V,ν)∈Gch(G) γV,νX,l[V, ν]G where γV,νX,l= x0<...<xn∈GX (Gx0,...,xn, ResGx0Gx0,...,xnlx0)=G(V,ν) (−1)n = 1 |NG(V, ν) : V| x0<...<xn∈X (Gx0,...,xn, ResGx0Gx0,...,xnlx0)=(V,ν) (−1)n.
Given aC-monomialG-poset (X,l) welettheset(X,l)U,μ tobe
(X, l)U,μ={x ∈ XU | ResGx
U lx= μ}
where (U,μ) is a subcharacter of G. Then given a C-subcharacter (U,μ) ∈ ch(G) we
χ(X, l)U,μ= n∈N x0<...<xn∈XU ResGx0Gx0,...,xnlx0=μ (−1)n= (V,ν)∈ ch(G) U⊆V ResV Uν=μ mX,lV,ν where mX,lV,ν = n∈N x0<...<xn∈X (Gx0,...,xn, ResGx0Gx0,...,xnlx0)=(V,ν) (−1)n. Now|NG(V,ν): V|mX,lV,ν= γ X,l
V,ν.Usingthis factweprovethefollowinglemma.
Lemma 20. Let(X,l) and (Y,m) be C-monomial G-posets then Λ(X,l) = Λ(Y,m) if and
onlyif χ(X,l)U,μ= χ(Y,m)U,μ forevery C-subcharacter (U,μ) of G.
Proof. AssumeΛ(X,l)= Λ(Y,m).Then
(V,ν)∈Gch(G) γV,νX,l[V, ν]G = (V,ν)∈Gch(G) γV,νY,m[V, ν]G (V,ν)∈Gch(G) (γV,νX,l− γV,νY,m)[V, ν]G= 0.
SoγV,νX,l = γV,νY,mandthen mV,nuX,l = mY,mV,nuforeveryC-subcharacter(V,ν) ofG.Weget
(U,μ)≤(V,ν)∈Gch(G)
mX,lV,ν =
(U,μ)≤(V,ν)∈Gch(G)
mY,mV,ν.
Thus,χ(X,l)U,μ= χ(Y,m)U,μforeveryC-subcharacter(U,μ) ofG.
Conversely,assumethatχ(X,l)U,μ= χ(Y,m)U,μforeveryC-subcharacter(U,μ)
ofG.Then (U,μ)≤(V,ν)∈ ch(G) mX,lV,ν = (U,μ)≤(V,ν)∈ ch(G) mY,mV,ν, (U,μ)≤(V,ν)∈ ch(G) (mX,lV,ν− mY,mV,ν) = 0.
Letz bethematrixwith thecoefficients
z(U, μ; V, ν) =(U, μ) ≤ (V, ν) =
1 if (U, μ)≤ (V, ν) 0 otherwise
forany C-subcharacters (U,μ),(V,ν). Ifwe list theC-subcharacters innon-decreasing order of size of thesubgroups, the matrixz is uppertriangular with nonzero diagonal
coefficients. Thus, z isnonsingular andso mX,lV,ν = mV,νY,m. Thisimplies γX,lV,ν = γV,νY,m. We get Λ(X,l)= (V,ν)∈Gch(G) γV,νX,l[V, ν]G= (V,ν)∈Gch(G) γV,νY,m[V, ν]G= Λ(Y,m).
This provesthelemma. 2 3. Generalizedtensorinduction
LetG andH befinitegroups.A setU isa(G,H)-biset ifU isaleftG-setandright
H-set such thattheG-actionand the H-actioncommute.Any (G,H)-biset U is aleft (G× H)-setwith thefollowing action:
∀u ∈ U, (g, h) ∈ G × H (g, h) · u = guh−1.
AC-monomial(G× H)-set(U,λ) willbecalledaC-monomial(G,H)-biset,andusually denotedbyUλ forsimplicity.
Now letUλ beaC-monomial(G× H)-setandu,u∈ U.Thenthesetofmorphisms
from u tou inU is
HomU(u, u) ={(g, h) ∈ G × H | gu = uh}.
If(g,h)∈ HomU(u,u),wedenote theimageof(g,h) underλ byλ(g,h,u,u).
LetUλbeaC-monomial(G,H)-bisetandVρbeaC-monomial(H,K)-biset.Consider
theset
Uλ◦ Vρ={(u, v) ∈ U × V | ∀h ∈ Hu∩ Hv, λ(1, h, u, u)ρ(h, 1, v, v) = 1}.
Theset Uλ◦ Vρ isanH-setwiththeaction
∀(u, v) ∈ Uλ◦ Vρ,∀h ∈ H, h(u, v) = (uh−1, hv).
Indeed, theconditionthatweimpose onUλ◦ Vρ amountstosaying thatgiven(u,v)∈
Uλ◦ Vρ, the linear character ξu,v : h → λ(1,h,u,u)ρ(h,1,v,v) of Hu∩ Hv is trivial.
Moreover we haveξux,x−1v(h)= ξu,v(xhx−1)= 1 for x∈ H and h∈ Hux∩ Hx−1v,i.e.
xhx−1 ∈ Hu∩ Hv.
WeletUλ◦HVρdenotethesetofH-orbitsonUλ◦ Vρ and(u,Hv) denotetheH-orbit
containing(u,v).Theset Uλ◦HVρ is(G,K)-bisetwiththeaction
(u,Hv)∈ Uλ◦HVρ, (g, k)∈ G × K, g(u,Hv)k = (gu,Hvk).
WeobtainaC-monomial(G,K)-biset(Uλ◦HVρ,λ×ρ),whereλ×ρ isdefinedasfollows:
thenthere exists h∈ H such thatgu= uh and hv = vk. This element h neednotbe unique,butitiswelldefineduptomultiplicationontherightbyanelementofHu∩ Hv.
Weset
(λ× ρ)g, k, (u,Hv), (u,Hv) = λ(g, h, u, u)ρ(h, k, v, v),
whichdoesnotdependonthechoiceofh,bythedefiningpropertyofUλ◦ Vρ.Notethat
Uλ◦HVρ= U×HV whenV isaleftfree(H,K)-biset,orwhenλ andρ arebothequal
tothetrivialfunctor.
GivenaC-monomialG-poset(X,l),welettU,λ(X,l) bethesetofG-equivariantmaps
f : U→ X suchthat
lg, f (u), f (u)= λ(g, 1, u, u)
forallu∈ U andg∈ Gu.ThentU,λ(X,l) isanH-posetwiththeaction(hf )(u)= f (uh),
foranyh∈ H,foranyf ∈ tU,λ(X,l),foranyu∈ U.Theorder≤ isgivenas follows:
∀f, f∈ t
U,λ(X, l), f ≤ f ⇔ ∀u ∈ U, f(u) ≤ f(u) in X.
NowwedefineafunctorLU,λ:tU,λ(X, l)→ •C.Letf ,f∈ tU,λ(X,l) andh∈ H such
thathf ≤ f. Wechoose a set [G\U] of representatives of G-orbits of U . Then for all
u∈ U thereexistsomegh,u∈ G andauniqueσh(u)∈ [G\U] suchthat
uh = gh,uσh(u).
Sincehf ≤ f,wegetgh,uf
σh(u)
≤ f(u),and weset
LU,λ(h, f, f) = u∈[G\U] l gh,u, f σh(u) , fuλ−1gh,u, h, σh(u), u .
Nowweshowthatthisdefinitiondoesnotdependonthechoiceofgh,u.Assumethat
thereexist gh,u,gh,u∈ G suchthat
uh = gh,uσh(u) = gh,u σh(u).
Sothereexistsw∈ Gσh(u) suchthatgh,u= g
h,uw. Weget l w, fσh(u) , fσh(u) = λw, 1, σh(u), σh(u) .
Furthermore,wegetthefollowingcommutativediagram:
σh(u) w gh,u σh(u) gh,u u
Thus, LU,λ(h, f, f) = u∈[G\U] l gh,u, f σh(u) , fuλ−1gh,u, h, σh(u), u = u∈[G\U] l gh,u w, fσh(u) , fuλ−1gh,u w, h, σh(u), u = u∈[G\U] l gh,u , fσh(u) , fuλ−1gh,u, h, σh(u), u .
Definition 21.The above construction TU,λ : (X,l) →
tU,λ(X,l),LU,λ
is called the
generalized tensorinduction forC-monomialG-posets,associatedto (U,λ).
Lemma 22. LetG and K befinite groupsand U be a(G,K)-biset. Thenthere exists a bijection betweenthesets{(u,t)| u∈ [G\U/K],t∈ [(K ∩ Gu)\K]} and [G\U].
Proof. Let u ∈ [G\U/K] andt ∈ [(K ∩ Gu)\K] then there exist somegt,u ∈ G and a
uniqueσt(u)∈ [G\U] suchthat
ut = gt,uσt(u).
Wedefineψ :{(u,t)| u∈ [G\U/K], t∈ [(K ∩ Gu)\K]}→ [G\U] byψ(u,t)= σ
t(u). 2
Lemma 23. LetG and H befinite groups, (U,λ) be amonomial (G,H)-biset and(X,l)
be aC-monomialG-poset. 1. tU,λ(X,l),LU,λ is aC-monomialH-poset. 2. tU,λ(X,l),LU,λ
does not depend on the choice of representative set [G\U], up to isomorphism.
Proof. 1. We show that LU,λ : tU,λ(X, l) → •C is a functor. Let h, h ∈ H and
f,f, f ∈ tU,λ(X,l) such that hf ≤ f and hf ≤ f. Let u ∈ [G\U]. Then
there exist somegh,u,gh,u, ghh,u inG anduniqueelements σh(u), σh(u), σhh(u)
in[G\U] such that
uh = gh,uσh(u), uh= gh,uσh(u), uhh = ghh,uσhh(u).
Alsothereexist somegh,σh(u)∈ G andauniqueσh
σh(u)
∈ [G\U] suchthat
σh(u)h = gh,σh(u)σh σh(u) . Nowweget uhh = gh,ugh,σh(u)σh σh(u)
and
σhh(u) = σh
σh(u).
Thenthere existsw∈ Gσh h(u) suchthat
ghh,u= gh,ugh,σh(u)w.
Wehavethefollowingcommutativediagram:
σhh(u) w gh h,u σhh(u) gh ,ugh,σh(u) uhh
Ontheotherhandsincew∈ Gσh h(u),weget
l w, fσhh(u) , fσhh(u) = λw, 1, σhh(u), σhh(u) .
Thus,setting L= LU,λ(hh,f,f),wehave
L = u∈[G\U] l ghh,u, f σhh(u) , fuλ−1ghh,u, hh, σhh(u), u = u∈[G\U] l gh,ugh,σh(u)w, f σhh(u) , fuλ−1gh,ugh,σh(u)w, h h, σ hh(u), u = u∈[G\U] l gh,ugh,σh(u), f σhh(u) , fuλ−1gh,ugh,σh(u), h h, σ hh(u), u = L(h, f, f)L(h, f, f).
Moreover,given f ∈ TU,λ(X,l) wehave
L(1, f, f ) =
u∈[G\U]
l1, f (u), f (u)λ−1(1, 1, u, u) = 1.
Thus,LU,λ:tU,λ(X, l)→ •C isafunctor.
2. Let h∈ H and f ,f ∈ tU,λ(X,l) such thathf ≤ f. LetS = [G\U] and letS be
theanotherchoiceofrepresentatives.Ifu∈ S thenthereexistsomeau∈ G,anda
uniqueu∈ S suchthatu= auu.Then thereexist somegh,auu,gh,u∈ G, a unique
σh(auu)∈ S,and auniqueσh(u)∈ S suchthat
auuh = gh,auuσ
h(auu)
and
uh = gh,uσh(u).
Then
auuh = augh,uσh(u) = augh,ua−1σh(u)aσh(u)σh(u).
So σh(auu)= aσh(u)σh(u). Note that aσh(u)σh(u) ∈ S
.We get the following
com-mutativediagram: aσh(u)f σh(u) augh,ua−1σh(u) a−1σh(u) auf u fσh(u) gh,u f (u). au
Thus,settingL= LU,λ(h,f,f),wehave
L = auu∈S l augh,ua−1σh(u), f aσh(u)σh(u) , fauu
λ−1augh,ua−1σh(u), h, aσh(u)σh(u), auu
= LU,λ(h, f, f) = LU,λ(h, f, f)αfα−1f where αf = u∈S lau, f(u), auf(u) λ−1au, 1, u, auu and α−1f = u∈S la−1u , auf (u), f (u) λ−1a−1u , 1, auu, u . 2
Proposition 24.LetG andH be finitegroupsand(U,λ) beaC-monomial(G,H)-biset.
1. Let(X,l),(X,l) be C-monomialG-posetsthen
TU,λ
(X, l)× (X, l) ∼= TU,λ(X, l)× TU,λ(X, l).
Proof. 1. is clear.
2. Let(ϕ,β): (X,l)→ (Y,m) beamap ofC-monomial G-posets.Wedefineamap of
C-monomialG-posets TU,λ(ϕ), TU,λ(β) :tU,λ(X, l), L →tU,λ(Y, m), M where
TU,λ(ϕ) : tU,λ(X, l)→ tU,λ(Y, m)
suchthatTU,λ(ϕ)(f )= ϕ◦ f and
TU,λ(β) : L(f )→ M ◦ TU,λ(ϕ)(f ) suchthat TU,λ(β) = u∈[G\U] βf (u)
for any f ∈ tU,λ(X,l). Clearly, ϕ◦ f : U → X → Y is a map of G-posets. Since
given g ∈ Gu and u∈ U the map β : l→ m◦ ϕ is natural,we have thefollowing
commutativediagram: lf (u) βf (u) l(g,f (u),f (u)) m◦ ϕf (u) m(g,ϕ◦f(u),ϕ◦f(u)) lf (u) βf (u) m◦ ϕf (u). So βf (u)l g, fu, fu= m
g, ϕf (u), ϕf (u)βf (u).
Sinceg∈ Gf (u),wehave
l g, fu, fu= λ(g, 1, u, u). Thenweget m
g, ϕf (u), ϕf (u)= λ(g, 1, u, u). Thus,ϕ◦ f ∈ tU,λ(Y,m).
Now weshowthat
TU,λ(β) : L→ M ◦ TU,λ(ϕ)
is anatural transformation. Let f,f ∈ tU,λ(X,l) and h ∈ H such that hf ≤ f. We
show thatthefollowingdiagram iscommutative: Lf TU,λ(β)f L(h,f,f) M◦ TU,λ(ϕ)(f ) M(h,ϕ◦f,ϕ◦f) Lf TU,λ(β)f M◦ TU,λ(ϕ)(f).
Letu∈ [G\U].Thenthereexistsomegh,u∈ G andauniqueσh(u)∈ [G\U] suchthat
uh = gh,uσh(u).
Since β : l→ m◦ ϕ is a naturaltransformation,we obtain the following commutative diagram:
l
fσh(u)
βf (σh(u)) lgh,u, f (σh(u)), f(u)
m◦ ϕ fσh(u) m gh,u, ϕ fσh(u) , ϕfu lf(u) βf (u) m◦ ϕf(u).
Usingthecommutativityoftheabovediagram,andsettingT = TU,λ(β)f◦L(h,f,f),
we get T = TU,λ(β)f u∈[G\U] l gh,u, f σh(u) , fuλ−1gh,u, h, σh(u), u = u∈[G\U] βf(u)l gh,u, f σh(u)
, f(u)λ−1gh,u, h, σh(u), u
= u∈[G\U] m gh,u, ϕ fσh(u) , ϕfuλ−1gh,u, h, σh(u), u βfσh(u) = M(h, ϕ◦ f, ϕ ◦ f)βf.
SoTU,λ(β): L→ M◦ TU,λ(ϕ) isanaturaltransformation.Thus,
TU,λ(ϕ), TU,λ(β) :tU,λ(X, l), L →tU,λ(Y, m), M is amapofC-monomialG-posets. 2
Lemma25. LetG,H andK be finite groups.If U isa(G,H)-biset andV is aleft free
(H,K)-biset, then themap (u,v)∈ U × V → (u,Hv)∈ U ×HV restricts to abijection
π : [G\U]× [H\V ] → [G\(U ×H V )], where brackets denote sets of representatives of
orbits.
Proof. For(u,v)∈ U × V ,there existsv0∈ [G\V ] and h∈ H suchthatv = hv0. Then
thereexistsu0∈ [G\U] andg∈ G suchthatuh= gu0.Then(u,Hv)= g(u0,Hv0).Hence
π issurjective.Now if(u0,v0) and (u1,v1) are pairsin[G\U]× [H\V ] which lieinthe
sameG-orbit,there exists g∈ G and h∈ H such that(gu0,v0)= (u1h−1,hv1).Hence
hv1 = v0, so v0 = v1 = hv1, and h = 1 since H act freely on V . Then gu0 = u1, so
u0= u1, andπ isinjective. 2
Proposition26. LetG,H andK befinite groups.
1. Let (•,1) be the C-monomial G-posetwhere • is G-posetwith one element and 1:
•→ •C isthefunctor suchthat1(g,•,•)= 1.ThenTU,λ(•,1)= (•,1).
2. Let(∅,z) betheempty C-monomial(G,H)-poset.ThenT∅,z istheconstant functor with value (•,1).
3. Let(U,λ) and(U,λ) be C-monomial(G,H)-bisetsandlet(X,l) beaC-monomial G-poset then
TUU,λλ(X, l) = TU,λ(X, l)TU,λ(X, l).
4. Let idG stand for the identity (G,G)-biset. Then TidG,1(X, l) = (X,l) for any
C-monomialG-poset (X,l).
5. Let (V,ρ) be a C-monomial left free (H,K)-biset, and (U,λ) be a C-monomial
(H,G)-biset.Then
TV,ρ◦ TU,λ= TU×HV,λ×ρ.
Proof. 1.,2.,3.and 4.areclear.
5. Note that since V is left free, we have Uλ ◦H Vρ ∼= GU×HVK. Let (X,l) be a
C-monomialG-poset.Weneedto showthat tV,ρ tU,λ(X, l), LU,λ , LV,ρ◦ LU,λ =tU×HV,λ×ρ(X, l), LU×HV,λ×ρ .
WedefineaK-posetmapϕ: tV,ρ
tU,λ(X,l),LU,λ
→ tU×HV,λ×ρ(X,l) suchthat
ϕ(f )(u,Hv) = f (v)(u)
foranyf ∈ tV,ρ
tU,λ(X,l),LU,λ
and(u,Hv)∈ U ×HV .It’s clearthatthemapϕ(f ) is
Letg∈ G(u,Hv).NotethatsinceV isH-free,wehaveg∈ Gu.Then l g, ϕf(u,Hv), ϕ f(u,Hv) = l g, f (v)(u), f (v)(u) = λ(g, 1, u, u)ρ(1, 1, v, v) = (λ× ρ)g, 1, (u,Hv), (u,Hv)
.
and soϕ(f )∈ tU×HV,λ×ρ(X,l).
Now wedefineamap
θ : tU×HV,λ×ρ(X, l)→ tV,ρ
tU,λ(X, l), LU,λ
suchthatθ(t)(v)(u)= t(u,Hv) foranyt∈ tU×HV,λ×ρ(X,l),u∈ U and v∈ V .Weshow
that θ(t)∈ tV,ρ
tU,λ(X,l),LU,λ
. Indeed, the map θ(t) isclearly a map of H-sets and moreover,sinceV isH-free,wehaveHv= 1 for anyv∈ V .Then
LU,λ
1, θ(t)(v), θ(t)(v)= 1 = ρ(1, 1, v, v).
Clearly,θ(t)(v) is amapof G-sets.Letg∈ Gu.Then g∈ G(u,Hv),and weget
l
g, θt(v)(u), θt(v)(u)= lg, t(u,Hv), t(u,Hv)
= λ(g, 1, u, u)ρ(1, 1, v, v) = λ(g, 1, u, u). Soθ(t)∈ tV,ρ tU,λ(X,l),LU,λ .
NowweshowthatLV,ρ◦LU,λ=LU×HV,λ×ρ.Letk∈ K andf,f ∈ tV,ρ
tU,λ(X,l),LU,λ
such thatkf ≤ f. Letv∈ [H\V ].Then there existauniqueσk(v)∈ [H\V ] and some
hk,v ∈ H suchthat
vk = hk,vσk(v).
Let u∈ [G\U]. Thenthere exist auniqueσhk,v(u)∈ [G\U] and someghk,v,u∈ G such
that
uhk,v= ghk,v,uσhk,v(u).
Then
(u,Hv) = (uhk,vhk,v−1,Hv) = (uhk,v,Hhk,v−1v)
=ghk,v,uσhk,v(u),Hσk(v)k−1 = ghk,v,u σhk,v(u),Hσk(v) k−1. Weget (u,Hv)k = ghk,v,u σhk,v(u),Hσk(v) .