ON EVALUATION FIBER SEQUENCES
MITSUNOBUTSUTAYA
ABSTRACT. Wedenote the n-th projectivespaceofatopological monoid$G$by$B_{n}G$andthe classifying
spaceby$BG$
.
Ourmain resultstatesthat,if$G$isatopologicalgroup,then theevaluation fibersequence$Map_{0}(B_{n}G, BG)arrow Map(B_{n}G, BG)arrow BG$extends totheright. This theorem isprovedbythe technique of$A_{n}$-maps.
1. INTRODUCTION
Theaimof this noteisto presentthe result oftheauthor’spaper[Tsu].
Firstofall, let
us
review thehistory of$A_{n}$-theory. $H$-space was
introducedbyJ.-P. Serre,which isnamed after H. Hopf. Apointed
space
$G$is
said tobean
$H$-space
ifa
continuous
binaryoperation
$m$
:
$G\cross Garrow G$of which the identity elementis
thebasepointis given. Every topological monoidis
of
courese an
$H$-space.
But theconverse
does nothold.J. F.Adams[Ada60]provedthat
an
$n$-dimensionalsphere$S^{n}$admitsa
structureofan
$H$-space
ifandonlyif$n=0$,1,
3
or
7.
But I. M.James [Jam57] provedthat$S^{7}$never
admita
homotopyassociative$H$-structure. An$H$-space $(G, m)$ is said to be homotopy associative ifthe maps $m\circ(id_{G}\cross m)$ and
$m\circ(m\cross id_{G})$
are
homotopic. So,there issome
difference between$H$-spaces
and topologicalmonoids.Then,howabout thedifference between homotopy
associative
$H$-spaces
andtopologicalmonoids?J. D. Stasheff $[Sta63a]$ introduced the notion of$A_{n}$-spaces. $H$
-space
and homotopyassociative
$H$-space
are
nothing but$A_{2}$-spaces
and$A_{3}$-spaces,
respectively. Every topologicalmonoid isan
$A_{\infty}-$space.
For general $n,$ $A_{n}$-space
isan
$H$-space
withsome
higher homotopyassociativityof
order $n$in
some
sense.
This homotopy associativity datais called$A_{n}$-form.
His work revealed thatbeinga
topologicalmonoidismuchfarfrombeing
an
$H$-space.
The
same
thingcan
besaid formaps
between topological monoids. Amap
$f$:
$Garrow G’$is
saidtobean
$H$-map
ifthemaps
$(g_{1},g_{2})\mapsto f(g_{1}g_{2})$ and$(g_{1},g_{2})\mapsto f(g_{1})f(g_{2})$are
homotopic. Everycontinuous
homomorphism between topological monoids is
an
$H$-map
but theconverse
is false.M. Sugawara [Sug60] given
a
condition for thata
map
between topological monoid inducesa
map
between their classifyingspaces.
As is well-known,a
homomorphism$Garrow G’$ inducesa
map
$Bf$
:
$BGarrow BG’$ between theclassifying spaces. He introduced the notionstrongly multiplicativemap and proved that
a
strongly multiplicativemap
inducesa map
between the classifyingspaces.
After that, Stasheff $[Sta63b]$ proved theconverse
for path-connectedtopological monoids. Beinga
strongly multiplicative
map
requiresa
map
topreserve
the infinitely higher homotopy associativityoftopological monoids. Stasheffweakenedthe condition and defined$A_{n}$-map between topological
monoids. $A_{n}$
-map
isalso definedby the dataofpreservationofhigher homotopyassociativity, called$A_{n}$
-form.
By definition of$A_{n}$
-map,
one may
expect that$A_{n}$-map is generalized for morphism between$A_{n^{-}}$maps.
Stasheff [Sta70] described the condition only for $n\leq 4$ because it needs combinatoriallycomplicated cellcomplex,calledmultiplihedra. Abstractly, this
was
done byBoardman-Vogt[BV73]Keywordsandphrases. mappingspace,homotopy fibersequence,$A_{n}$-space,$A_{n}$-map, gauge group.
数理解析研究所講究録
MITSUNOBU TSUTAYA
using their”’$W$-constmction”’. Independently, N. Iwase also constructed those complexes in hismaster
thesis [Iwa83]. Hedescribedtheircombinatorialstructure
as
well. 2. SPACE0F$A_{n}$-MApsNow, to state
our
result,we
get backtothe workofStasheffabout$A_{n}$-maps
between topologicalmonoids. Consider
a
path-connectedtopological monoid$G$.
Letus
denote the n-thprojectivespace
and theclassifying
space
by$B_{n}G$and$BG=B_{\infty}G$,respectively. Weremarkthat thereisthecanonicalinclusion$B_{n_{1}}G\subset B_{n_{2}}G$for$n_{1}<n_{2}$and the homeomorphism$B_{1}G\cong\Sigma G$with thereduced suspension.
There is
a
universal principalG-fibration $EGarrow BG$ with $EG$ contractible. Therestriction$E_{n}Garrow$$B_{n}G$
over
$B_{n}G$isalsoa
principalG-fibration. Using thecanonicalhomotopycofibersequence
$E_{n}Garrow B_{n}Garrow B_{n+1}G$
and the homotopy fiber
sequence
$E_{n}Garrow B_{n}Garrow BG,$
Stasheff’sresult[$Sta63b$,Theorem4.5]isrephrased
as
follows.Theorem
2.1
(Stasheff, 1963). Let $G$ and $G’$ be connected topological monoids which are $CW$complex. Then, a pointedmap $f$
:
$Garrow G’$ isan
$A_{n}$-mapif
and onlyif
the reduced suspension$\Sigma f$
:
$\Sigma Garrow\Sigma G’\subset BG’$can
beextendedtoa
map$B_{n}Garrow BG’.$Theorem
2.1
statesnothing about the correspondence of$A_{n}$-forms. Our result refines this point.Let
us
denote the space ofcontinuous mapsbetween $X$and $Y$ byMap$(X, Y)$ and that of pointedones
by$Map_{0}(X, Y)$.
To guarantee the exponentiallaw,we
always workinthe category of compactlygeneratedspacesandthemappingspaces
are
considered in themanner
ofcompactly generatedspaces. Thespaceof$A_{n}$-mapswith$A_{n}$-forms between$G$and$G$’ is denoted by$ffl_{n}(G, G’)$.
Theorem
2.2
(T). Let$G$bea
well-pointedtopologicalmonoidof
homotopytypeof
a$CW$complex and $G’$a well-pointedgrouplike topologicalmonoid. Thenthefollowingcomposite isa
weakequivalence.$\ovalbox{\tt\small REJECT}_{n}(G,G’)arrow Map_{0}(B_{n}G, B_{n}G’)B_{n}arrow Map_{0}(B_{n}G, BG(\iota_{n})_{\#}$
A pointed space $X$ is saidto be well-pointed ifthe basepoint of$X$ has the homotopy extension
property. A topological monoid $G$ is said to be grouplike if$\pi_{0}(G)$ is a group with respect to the
multiplicationinducedfrom that of$G$
.
Themap
$B_{n}$ is givenby Sugawara’sconstruction [Sug60] andthe
map
$(\iota_{n})_{\#}$ is the composition withtheinclusion $\iota_{n}$:
$B_{n}Garrow BG$.
In [Tsu], the author constructsa
topological category$\ovalbox{\tt\small REJECT} l_{n}$ oftopological monoids and$A_{n}$-maps
between them and realizes $B_{n}$as a
continuous functor from$ffl_{n}$ tothe category of(compactly generated)pointedspaces.
3.
EVALUATIONFIBERSEQUENCENext,
we
explain the main resulton
evaluation fibersequences.
If $X$ is well-pointed, then theevaluation
Map$(X, Y)arrow Y$
at the basepoint is
a
Hurewicz fibration of which the fiberover
the basepoint is $Map_{0}(X, Y)$.
Thisfiber
sequence
is called the evaluationfiber
sequence. Roughly, our second result states that this fibersequence
extends to the right if$X=B_{n}G$ and $Y=BG$.
Onemay
think that this is strange.Because itimplies not only that the mapping
space
$Map_{0}(B_{n}G, BG)$ is equivalenttoa
topologicalmonoid,butalsotheconnecting
map
$Garrow Map_{0}(B_{n}G, BG)$isequivalenttoa
homomorphism betweenON EVALUATIONFIBER SEQUENCES
topological
monoids. But Theorem2.2
claims that $Map_{0}(B_{n}G, BG)$is
equivalent to the topologicalmonoid$fl_{n}(G,G)$
.
There is
a
well-known adjunction$Map_{0}(\Sigma X, Y)\cong Map_{0}(X, \Omega Y)$
for pointed
spaces
$X$and $Y$, where$\Omega Y$is thespace
of based loopsin $Y.$For
a
topological monoid $G$, define the subspace $\overline{Map}_{0}(B_{n}G, BG)\subset Map_{0}(B_{n}G, BG)$ consistingof
maps
$B_{n}Garrow BG$ which restricts toa map
$\Sigma Garrow BG$ with the adjoint $Garrow\Omega BG\simeq G$ isa
homotopy equivalence. We also define the subspace $\overline{Map}(B_{n}G, BG)\subset Map(B_{n}G, BG)$
as
the unionof the path-components thatintersectwith$\overline{Map}_{0}(B_{n}G, BG)$
.
For
a
topologicalgroup
$G$,the composition ofthe conjugation$Garrow G$ byan
element of$G$ toan
$A_{n}$
-map
from the left defines the leftaction of$G$on
$\ovalbox{\tt\small REJECT}_{n}(G, G)$.
In particular, thecomposition to theidentity
map
definesa
map
$\delta$:
$Garrow\ovalbox{\tt\small REJECT}_{n}(G, G)$
.
Note that there
is
a
natural well-pointed replacement$tWGarrow G$ofa
topological monoid$G$,whichis
a
homomorphism
between topological monoids andis
a
weak equivalence.Theorem
3.1
(T). Let $G$ bea
well-pointed topological groupof
homotopy typeof
a
$CW$complex.Consider
a
map$BGarrow B’W\ovalbox{\tt\small REJECT}_{n}(G, G; eq)$defined
bythecomposite$BGarrow\simeq KWGarrow KW\ovalbox{\tt\small REJECT}_{n}(G, G;KW\delta eq)$
.
Thenthe
sequence
$\overline{Map}_{0}(B_{n}G, BG)arrow$Map$(B_{n}G, BG)arrow BGarrow KW\ovalbox{\tt\small REJECT}_{n}(G, G; eq)$,
is
a
homotopyfiber
sequence.When$n=\infty$, the above
sequence
can
be extendedas
$\overline{Map}_{0}(BG, BG)arrow\overline{Map}(BG, BG)arrow eBGarrow B^{t}W\overline{Map}_{0}(BG, BG)arrow B^{t}W\overline{Map}(BG, BG)$,
where the subspaces$\overline{Map}_{0}(BG, BG)\subset Map_{0}(BG, BG)$ and Map$(BG, BG)\subset Map(BG, BG)$
are
ex-actlythoseof homotopy equivalences$BGarrow BG$
.
Fordetails,see
Gottlieb’spaper
[Got73].4.
$A_{n}-$lYPESOFGAUGE GROUPSTheorem
2.2
and3.1
may
beapplied inmany
situations. Asan
application ofthem,we
obtain the resultfor$A_{n}$-types ofgauge groups.
Fora
principal $G$-bundle $Parrow B$, the topologicalgroup
$\mathcal{G}(P)$consistingof$G$-equivariant
maps
$Parrow P$that induces theidentityon
$B$is called thegauge
group of$P.$Theassociatedbundle$adP=P\cross_{G}G$withrespect totheadjointactionof$G$iscalled the adjoint bundle
andbecomes
a
bundle oftopologicalgroups.
Thespace
ofsections$\Gamma(adP)$isnaturallyisomorphictothe
gauge group
$\mathcal{G}(P)$.
For example,theclassification of the$A_{n}$-typesof thegaugegroups
of principal$SU(2)$-bundles
over
$S^{4}$ is investigated by Kono [Kon91], byCrabb-Sutherland[CS00], by Tsukuda[TsuOl] and bytheauthor[Tsu12]. Forhomotopy types, there
are
many
related works.Theorem
4.1
(Kishimoto-Kono,2010
andT). Let$G$ bea
well-pointed topologicalgroup and$B$ beapointedspace, both
of
which have the pointed homotopy typesof
$CW$complexes. Fora
principal$G$-bundle$P$
over
$B$classified
by$\epsilon$:
$Barrow BG$, the followingconditionsare
equivalent:(i) ad$P$is$A_{n}$
-trivial
(ii) the map$(\epsilon, \iota_{n})$
:
$B\vee B_{n}Garrow BG$extendsovertheproduct$B\cross B_{n}G,$(iii) the composite$Barrow BG\epsilonarrow\theta W\ovalbox{\tt\small REJECT}_{n}(G, G; eq )$is null-homotopic.
MITSUNOBU TSUTAYA
A bundle of topologicalmonoids$Earrow B$is saidtobe$A_{n}$-trivialifthere exist
a
topologicalmonoid$G$anda“fiberwise$A_{n}$-equivalence”’$B\cross Garrow adP$
.
The$A_{n}$-trivialityof ad$P$impliesthe$A_{n}$-equivalenceof$\mathcal{G}(P)$and Map$(B, G)$
.
Theequivalence of(i)and(ii)have already been known byKishimoto-Kono[KK10].
In[Tsu], the authorappliedTheorem
2.2
and3.1
tohigher homotopycommutativityand tocyclicmaps
as
well.5. FUTUREW0RK
It
seems
thatTheorem3.1
hasseveralapplicationsto homotopytheory ofgauge groups
andhigher homotopycommutativity. Another directionis
generalizations ofTheorem2.2.
The authoris
trying togeneralize Theorem2.2for the$A_{\infty}$-functors betweensmalltopological categories.REFERENCES
[Ada60] J. F.Adams,Onthenon-existenceofelementsofHopfinvariantone,Ann.ofMath.72(1960),20-104.
[BV73] J.M. Boardman and R. M. Vogt, Homotopyinvariantstructuresontopologicalspaces,LectureNotesin
Mathe-matics347,Springer-Verlag,Berlin, 1973.
[CS00] M. C. Crabb andW. A. Sutherland, Counting homotopytypes ofgauge groups, Proc. LondonMath. Soc. 81
(2000),747-768.
[Got73] D. H.Gottlieb,The totalspaceofuniversalfibrations,Pacific J.Math.46(1973),415-417.
[Iwa83] N.Iwase,Onthe ring structureof$K^{*}(XP^{n})$(Japanese),MasterThesis,KyushuUniv., 1983.
[Jam57] I. M.James,Multiplicationonspheres.II,Trans. Amer.Math.Soc.84(1957),545-558.
[KK10] D.Kishimotoand A.Kono,Splittingofgaugegroups,Trans. Amer. Math. Soc.362(2010),6715-6731.
[Kon91] A.Kono,A noteonthe homotopytypeofcertain gauge groups,Proc.Roy.Soc. Edinburgh: Sect. A117(1991),
295-297.
[Sta63a] J.D. Stasheff,Homotopy associativity
of
$H$-spaces.I,Trans.Amer. Math. Soc.108(1963),275-292.[Sta63b] J.D.Stasheff,Homotopyassociativityof$H$-spaces.II,Trans. Amer. Math. Soc.108(1963),293-312.
[Sta70] J. D. Stasheff,$H$-spacesfromahomotopy point ofview,Lecture Notes in Mathematics 161,1970.
[Sug60] M.Sugawara, On thehomotopy-commutativityofgroupsand loopspaces,Mem.CollegeSci. Univ.Kyoto Ser.
AMath.33(1960),257-269.
[TsuOl] S.Tsukuda,Comparing the homotopytypesofthecomponentsofMap$(S^{4}, BSU(2))$,J.Pure and Appl. Algebra
161(2001),235-247.
[Tsu12] M.Tsutaya, Finitenessof$A_{n}$-equivalencetypesofgauge groups,J. London Math. Soc.85(2012),142-164. [Tsu] M. Tsutaya, Mappingspacesfromprojectivespaces,arXiv:1408.2010,preprint.
DEPARTMENT0FMATHEMATICS,KYOTOUNIVERSITY, KYOTO606-8502,JAPAN
$E$-mail address: tsutaya@math.kyoto-u.ac.jp