• 検索結果がありません。

Homotopy spherical space formsÐa numerical bound for homotopy types

N/A
N/A
Protected

Academic year: 2024

シェア "Homotopy spherical space formsÐa numerical bound for homotopy types"

Copied!
10
0
0

読み込み中.... (全文を見る)

全文

(1)

Homotopy spherical space formsÐa numerical bound for homotopy types

Marek GolasinÂski and Daciberg Lima GoncËalves

(Received July 27, 1999) (Revised July 5, 2000)

Abstract. Let G be a ®nite group. We show that for a ®xed nV1 the set of homotopy typesof orbit spacesof all free G-actionson homotopy…2n 1†-spheres is

®nite and bounded by the order of some quotient group associated with G. In particular, we deduce that there are at most two homotopy types of lens spaces de- termined by all freeZ=pm-actionson homotopy 3-sphereswhenpisan odd prime, and only one homotopy type of those spaces provided that 4ap 1. There isalso only one homotopy type of lens spaces of dimension 2n 1 determined by allZ=2m-free actions provided that n isodd.

Introduction

A ®nite group hasthe periodic homology if it actsfreely on a ®nite homology sphere [1, Chap. III] and [3, 4]. On the other hand, Swan [10]

showed that any ®nite group with periodic homology of period lactsfreely on a ®nite CW complex of the homotopy type of an …nl 1†-sphere for some positive integer n. ThisCW complex may not have the homotopy type of a closed manifold. The symmetric group S3 hasa periodic homology but does not act freely on any manifold which hasthe homotopy type of a sphere due to the result of Milnor [8]: Every element of order two must be in the center.

The study of free actions of a group on homotopy spheres is related to the study of their orbit spaces. In the case of the cyclic group Z=2 of order two, the orbit space has the homotopy type of the real projective space and the problem of classifying manifolds of this homotopy type has been studied extensively in [7, 11]. An old result of homotopy type theory (see e.g. [9]) says that, up to homotopy, the set of lens spaces exhausts all the homotopy types of orbit spaces of free actions of the cyclic group Z=mof order m on homotopy …2n 1†-spheres. Thus the classi®cation of all the Z=m-actionson those homotopy spheres is equivalent to the classi®cation of manifolds of the homotopy type of lens spaces studied by Browder in [2]. The set of homotopy

2000 Mathematics Subject Classi®cation. Primary 55M35, 55P15; secondary 57S17.

Key words and phrases: automorphism group, CW complex, freeG-action, homotopy sphere, lens space, orbit space.

(2)

types of lens spaces is evidently ®nite and their homotopy classi®cation has been presented in [5, 9]. The main purpose of this paper is to present in Corollary 1.3 a proof of the following result to estimate the number of homotopy types of the orbit spaces.

Let G be a ®nite group with H2n…G;Z† ˆZ=jGjfor a ®xed nV1,wherejGj is the order of the group G. Then the set of homotopy types of orbit spaces of all free G-actions on all homotopy …2n 1†-spheres is ®nite and bounded by the order of the quotient group Aut…Z=jGj†=fGj;jAAutGg of the automorphism group Aut…Z=jGj†, where j is the induced automorphism on the cohomology H2n…G;Z† ˆZ=jGj by j in the automorphism group AutG.

In particular, under some assumptions on n and m, we calculate in Corollary 2.3 the number of homotopy typesof lensspacesgiven by freeZ=pm- actionson homotopy …2n 1†-spheres and we estimate in Corollary 2.5 the number of homotopy typesof orbit spacesgiven by free actionsof the generalized quaternion group Q2m‡2 on homotopy …4n 1†-spheres.

The paper is divided into two sections. In § 1 we use tom Dieck's theorem [6, p. 126] to prove our main result. Then in Proposition 1.5 we deduce that the group of homotopy classes of homotopy equivalences of the orbit space of a free G-action on a homotopy sphere is isomorphic to the subgroup of AutG which consists of all the automorphisms j with j1 G1 …modjGj† provided that jGj>2.

In § 2 we examine the quotient group Aut…Z=jGj†=fGj;jAAutGg for a cyclic group Gof ®nite order. In particular, there are at most two homotopy typesof lensspacesdetermined by all the free Z=pm-actionson homotopy 3- spheres when p is an odd prime, and only one homotopy type of those spaces provided that 4ap 1. There is also only one homotopy type of lens spaces of dimension 2n 1, determined by all the Z=2m-free actionsprovided thatn is odd. We prove that the number of homotopy typesof the orbit spacesof free actionsof the quaternion groupQ8 on homotopy 3-spheres is bounded by 2.

At the end, we deal with the group of homotopy typesof self-homotopy equivalencesof a lensspace given by a free Z=p-action on a homotopy …2n 1†-sphere, provided that p isa prime.

The authorsare extremely grateful to the referee for carefully reading the original manuscript and his many valuable suggestions. The ®rst author acknowledgesthe SaÄo Paulo University hospitality and the FAPESP-SaÄo Paulo (Brasil) ®nancial support during his visit when the work has been done.

1. Main result

An n-dimensional CW complex Xwith the homotopy type of an n-sphere Sn iscalled a homotopy n-sphere. An orientation for X consists of a choice of

(3)

a generator z…X† in the cohomology group Hn…X;Z† ˆZ. Suppose that a

®nite group G of order jGj actsfreely on X. Then by [4], the group G has periodic cohomology groupsof period n‡1 with Hn‡1…G;Z† ˆZ=jGj. Given two oriented homotopy n-spheres X1 and X2 with free G-actions, we can associate to each G-map f :X1!X2 a degree d…f†, de®ned by fz…X2† ˆ d…f†z…X1†, where f isthe induced map on cohomology groups. Throughout thissection G will be a ®nite group acting freely and cellularly on an odd dimensional homotopy sphere X. Then all the hypotheses of Theorem 4.11 in [6, p. 126] are satis®ed for homotopy spheres and it can be formulated as follows.

Theorem 1.1. Suppose that X1 and X2 are oriented homotopy n-spheres with free cellular actions of a ®nite group G. Then the following holds:

(1) The set ‰X1;X2ŠG of G-homotopy types of all G-maps X1!X2 is not empty.

(2) Let f :X1!X2 be a G-map. Suppose that d1d…f† …modjGj†. Then there exists a G-map h:X1!X2 such that d…h† ˆd.

(3) Suppose that f0;f1:X1!X2 are G-maps. Then d…f0†1d…f1† …modjGj†.

(4) Two G-maps f0;f1:X1!X2 are G-homotopic if and only if d…f0† ˆd…f1†.

From Theorem 1.1 it followsthat for a G-mapf :X1 !X2 of homotopy …2n 1†-spheres with free G-actionsthere isa G-map g:X2!X1 with d…gf†

ˆd…g†d…f†11 …modjGj†. Hence the degree d…f†determinesa unit in Z=jGj and then an automorphism of the cyclic group Z=jGj. Since H2n…G;Z† ˆ Z=jGj, we have the induced map of automorphism groups AutG! Aut…H2n…G;Z†† ˆAut…Z=jGj†. For an automorphism jof the groupG, letj be itsimage in Aut…Z=jGj†. Observe that j can be identi®ed with a unit of the ring Z=jGj so in the sequel we may also identify j with some positive integer less than the orderjGjof the groupG, which isinvertible in this ring.

Write Xg for a homotopy …2n 1†-sphere X with a free cellular G- action g:GX !X and let X=g be the associated orbit space. By the Serre spectral sequence argument Hk…X=g;Z† ˆHk…G;Z† for 0Uk<2n 1, H2n 1…X=g;Z† ˆZ and Hk…X=g;Z† ˆ0 for k>2n 1. Consequently, with any map between G-orbit spaces we can also associate its degree. Fix an arbitrary free cellular G-action g0 on a homotopy …2n 1†-sphere X0. Then by Theorem 1.1, there isa G-map fg:Xg!Xg00.

Theorem 1.2. Two orbit spaces X1=g1 and X2=g2 of homotopy …2n 1†- spheres X1 and X2 have the same homotopy type if and only if d…fg1†1 Gjd…fg2† …modjGj†for somej in the group AutG and G-maps fg1:Xg11!Xg00 and fg2 :Xg22 !Xg00.

(4)

Proof. Suppose that for two orbit spaces X1=g1 and X2=g2 there is an automorphism j of the group G such that d…fg1†1 Gjd…fg2† …modjGj†.

Then, in view of Theorem 1.1, there isa G-map fg02 :Xg00!Xg22 such that d…fg02fg1† ˆd…fg02†d…fg1†1 Gj …modjGj†.

Let …X1=g1†2n 1 be the …2n 1†-stage of the Postnikov tower of X1=g1 and consider the canonical maps …X1=g1†2n 1 !K…G;1† to the Eilenberg-MacLane space K…G;1† and X1=g1! …X1=g1†2n 1. The map K…j 1;1†:K…G;1† ! K…G;1† determined by the automorphism j 1 yieldsa map j:…X1=g1†2n 1! …X1=g1†2n 1 with a homotopy commutative square

…X1=g1†2n 1 ƒƒƒƒ!j …X1=g1†2n 1

??

?y

??

?y

K…G;1† ƒƒƒƒ!K…j 1;1† K…G;1†:

Since the homotopy ®ber of the map X1=g1! …X1=g1†2n 1 is …2n 2†- connected, by obstruction theory the mapj:…X1=g1†2n 1! …X1=g1†2n 1 admits a map F:X1=g1!X1=g1 such that the diagram

X1=g1 ƒƒƒ!F X1=g1

??

?y

??

?y

…X1=g1†2n 1 ƒƒƒ!j …X1=g1†2n 1

commutesup to homotopy. Ifx0 isa base point in X1=g1 andF~:X1!X1 is a cover of the mapF, thenF…gx† ˆ~ j 1…g†F…x†~ for all g in the groupG andx in the space X1. Thismeansthat F~:Xg10

1!Xg11 isa G-map with d…F†~ 1 …j† 1 …modjGj†, where g10 :GX1!X1 isthe G-action such that g10…g;x† ˆ j…g†x for all gAG andxAX1. For the composite G-map fg02fg1F~:Xg10

1!Xg22, we get that d…fg02fg1F†~ 1 G1 …modjGj†. Then Theorem 1.1 yieldsa G-map h:Xg10

1 !Xg22 with d…h† ˆG1 which in view of [9] inducesa homotopy equivalence X1=g1!X2=g2 of the orbit spaces, since the spaces X1=g1 and X1=g10 are homeomorphic.

Suppose now that f :X1=g1!X2=g2 isa homotopy equivalence and ®x a base point x0 in the orbit space X1=g1. Then we get an automorphism jˆ p1…f† of the group G, determined by the fundamental group functor p1. Let f~:X1!X2 be a cover of the map f andF~:X1 !X1 the map associated to the automorphism j 1 in the light of the ®rst part of this proof. Then f~…gx†

ˆj…g†f~…x† andF…gx† ˆ~ j 1…g†F…x†~ for allgAG andxAX1, so the composite

(5)

f~F~:Xg11!Xg22 isa G-map with d…f~F†~ 1 G…j 1† …modjGj†. Finally, in the light of Theorem 1.1 we get that d…fg1†1 G…j 1†d…fg2† …modjGj† and the

proof iscomplete. r

In particular, if two orbit spaces X1=g1 and X2=g2 of homotopy …2n 1†- spheres X1 andX2 are homeomorphic, then d…fg1†1 Gjd…fg2† …modjGj† for some jAAutG and G-maps fg1:Xg11!Xg00 and fg2:Xg22!Xg00.

Write KG2n 1 for the set of all homeomorphism classes ‰X=gŠ of orbit spaces X=gof all homotopy …2n 1†-spheres X with respect to all free cellular G-actions. Then the map

D2nG 1:KG2n 1!Aut…Z=jGj†=fGj;jAAutGg

given by DG2n 1…‰X=gІ ˆ ‰d…fg†Š, where ‰d…fg†Šis the class of the automorphism in Aut…Z=jGj† determined by the integer d…fg†, iswell-de®ned. Here we note that the equality d…fg† ˆ 1 holdsif fg isthe identity map of a homotopy …2n 1†-sphere to itself with the opposite orientation. Let KG2n 1=F be the quotient set of KG2n 1 by the homotopy relation F. Then we can state

Corollary 1.3. The mapDG2n 1:KG2n 1!Aut…Z=jGj†=fGj;jAAutGg induces an injection DgG2n 1:KG2n 1=F!Aut…Z=jGj†=fGj; jAAutGg. Conse- quently, the setKG2n 1=Fis ®nite and bounded by the order of the quotient group Aut…Z=jGj†=fGj;jAAutGg.

If G isthe cyclic group Z=m of order m, then by [5, 9] the map Dg2nG 1 is surjective on the homotopy types of the lens spaces. Thus we deduce from Theorem 1.2 one of the classical results of homotopy theory.

Corollary 1.4. Given a free cellular action gof the cyclic group Z=m on a homotopy …2n 1†-sphere X, the orbit space X=g is homotopy equivalent to a lens space.

Let E‰X1=g1;X2=g2Š denote the set of homotopy classes of homotopy equivalences f :X1=g1!X2=g2. Then from the proof of Theorem 1.2, we may deduce the following result which is a generalization of the result in [5, p. 96] stated for cyclic groups only.

Proposition 1.5. If jGj>2 and orbit spaces X1=g1, X2=g2 of homotopy spheres are homotopy equivalent, then there is a bijection from the set E‰X1=g1;X2=g2Š to the subgroup of the group AutG consisting of all auto- morphisms j such that j1 G1 …modjGj†. In particular, the group E‰X=g;

X=gŠis isomorphic to that subgroup of the group AutG, for the orbit space X=g of a homotopy sphere X.

Proof. Let f :X1=g1!X2=g2 be a homotopy equivalence, ®x a base

(6)

point inX1=g1 and letjˆp1…f†be the automorphism of the group Ginduced by the fundamental group functor p1. Then d…f† ˆG1 and by the Serre spectral sequence argument j1d…f† ˆG1 …modjGj†. De®ne l…‰fІ ˆ p1…f† for ‰fŠAE‰X1=g1;X2=g2Š. We show that l givesthe required bijection from the set E‰X1=g1;X2=g2Š.

Consider two homotopy equivalences f;g:X1=g1!X2=g2 with p1…f† ˆ p1…g† ˆj. If f~;g~:X1!X2 are coversof f and g, respectively then f~F;~ g~F~:Xg11!Xg22 are G-maps, where F~:X1!X1 is a map associated with the automorphism j 1 asin the proof of Theorem 1.2. In view of Theorem 1.1, d…f†…j† 11d…f~F†~ 1d…~gF†~ 1d…g†…j† 1 …modjGj†. Consequently, d…f†1d…g† …modjGj†, hence d…f† ˆd…g† since d…f†;d…g† ˆG1 and jGj>2.

Finally, in the light of [9], the maps f;g:X1=g1!X2=g2 are homotopic.

Let now c be an automorphism of the group G such that c1 G1 …modjGj† and f :X1=g1!X2=g2 be a ®xed homotopy equivalence with jˆ p1…f†. Then, by the proof of Theorem 1.2, there isa map W:X1=g1! X1=g1 associated with the automorphism oˆj 1c and d…W† ˆG1, so it is a homotopy equivalence by [9]. But p1…fW† ˆc and the proof is

complete. r

Remark 1.6. It iswell-known that the group E‰X=g;X=gŠ isisomorphic to the cyclic group Z=2 provided that jGjU2 for any homotopy sphere X.

In the next section we calculate the group E‰X=g;X=gŠ, when Gisa cyclic group Z=p with p an odd prime and X a homotopy …2n 1†-sphere.

2. Estimation of the number of orbit spaces

Let g2 be a generator of the cohomology groupH2…Z=m;Z† ˆZ=m. The results of [4, Chap. XII, § 11] show that …g2†n generates H2n…Z=m;Z† ˆZ=m.

Then, for any jAAut…Z=m†, the induced automorphism j of the group H2n…Z=m;Z† ˆZ=m isdetermined by the power kn, for some k with …k;m† ˆ 1, that is kA…Z=m†.

Let now Q4mˆ fx;y;xmˆy2;xyxˆyg be the generalized quaternion group of order 4mand g4 a generator of H4…Q4m;Z† ˆZ=4m. Then by [10], the element …g4†n generates H4n…Q4m;Z† ˆZ=4m, and for any j in the group AutQ4m the induced automorphism j of the group H4n…Q4m;Z† ˆZ=4m is determined by the power k2n, for some k with …k;4m† ˆ1.

The automorphism group Aut…Z=m† isisomorphic to the unit group …Z=m† of the ringZ=m. Therefore, we are led to compute the quotient group …Z=m†=fGxn;xA…Z=m†g. If mˆp1k1 psks isthe prime factorization of an integer mV1 with kiV1 for iˆ1;. . .;s, then it iswell-known (see e.g. [12, Chapter IV]) that …Z=m†ˆ …Z=p1k1† …Z=psks†. Moreover, …Z=pl†ˆ

(7)

Z=…p 1† Z=pl 1 for p an odd prime or l<3 and …Z=2l†ˆZ=2Z=2l 2 forlV3 with generators, multiplication by 1 giving the element of order two, and multiplication by 5 or 3 giving the element of order 2l 2.

In the group Z=…p 1† Z=pm 1, if p isan odd prime, …p 1†=2 isthe only element of order 2, which corresponds to 1 in the group …Z=pm† by the isomorphism above. Therefore, the equation 1ˆxn hasa solution in the multiplicative group…Z=pm† if and only if …p 1†=2 isdivisible by the integer n in the additive group Z=…p 1†. Note that …2k‡1†…p 1†=21…p 1†=2 …mod…p 1††. On the other hand 10xn for any element x in the group …Z=2m† if n is an even integer. Thus, we will get the proposition below and we may concentrate in the sequel on the quotient group …Z=pm†=fGxn; xA…Z=pm†g, for positive integers m;n and a prime p.

Proposition 2.1. If mˆp1k1 psks is the prime factorization of an integer mV1, then for any integer nV1 there is an extension

0! …Z=2†t! …Z=m†=fGxn;xA…Z=m†g

! …Z=p1k1†=fGxn;xA…Z=p1k1†g …Z=psps†=fGxn;xA…Z=psks†g

!0;

where t is determined as follows:

(1) If 1Afxn;xAZ=pikig for any iˆ1;. . .;s, then tˆ0;

(2) Otherwise tˆs 1 afi; 1Afxn;xAZ=pikigg. In particular, when n is an odd integer, 1Afxn;xAZ=pikig for any iˆ1;. . .;s and hence tˆ0;

when n is an even integer, t‡1 is the number of odd primes pi which appears in the prime factorization of m and …pi 1†=2 is not divisible by n in the group Z=…pi 1†.

Proof. Of course, we have the extension

0! fGxn;xA…Z=p1k1†g fGxn;xA…Z=psks†g=fGxn;xA…Z=m†g

! …Z=m†=fGxn;xA…Z=m†g ! …Z=p1k1†=fGxn;xA…Z=p1k1†g …Z=psks†=fGxn;xA…Z=psks†g !0:

The square of every element in fGxn;xA…Z=p1k1†g fGxn;xA …Z=psks†g liesin fxn;xA…Z=m†g, because …Gxn†2ˆ …x2†n. So, it follows that the group

fGxn;xA…Z=p1k1†g fGxn;xA…Z=psks†g=fGxn;xA…Z=m†g is isomorphic to a direct sum of Z=2's. Certainly, tˆs 1 in the case (1).

(8)

In the case (2), 1Afxn;xA…Z=m†g; we can replace Gxn withxn for…Z=m†

and for …Z=piki† with 1Afxn;xA…Z=piki†g. Then it isnot hard to see the

result. r

Of course, we can assume that nˆq1t1 qltl and p 1ˆq1u1 qlul, where q1;. . .;ql are di¨erent primes, t1;. . .;tl;u1;. . .;ul are non-negative integersand ti or ui ispositive for all iˆ1;. . .;l.

Proposition 2.2. Let m;n be positive integers, p a prime and q1;. . .;ql

primes from the factorization above. Then the quotient group …Z=pm†= fxn;xA…Z=pm†g is isomorphic to one of the following groups:

(1) Z=qmin…t1 1;u1† Z=qmin…tl l;ul†, if p02 and p0q1;. . .;ql;

(2) Z=qmin…t1 1;u1† Z=qmin…ti0 1i0 1;ui0 1†Z=pmin…m 1;ti0†Z=qmin…ti0‡1i0‡1;ui0‡1† Z=qmin…tl l;ul†, if p02 and pˆqi0 for some 1Ui0Ul;

(3) the trivial group E, if pˆ2 and n is odd;

(4) Z=2Z=2min…ti0;m , if pˆ2 and qi0 ˆ2 for some 1Ui0Ul.

Proof. Let A be a ®nite additive abelian group and n a positive integer.

Then, for the endomorphism n:A!A given by the multiplication by n, the quotient group A=Imn isisomorphic to the kernel Kern, where Imn is the image of n. We consider the endomorphism of the multiplicative group …Z=pm†given by taking then-th power. But the group…Z=pm†isisomorphic to the additive group Z=…p 1† Z=pm 1, for p an odd prime or m<3 and to the group Z=2Z=2m 2 for pˆ2 and mV3. Thus the result follows.

Remark that ui0 ˆ0 in (ii). r

Consequently, the following holds.

Corollary 2.3. Let m;n and p be as in Proposition 2.2 and let Tp…m;n†

denote the number of homotopy types of lens spaces given by free Z=pm-actions on homotopy …2n 1†-spheres.

(1) If p is an odd prime and …p 1†=2 is divisible by n in the group Z=…p 1†, or pˆ2 and n is an odd integer, then Tp…m;n† is equal to:

( i ) qmin…t1 1;u1† qmin…tl l;ul†, if p02 and p0q1;. . .;ql;

( ii ) qmin…t1 1;u1† qmin…ti0 1i0 1;ui0 1†pmin…m 1;ti0†qmin…ti0‡1i0‡1;ui0‡1† qmin…tl l;ul†, if p02 and pˆqi0 for some 1Ui0Ul;

(iii) 1, if pˆ2.

(2) Otherwise Tp…m;n† is equal to:

( i ) 2 1qmin…t1 1;u1† qmin…tl l;ul†, if p02 and p0q1;. . .;ql;

( ii ) 2 1qmin…t1 1;u1† qmin…ti0 1i0 1;ui0 1†pmin…m 1;ti0†qmin…ti0‡1i0‡1;uio‡1† qmin…tl l;ul†, if p02 and pˆqi0 for some 1Ui0Ul;

(iii) 2min…ti0;m , if pˆ2, mV3 and qi0ˆ2;

(iv) 1, if pˆ2, mU2 and qi0ˆ2.

(9)

In particular, we get

Corollary 2.4. Let p be a prime and m a positive integer. Then the number of homotopy types of lens spaces given by free Z=pm-actions on a homotopy 3-sphere is equal to:

(1) 2, if either p 1 is divisible by 4 or pˆ2 and mV2;

(2) 1, if either p 1 is divisible by2 but not divisible by 4 or pˆ2 and mˆ1.

Moreover, for the generalized quaternion group Q2m‡2 one can deduce Corollary 2.5. The number of homotopy types of orbit spaces given by free Q2m‡2-actions on homotopy …4n 1†-spheres is bounded by 2 for any m>0, provided that n is an odd positive integer.

Proof. In the light of [10, Proposition 8.1], for any j in the group AutQ2m‡2, the induced automorphism j of H4n…Q2m‡2;Z† ˆZ=2m‡2 isde- termined by the power k2n for some k with …k;2m‡2† ˆ1. The result follows

from the same argument as for Corollary 2.3. r

Finally, let L be a lensspace determined by a Z=p-free action on a homotopy …2n 1†-sphere with p an odd prime. Then, by Proposition 1.5, the group E‰L;LŠ of homotopy typesof all itsself-homotopy equivalencesis isomorphic to the subgroup of…Z=p† consisting of all elementskwithkn1 G1 …modp†. But …Z=p†ˆZ=…p 1†, so the corresponding subgroup of Z=

…p 1† consisting of all elements l with nlˆ0 or nlˆ …p 1†=2. If d ˆ …n;p 1† is the greatest common divisor of the integers n and p 1, then the subgroup Gd of Z=…p 1† generated by …p 1†=d consists of all elements l in the group Z=…p 1† with nlˆ0. Moreover, if there isa solution l0 in the group Z=…p 1† of the equation nlˆ …p 1†=2, then the set of all solution of that equation in the group Z=…p 1† isequal to the coset l0‡Gd. We can conclude the paper with the following result deduced from the consideration above.

Proposition2.6. Let L be a lens space determined by a freeZ=p-action on a homotopy …2n 1†-sphere with an odd prime p. Then the group E‰L;LŠ of homotopy classes of all its self-homotopy equivalences is isomorphic either to the subgroup Gd of the cyclic group Z=…p 1† or to the subgroup GdU…l0‡Gd†, provided that there is a solution l0 of the equation nlˆ …p 1†=d in the group Z=…p 1†.

References

[ 1 ] G. E. Bredon, Introduction to Compact Transformation Groups, Academic Press, New York-London (1972).

(10)

[ 2 ] W. Browder, FreeZp-actions on homotopy spheres, Topology of Manifolds, Proceedings of the University of Georgia, Topology of Manifolds Institute (1969), 217±226.

[ 3 ] K. S. Brown, Cohomology of Groups, Springer-Verlag, New York-Heidelberg-Berlin (1982).

[ 4 ] H. Cartan and S. Eilenberg, Homological Algebra, Princeton, New Jersey (1956).

[ 5 ] M. M. Cohen, A Course in Simple-Homotopy Theory, Springer-Verlag, New York- Heidelberg-Berlin (1973).

[ 6 ] T. tom Dieck, Transformation Groups, Walter de Gruyter, Berlin-New York (1987).

[ 7 ] S. Lopez de Medrano, Some results on involutions of homotopy spheres, Proceedings of the Conference on Transformation Groups, New Orleans, Springer-Verlag, Berlin (1967), 167±174.

[ 8 ] J. Milnor, Groupswhich act on Sn without ®xed points, Amer. J. Math.79(1957), 623±

630.

[ 9 ] P. Olum, Mappingsof manifoldsand the notion of degree, Ann. of Math.56, no. 3 (1953), 458±480.

[10] R. G. Swan, Periodic resolutions for ®nite groups, Ann. of Math.72, no. 2 (1960), 267±

291.

[11] C. T. C. Wall, Surgery on Compact Manifolds, Academic Press, London-New York (1970).

[12] H. J. Zassenhaus, The Theory of Groups, Chelsea, New York (1958).

Marek GolasinÂski

Faculty of Mathematics and Informatics Nicholas Copernicus University Chopina 12/18, 87-100 TorunÂ, Poland

e-mail: [email protected] Daciberg Lima GoncËalves Department of Mathematics-IME

University of SaÄo Paulo

Caixa Postal 66.281-AG. Cidade de SaÄo Paulo 05315-970 SaÄo Paulo, Brasil

e-mail: [email protected]

参照

関連したドキュメント

3.2 Bounded-SVD bias In bounded-SVD bias, the evaluation values are estimated not only by the product of the feature matrices but also by taking into account the user and

In [BH] it was shown that the singular orbits of the cohomogeneity one actions on the Kervaire spheres have codimension 2 and 2n, and that they do not admit a metric with

In this paper, we show that the rational homotopy Lie alge- bra of classifying spaces for certain types of hyperbolic coformal 2-cones is not nilpotent.. It is an algebraic model of

Famouri, “The application of homotopy analysis method to solve nonlinear differential equation governing Jeffery-Hamel flow,” Communications in Nonlinear Science and

Note on the space of polynomials with roots of bounded multiplicity 山口耕平 (Kohhei Yamaguchi) 電気通信大学情報理工学研究科 (University

The classifying space of the monoid of self-homotopy equivalences of a space and the Kedra-McDuff $\mu$ -classes.. Katsuhiko

In the fourth column, we indicate that in most cases we are able to verify the stronger result of homotopy equivalence of $\Delta$ with the Quillen elementary complex

Especially, the chain type level of the homotopy fibre of a map f gives an upper estimate for the E-category in the sense of Kahl [26], which is an algebraic approximation of