(Z2)k-ACTIONS WITH FIXED POINT SET OF CONSTANT

Indian J. Pure Appl. Math., 41(4): 607-623, August 2010
c Indian National Science Academy
°
(Z2 )k -ACTIONS WITH FIXED POINT SET OF CONSTANT
CODIMENSION 2k + 81
Jingyan Li∗ and Yanying Wang∗∗
∗ Department
of Mathematics and Physics, Shijiazhuang Railway Institute,
Shijiazhuang, 050043, Peoples’ Republic of China
e-mail : [email protected]
∗∗ College of Mathematics and Information Science,
Hebei Normal University, Shijiazhuang, 050016, Peoples’ Republic of China
e-mail : [email protected]
(Received 19 September 2008; after final revision 26 March 2010;
accepted 19 May 2010)
k
2 +8
The ideal J∗,k
of cobordism classes in the unoriented cobordism ring
M O∗ containing a representative admitting a (Z2 )k -action with fixed
point set of constant codimension 2k + 8 is determined for k ≥ 4.
Key words : (Z2 )k -action, indecomposable cobordism class, fixed
point set, constant codimension, Dold manifold.
1. Introduction
Let φ : (Z2 )k × M n → M n be a smooth action of the group (Z2 )k =
{T1 , T2 , · · · , Tk | Ti2 = 1, Ti Tj = Tj Ti } on a closed n-dimensional manifold.
The fixed point set F of the action is a disjoint union of closed submanifolds
1
Project supported by the Academic Research Fund from Shijiazhuang Railway Institute and NSFC (10971050).
608
JINGYAN LI AND YANYING WANG
of M n . If each component of F is (n − r)-dimensional, then F has constant
r denote the set of n-dimensional cobordism classes
codimension r. Let Jn,k
containing a representative M n admitting a (Z2 )k -action with fixed point
P
r =
r
set of constant codimension r and J∗,k
n≥r Jn,k . From [1] we know that
r is a subgroup in the unoriented cobordism group M O and J r forms
Jn,k
n
∗,k
P
an ideal of the cobordism ring M O∗ = n≥0 M On . It is not difficult to see
r1 r2
r1 +r2
r ⊂ Jr
that J∗,k
.
∗,k+1 and J∗,k J∗,k ⊂ J∗,k
1 = (0) in [2]. Inspired in
In 1965, Conner and Floyd showed that J∗,1
r and
this fact, Stong introduced the question of determining the ideal J∗,1
2
3
4
computed J∗,1 in [3]. Soon after, Capobianco computed J∗,1 in [4] and J∗,1
5 , J 6 , J 7 and
in [5]. After this, Iwata, Wada, Wu and Kikuchi obtained J∗,1
∗,1 ∗,1
8 respectively in [6-9]. In 1992, Pergher introduced the same question of
J∗,1
r with k ≥ 2 in [10]. Also, he computed J 1 and J 2 for
determining J∗,k
∗,k
∗,k
k ≥ 2, and made a lot of characteristic numbers calculations to determine
3 defined in terms of certain restrictions on the fixed
certain sub-ideals of J∗,2
data of the actions. In 1994, Shaker got the most important advance in
r in [11]. The crucial point was the discovery
the direction of computing J∗,k
of a method for constructing models of (Z2 )k −actions on certain manifolds
so that one can control two involved properties: the indecomposability of
the manifold and the codimension of the fixed point set of the action, thus
providing a method to construct generators for the unoriented cobordism
r for
ring with desired properties. Using this method, Shaker computed J∗,k
r < 2k , and concluded that


r = 1,
 (0),
r
∞
J∗,k =
⊕n=r M On ,
r even,

 ⊕∞ M O ∩ Kerχ, r odd, r ≥ 3,
n
n=r
where χ : M O∗ → Z2 denotes the mod 2 Euler characteristic. Later, in
[12], Shaker solved the case r = 2k . The key point was a way to equip Dold
manifolds with certain (Z2 )k −actions and the fact that there is a method
to recognize the indecomposability of Dold manifolds. These two articles
of Shaker gave a clear scheme to attack the next cases. Using this scheme
and the complicated formula of Kosniowski and Stong given in [13], Wang,
Wu and Ma solved the case r = 2k + 1 in [14], leaving open some particular
cases. In [15], Wang, Wu and Ding gave a crucial contribution concerning
the determination of the next cases and made an important advance in the
r for r > 2k . The authors showed that, for a given
setting of computing J∗,k
(Z2 )k -ACTIONS WITH FIXED POINT SET
609
r > 2k , there exists a number g(r) much bigger than r so that, if n ≥ g(r),
r
then Jn,k
is explicitly computed. For r > 2k , this completely solves the
r
question of computing Jn,k
for all sufficiently large n. In other words, for
k
r to a finite and explicit list of
r > 2 , this reduces the computation of Jn,k
values of n. The authors also solved the case r = 2k + 3 for k ≥ 4 and
n ≥ 2k + 4. In [16], Liu and Wu studied the case r = 2k + 2. They used the
generalized Dold manifold given in [17] to construct certain indecomposable
manifolds equipped with (Z2 )k −actions. This allowed to obtain, besides the
case r = 2k + 2, the solution for certain cases left previously open. For
r have been determined in [18-21]
r = 2k + 4, 2k + 5, 2k + 6 and 2k + 7, J∗,k
respectively.
In this paper, we will construct some special generators of M O∗ to de2k +8
termine the ideal J∗,k
. The main result is
k
2 +8
Theorem — For k ≥ 4, the ideal J∗,k
of M O∗ consists of all classes
k
in dimensions greater than 2 + 8 and the decomposable classes in dimension
2k + 8 which contain only factors with dimension less than 2k .
Throughout this paper manifolds will be smooth, compact and without
boundary but not necessarily connected. S m denotes the m-dimensional
sphere. The coefficient group is Z2 (the integers mod 2) and ≡ denotes
¡ ¢
congruence mod 2. Binomial coefficients are m
n = m!/n!(m − n)!.
2. Preliminaries
It is well known that the unoriented cobordism ring M O∗ is a Z2 -polynomial
algebra with a single generator in each dimension n which is not of the form
2u − 1 (see [1]). If the cobordism class [M n ] of the smooth closed manifold
M n can be expressed as a sum of products of lower dimensional cobordism
classes, then [M n ] is called decomposable. Otherwise it is indecomposable.
The indecomposable classes can be chosen as generators of the Z2 -polynomial
algebra M O∗ . For our purpose, the indecomposable classes will come from
the following sources.
Lemma 2.1 — [3; Lemma 3.4] Let RP (n1 , n2 , · · · , nl ) be the projective
space bundle of λ1 ⊕ λ2 ⊕ · · · ⊕ λl over RP (n1 ) × RP (n2 ) × · · · × RP (nl ),
where λi is the pullback of the canonical line bundle over the i-th factor.
Then for l > 1, [RP (n1 , n2 , · · · , nl )] is indecomposable in M O∗ if and only
610
JINGYAN LI AND YANYING WANG
if
Ã
n+l−2
n1
!
Ã
+
n+l−2
n2
!
Ã
+ ··· +
n+l−2
nl
!
≡ 1 mod 2,
where n = n1 + n2 + · · · + nl .
The manifold RP (n1 , n2 , · · · , nl ) has dimension n + l − 1. If ni+1 =
ni+2 = · · · = nl = 0, then RP (n1 , n2 , · · · , nl ) will sometimes be written as
RP (n1 , n2 , · · · , ni ; l).
In [17], Brown constructed the generalized Dold manifold. For a space
X and a positive integer m, let P (m, X) be formed from S m × X × X by
identifying (u, x, y) with (−u, y, x). If X is an n-dimensional manifold, then
P (m, X) is an m + 2n-dimensional manifold. He also gave
Lemma 2.2 — [17; Proposition 4.1] [P (m, M n )] is indecomposable in
M O∗ if and only if [M n ] is indecomposable in M O∗ and the binomial coefficient
!
Ã
m+n−1
≡ 1 mod 2.
m−1
To calculate binomial coefficients mod 2, we recall Kummer’s result: If
m=
l
X
mi 2i and n =
i=0
Ã
with 0 ≤ mi , ni ≤ 1, then
m
n
l
X
ni 2i
i=0
!
≡ 1 mod 2 if and only if ni ≤ mi for every
i.
Lemma 2.3 — [11; Lemma 3.1] Let λi −→ Xi be line bundles, and let
(Z2 )ki (ki ≥ 0) act on λi as bundle maps with fixed point set Fi on Xi for all
l
P
1 ≤ i ≤ l with
2ki ≤ 2k . Then RP (n1 , n2 , · · · , nl ) over X1 × X2 × · · · × Xl
i=1
admits a (Z2 )k -action with fixed point set F1 × F2 × · · · × Fl × E with E
being a set of l points.
Lemma 2.4 — [13] If (M n , φ) is a (Z2 )k -action on an indecomposable nmanifold, then some component of the fixed point set of M is of dimension
at least [n/2k ].
(Z2 )k -ACTIONS WITH FIXED POINT SET
611
Lemma 2.5 — [13] If (M n , φ) is a (Z2 )k -action on a manifold with
s(λ1 ,··· ,λi ) [M ] 6= 0,
then some component of the fixed point set of M is of dimension at least
[λ1 /2k ] + · · · + [λi /2k ].
3. Existence of Indecomposables
k
2 +8
The main task of this section is to exhibit indecomposable classes in J∗,k
.
k
2 +8
Lemma 3.1 — There exist indecomposable classes xn ∈ J∗,k
for k ≥ 4,
k
u
n ≥ 2 + 9 odd and not of the form 2 − 1.
Proof : Since n is not of the form 2u − 1, n−1
is not of the form
2
n−1
2k +9−1
u
k−1
2 − 1. We have 2 ≥
= 2
+ 4. From [11; 5.1], there exists
2
2k−1 +4
an indecomposable class X ∈ J n−1 ,k (for 2k−1 + 4 < 2k ), so that X has
2
a representative M admitting a (Z2 )k -action with fixed point set F 0 , where
n−1
n − 2k − 9
dim(F 0 ) =
− (2k−1 + 4) =
. Let Ti0 (i = 1, 2, · · · , k) denote
2
2
the (Z2 )k -action on M . Then we can define a (Z2 )k -action on S 1 × M × M
as follows:
T1 (u, x, y) = (u, T10 (x), T10 (y)),
T2 (u, x, y) = (u, T20 (x), T20 (y)),
······ ,
Tk (u, x, y) = (u, Tk0 (x), Tk0 (y)).
These involutions commute with the map T (u, x, y) = (−u, y, x), so that
induce a (Z2 )k -action on P (1, M ) with fixed point set F = S 1 × F 0 × F 0 /T =
P (1, F 0 ), where dimF = n − 2k − 8. Taking xn = [P (1, M )], we have that
2k +8
xn ∈ J∗,k
. By Lemma 2.2, xn is indecomposable.
¤
k
2 +8
Lemma 3.2 — There exist indecomposable classes xn ∈ J∗,k
for k ≥ 4,
k
k+1
2 + 10 ≤ n ≤ 2
and n even.
Proof : We will consider the following cases.
612
JINGYAN LI AND YANYING WANG
Case 1 — n = 2k + 2r1 + 2r2 + · · · + 2rm + 2, k > r1 > r2 > · · · > rm ≥ 2
and m ≥ 1.
k−1
From [11; 5.1], there exists an indecomposable class X ∈ J 2n−2 ,k+4 (for
2
2k−1 + 4 < 2k ), so that X has a representative M admitting a (Z2 )k -action
k−1 + 4) = n−2k −10 . Take
with fixed point set F 0 , where dim(F 0 ) = n−2
2 − (2
2
xn = [P (2, M )], with the (Z2 )k -action given in the proof of Lemma 3.1.
The fixed point set of this action is F = S 2 × F 0 × F 0 /T = P (2, F 0 ) with
2k +8
. By Lemma 2.2, xn is indecomposable.
dimF = n − 2k − 8, then xn ∈ J∗,k
Case 2 — n = 2k + 2r1 + 2r2 + · · · + 2rm , k > r1 > r2 > · · · > rm ≥ 2 and
m ≥ 2.
(1) If m ≥ 3 or m = 2 and r1 ≥ 5, then 2k−1 − 2r1 −1 − · · · − 2rm−1 −1 + 8 <
2k−1 .
From [11; 5.1], there exists an indecomposable class
2k−1 −2r1 −1 −···−2rm−1 −1 +8
X ∈ J n−2rm
, so that X has a representative M admitting
2
,k−1
rm
a (Z2 )k−1 -action with fixed point set F 0 , where dim(F 0 ) = n−22 − (2k−1 −
2r1 −1 −· · ·−2rm−1 −1 +8) = 2r1 +2r2 +· · ·+2rm−1 −8. Let Ti0 (i = 1, 2, · · · , k−1)
denote the (Z2 )k−1 -action on M . Then we can define a (Z2 )k -action on
rm
S 2 × M × M as follows:
T1 (u, x, y) = (u, y, x),
T2 (u, x, y) = (u, T10 (x), T10 (y)),
······ ,
0
0
Tk (u, x, y) = (u, Tk−1
(x), Tk−1
(y)).
These involutions commute with the map T (u, x, y) = (−u, y, x), so that
rm
induce a (Z2 )k -action on P (2rm , M ) with fixed point set F = S 2 × ∆(F 0 ×
F 0 )/T , where dimF =dim(F 0 ) + 2rm = n − 2k − 8. Taking xn = [P (2rm , M )],
2k +8
we have xn ∈ J∗,k
. By Lemma 2.2, xn is indecomposable.
(2) If m = 2 and r1 = 4, we take xn = [RP (19, 2r2 − 1; 2k − 1)]. From
2k +8
[15; Lemma 2.4], xn ∈ J∗,k
.
Ã
! Ã
!
k
4
r
2 +2 +2 2 −1
2k + 24 + 2r2 − 1
+
+ 2k − 3 ≡ 1 mod 2.
24 + 2 + 1
2r2 − 1
By Lemma 2.1, xn is indecomposable.
(Z2 )k -ACTIONS WITH FIXED POINT SET
613
(3) If m = 2 and r1 = 3, we take xn = [RP (11, 11; 2k − 9)]. Let
T1 [x1 , x2 , · · · , x12 ] = [x1 , x2 , · · · , x6 , −x7 , −x8 , · · · , −x12 ],
T2 [x1 , x2 , · · · , x12 ] = [x1 , x2 , x3 , −x4 , −x5 , −x6 , x7 , x8 , x9 , −x10 , −x11 , −x12 ].
Then (T1 , T2 ) defines a (Z2 )2 -action on RP (11) with fixed point set being
four copies of RP (2). Let (Z2 )0 act as the identity on the rest of the base.
2k +8
Since 22 + 22 + 2k − 9 − 2 = 2k − 3 < 2k , by Lemma 2.3, xn ∈ Jn,k
. By
Lemma 2.1, xn is indecomposable.
Case 3 — n = 2k + 2r1 and k > r1 ≥ 4.
If r1 > 4, we take xn = [RP (2n − 2k+1 − 17, 1, 1; 2k+1 − n + 16)]. From
2k +8
[22; Lemma 2.3], xn ∈ Jn,k
.
!
n−1
+ 2k+1 − n + 16 − 3
+2
1
Ã
!
2k + 2r1 −1 + · · · + 2 + 1
≡
+1
2r1 + · · · + 25 + 23 + 22 + 2
Ã
n−1
2n − 2k+1 − 17
!
Ã
≡ 1 mod 2.
By Lemma 2.1, xn is indecomposable.
If r1 = 4, we take xn = [RP (19, 4; 2k − 6)]. Let
T1 [x1 , x2 , · · · , x20 ] = [x1 , x2 , · · · , x10 , −x11 , −x12 , · · · , −x20 ],
T2 [x1 , x2 , · · · , x20 ] = [x1 , · · · , x5 , −x6 , · · · , −x10 , x11 , · · · , x15 , −x16 , · · · , −x20 ].
Then (T1 , T2 ) defines a (Z2 )2 -action on RP (19) with fixed point set being
four copies of RP (4). Let (Z2 )0 act as the identity on the rest of the base.
2k +8
By Lemma 2.3, xn ∈ Jn,k
.
Ã
2k + 24 − 1
19
!
Ã
+
2k + 24 − 1
4
!
+ 2k − 8 ≡ 1 mod 2.
By Lemma 2.1, xn is indecomposable.
¤
k
2 +8
Lemma 3.3 — There exist indecomposable classes xn ∈ J∗,k
for k ≥ 4,
k+1
k+2
2
≤n<2
and n even.
614
JINGYAN LI AND YANYING WANG
Proof : We will consider the following cases.
Case 1 — 2k+1 ≤ n ≤ 2k+1 + 2k and k ≥ 5.
Let xn = [RP (2k + 1, 1, n − 2k − 2k−1 − 8; 2k−1 + 7)]. From [22; Lemma
2k +4
2.3], xn ∈ Jn,k
.
Ã
! Ã
! Ã
!
n−1
n−1
n−1
+
+
+ 2k−1 + 7 − 3
2k + 1
1
n − 2k − 2k−1 − 8
Ã
! Ã
!
n−1
n−1
≡
+
+ 1.
2k + 1
2k + 2k−1 + 7
If n = 2k+1 ,
! Ã
!
Ã
2k+1 − 1
2k+1 − 1
+
+ 1 ≡ 1 + 1 + 1 ≡ 1 mod 2.
2k + 1
2k + 2k−1 + 7
If n = 2k+1 + 2k ,
!
! Ã
Ã
2k+1 + 2k − 1
2k+1 + 2k − 1
+ 1 ≡ 0 + 0 + 1 ≡ 1 mod 2.
+
2k + 2k−1 + 7
2k + 1
If n = 2k+1 + 2r1 + · · · + 2rm and r1 < 2k ,
Ã
! Ã
!
2k+1 + 2r1 + · · · + 2rm − 1
2k+1 + 2r1 + · · · + 2rm − 1
+
+1
2k + 1
2k + 2k−1 + 7
≡ 0 + 0 + 1 ≡ 1 mod 2.
By Lemma 2.1, xn is indecomposable.
Case 2 — 2k+1 ≤ n ≤ 2k+1 + 2k and k = 4.
k
2 +8
If n = 25 , we take xn = [RP (26; 7)]. From [15; Lemma 2.5], xn ∈ Jn,k
.
By Lemma 2.1, xn is indecomposable.
If 25 < n ≤ 25 + 24 , we take xn = [RP (11, 11, n − 34; 13)]. From [22;
2k +8
Lemma 2.3], xn ∈ Jn,k
.
Ã
! Ã
!
Ã
!
n−1
n−1
n−1
2
+
+ 13 − 3 ≡
≡ 1 mod 2,
11
n − 34
25 + 1
By Lemma 2.1, xn is indecomposable.
(Z2 )k -ACTIONS WITH FIXED POINT SET
615
Case 3 — 2k+1 + 2k < n < 2k+2 and k ≥ 4.
Suppose n = 2k+1 +2r1 +2r2 +· · ·+2rs , k = r1 > r2 > · · · > rs ≥ 1, s ≥ 2.
Write S = {i|there exists some i in the 2-adic expansion of n such that ri >
ri+1 + 1}.
(1) If S 6= ∅, let l = min{i | i ∈ S} and p = 2rl+1 + 2rl+2 + · · · + 2rs .
(i) If p > 9, then k ≥ 4. Take xn = [RP (2p + 1, 1, n − 2k − p − 8; 2k −
2k +8
p + 7)]. From [22; Lemma 2.3], xn ∈ Jn,k
.
Ã
!
n−1
+
+
+ 2k − p + 7 − 3
n − 2k − 2p − 8
Ã
!
2k+1 + 2r1 + · · · + 2rl + 2rl+1 · · · + 2rs − 1
≡
+
2rl+1 +1 + 2rl+2 +1 + · · · + 2rs +1 + 1
Ã
!
2k+1 + 2r1 + · · · + 2rl + 2rl+1 · · · + 2rs − 1
+1
2k + 2r1 + · · · + 2rl − 1 − 7
n−1
2p + 1
!
Ã
n−1
1
!
Ã
≡ 1 mod 2.
By Lemma 2.1, xn is indecomposable.
(ii) If 2 ≤ p ≤ 8, the argument is divided into the following cases:
(a) p = 8 or p = 4.
Take xn = [RP (19, n − 2k − 17; 2k − 1)]. From [15; Lemma 2.4], xn ∈
Jn,k . If p = 8, then rl ≥ 5.
2k +8
Ã
Ã
!
n−1
19
Ã
+
!
n−1
n − 2k − 17
+ 2k − 1 − 2
!
2k+1 + 2r1 + · · · + 2rl + 8 − 1
≡
24 + 2 + 1
!
Ã
2k+1 + 2r1 + · · · + 2rl + 8 − 1
+1
+
2k + 2r1 + · · · + 2rl − 8 − 1
!
Ã
2k+1 + 2r1 + · · · + 2rl + 8 − 1
+1
≡ 0+
2k + 2r1 + · · · + 2rl −1 + · · · + 24 + 22 + 1
≡ 0 + 0 + 1 ≡ 1 mod 2.
616
JINGYAN LI AND YANYING WANG
If p = 4, then rl ≥ 4.
Ã
! Ã
!
n−1
n−1
+
+ 2k − 1 − 2
19
n − 2k − 17
Ã
!
2k+1 + 2r1 + · · · + 2rl + 22 − 1
≡
24 + 2 + 1
Ã
!
2k+1 + 2r1 + · · · + 2rl + 22 − 1
+
+ 1. (∗)
2k + 2r1 + · · · + 2rl − 1 − 8 − 4
For rl > 4
Ã
(∗) ≡ 0 +
2k+1 + 2r1 + · · · + 2rl + 2 + 1
2k + 2r1 + · · · + 2rl+1 + 2rl −1 + · · · + 2 + 1 − 8 − 4
!
+1
≡ 0 + 0 + 1 ≡ 1 mod 2.
For rl = 4
Ã
(∗) ≡ 1 +
2k+1 + 2r1 + · · · + 24 + 2 + 1
2k + 2r1 + · · · + 2rl+1 + 2 + 1
!
+1
≡ 1 + 1 + 1 ≡ 1 mod 2.
From above, we know that xn is indecomposable.
(b) p = 6.
If p = 6, then rl ≥ 4. For rl > 4, we take xn = [RP (19, n−2k −17; 2k −1)].
2k +8
From [15; Lemma 2.4], xn ∈ Jn,k
.
Ã
Ã
n−1
19
!
Ã
+
n−1
n − 2k − 17
!
+ 2k − 1 − 2
!
2k+1 + 2r1 + · · · + 2rl + 4 + 1
≡
24 + 2 + 1
!
Ã
2k+1 + 2r1 + · · · + 2rl + 4 + 1
+1
+
2k + 2r1 + · · · + 2rl − 1 − 8 − 2
≡ 0 + 0 + 1 ≡ 1 mod 2.
By Lemma 2.1, xn is indecomposable.
(Z2 )k -ACTIONS WITH FIXED POINT SET
617
For rl = 4, we take xn = [RP (2k+1 + 1, 2, 2, 2, 2, n − 2k+1 − 16, 0, 0)].
2k +8
From [15; Lemma 2.4], xn ∈ Jn,k
.
Ã
!
2k+1 + 2r1 + · · · + 24 + 22 + 2 − 1
2k+1 + 1
Ã
!
2k+1 + 2r1 + · · · + 24 + 22 + 2 − 1
+
2r1 + · · · + 24 + 22 + 2 − 24
≡ 1 + 0 ≡ 1 mod 2
By Lemma 2.1, xn is indecomposable.
(c) p = 2.
If p = 2, then rl ≥ 3.
For rl > 4, we take xn = [RP (19, n − 2k − 17; 2k − 1)]. From [15; Lemma
2k +8
2.4], xn ∈ Jn,k
.
Ã
!
2k+1 + 2r1 + · · · + 2rl + 2 − 1
24 + 2 + 1
Ã
!
2k+1 + 2r1 + · · · + 2rl + 2 − 1
+
+1
2k + 2r1 + · · · + 2rl − 1 − 8 − 4 − 2
≡ 0 + 0 + 1 ≡ 1 mod 2.
By Lemma 2.1, xn is indecomposable.
For rl = 4 or 3, we take xn = [RP (2k+1 + 1, 2, 2, 2, 2, n − 2k+1 − 16, 0, 0)].
k
2 +8
From [15; Lemma 2.4], xn ∈ Jn,k
.
Ã
2k+1 + 2r1 + · · · + 2rl + 2 − 1
2k+1 + 1
!
Ã
+
2k+1 + 2r1 + · · · + 2rl + 2 − 1
2r1 + · · · + 2rl + 2 − 24
≡ 1 + 0 ≡ 1 mod 2.
By Lemma 2.1, xn is indecomposable.
(2) If S = ∅, i.e. n = 2k+1 + 2r1 + 2r2 + · · · + 2rs , k = r1 > r2 > · · · >
rs ≥ 1, ri = ri+1 + 1 and s ≥ 2.
!
618
JINGYAN LI AND YANYING WANG
2rs +1 − 8 2rs +1 − 8
,
, n − 2k+1 −
2
2
2k +8
2rs +1 ); 8]. From [15; Lemma 2.4], xn ∈ Jn,k
.

 Ã
Ã
!
!
n
−
1
n−1
n
−
1
+ 2  2rs +1 − 8  +
+8−4
2k+1 + 1
n − 2k+1 − 2rs +1 )
2
Ã
! Ã
!
k+1
r
r
1
2
2
+ 2 + 2 + · · · + 2rs − 1
2k+1 + 2r1 + 2r2 + · · · + 2rs − 1
≡
+
2k+1 + 1
2r1 + 2r2 + · · · + 2rs − 2rs +1
Ã
!
2k+1 + 2r1 + 2r2 + · · · + 2rs−1 + 2rs −1 + · · · + 2 + 1
≡ 1+
2r1 + 2r2 + · · · + 2rs−2 + 2rs
For rs ≥ 2, we take xn = [RP (2k+1 + 1,
≡ 1 + 0 ≡ 1 mod 2.
By Lemma 2.1, xn is indecomposable.
For rs = 1, we take xn = [RP (2k+1 + 1, n − 2k+1 − 8; 8)]. From [15;
2k +8
Lemma 2.4], xn ∈ Jn,k
.
Ã
! Ã
!
2k+1 + 2r1 + 2r2 + · · · + 2 − 1
2k+1 + 2r1 + 2r2 + · · · + 2 − 1
+
2k+1 + 1
2r1 + 2r2 + · · · + 2 − 8
≡ 1 + 0 ≡ 1 mod 2.
By Lemma 2.1, xn is indecomposable.
¤
Proposition 3.4 — If n ≥ 2k + 9 and n 6= 2u − 1, then there exist inde2k +8
composable classes xn ∈ Jn,k
for k ≥ 4.
Proof : Take xn as in Lemma 3.1 for n odd, xn as in Lemma 3.2 and
Lemma 3.3 for 2k + 10 ≤ n < 2k+2 even, and xn as in [15; Lemma 3.1] for
2k +8
n ≥ 2k+2 even. Then, every xn is indecomposable and xn ∈ Jn,k
.
¤
To complete the proof of the main theorem, we need the following two
lemmas.
Lemma 3.5 — For k ≥ 4 and n = 2k + 7, there exist indecomposable
2k +6
classes xn ∈ Jn,k
.
Proof : From [11; 5.1], there exists an indecomposable class X ∈
k−1
J 2n−1 ,k+3 (for 2k−1 + 3 < 2k ), so that X has a representative M admitting a
2
(Z2 )k -action with fixed point set F 0 , where dim(F 0 ) =
n−1
2
− (2k−1 + 3) = 0.
(Z2 )k -ACTIONS WITH FIXED POINT SET
619
Take xn = [P (1, M )] with the (Z2 )k -action as in Lemma 3.1. The fixed point
set of this action is F = S 1 ×F 0 ×F 0 /T = P (1, F 0 ) with dimF = 1 = n−2k −6,
2k +6
then xn ∈ J∗,k
. By Lemma 2.2, xn is indecomposable.
¤
Lemma 3.6 — For k ≥ 4 and n = 2k + 8, there exist indecomposable
2k +6
classes xn ∈ Jn,k
.
Proof : Take xn = [RP (11; 2k −6)]. Just as in Lemma 3.2, there exsits a
(Z2 )2 -action on RP (11) with fixed point set being four copies of RP (2). Let
2k +6
(Z2 )0 act as the identity on the rest of the base. By Lemma 2.3, xn ∈ Jn,k
.
Ã
2k + 8 − 1
8+2+1
!
+ 2k − 6 − 1 ≡ 1 mod 2.
By Lemma 2.1, xn is indecomposable.
¤
4. Proof of the Main Theorem
Proof : We choose a system of generators xn as follows:
(a) Let x2 = [RP (2)]. Noticing that χ(x2 ) = 1, by [11; 5.1], we have
2 .
x2 ∈ J∗,k
(b) For 3 ≤ n ≤ 2k + 6 and n 6= 2u − 1, we can choose indecomposable
n/2
classes xn such that χ(xn ) = 0 (otherwise, replacing xn by xn + x2 ).
(c) Take xn as in Lemma 3.5 for n = 2k + 7 and xn as in Lemma 3.6 for
2k +8
n = 2k + 8. By Lemma 2.4, x2k +8 ∈
/ J∗,k
.
(d) For n ≥ 2k + 9, let xn be as in Proposition 3.4.
k
2 +8
Since J∗,k
is an ideal of M O∗ , to complete the proof it is necessary
only to show that it contains the decomposable classes xi1 xi2 · · · xim with
m ≥ 2, 2 ≤ i1 ≤ i2 ≤ · · · ≤ im ≤ 2k + 8 and i1 + i2 + · · · + im ≥ 2k + 8.
The argument is divided into two cases: Case 1. i1 +i2 +· · ·+im > 2k +8.
Case 2. i1 + i2 + · · · + im = 2k + 8.
Case 1 — i1 + i2 + · · · + im > 2k + 8.
(1) The case im > 2k .
620
JINGYAN LI AND YANYING WANG
8 . From
(i) If i1 + i2 + · · · + im−1 ≥ 8, from [11; 5.1] xi1 xi2 · · · xim−1 ∈ J∗,k
k
k
k
2 +8
2 and x x · · · x
8
2
[12], xim ∈ J∗,k
i1 i2
im ∈ J∗,k J∗,k ⊂ J∗,k .
4
(ii) If i1 + i2 + · · · + im−1 = 7, then xi1 xi2 · · · xim−1 ∈ J∗,k
T
7 and
J∗,k
k
2 +4
im > 2k +1. If im > 2k +4, by [18; Theorem] xim ∈ J∗,k
and xi1 xi2 · · · xim ∈
k
k
4 J 2 +4 ⊂ J 2 +8 .
J∗,k
∗,k
∗,k
k
2 +1
.
If im ≤ 2k + 4, from (b), χ(xim ) = 0. By [14; Theorem 1], xim ∈ J∗,k
k
k
7 J 2 +1 ⊂ J 2 +8 .
So xi1 xi2 · · · xim ∈ J∗,k
∗,k
∗,k
k
2 +2
(iii) If i1 + i2 + · · · + im−1 = 6, then im > 2k + 2. From [16], xim ∈ J∗,k
.
k
k
2 +8
2 +2
6 . Then x x · · · x
6
⊂ J∗,k
.
By [11; 5.1], xi1 xi2 · · · xim−1 ∈ J∗,k
i1 i2
im ∈ J∗,k J∗,k
T 5
4
(iv) If i1 + i2 + · · · + im−1 = 5, then xi1 xi2 · · · xim−1 ∈ J∗,k
J∗,k and
k
2 +4
im > 2k +3. If im > 2k +4, by [18; Theorem], xim ∈ J∗,k
and xi1 xi2 · · · xim ∈
k
k
4 J 2 +4 ⊂ J 2 +8 .
J∗,k
∗,k
∗,k
If im = 2k + 4, from (b), χ(xim ) = 0.
k
k
2 +3
5 J 2 +3 ⊂
By [15; Proposition 4.3], xim ∈ J∗,k
. Then xi1 xi2 · · · xim ∈ J∗,k
∗,k
k
2 +8
J∗,k
.
(v) If i1 +i2 +· · ·+im−1 = 4, then im > 2k +4. From [18; Theorem], xim ∈
2k +4
4 . Then x x · · · x
4
By [11; 5.1], xi1 xi2 · · · xim−1 ∈ J∗,k
⊂
i1 i2
im ∈ J∗,k J∗,k
2k +4
J∗,k
.
2k +8
J∗,k .
2 and
(vi) If i1 + i2 + · · · + im−1 = 2, then m = 2, i1 = 2, x2 ∈ J∗,k
k
k
k
2 +6
2 J 2 +6 ⊂ J 2 +8 .
2k +6 < i2 ≤ 2k +8. From (c), xi2 ∈ J∗,k
. Then x2 xi2 ∈ J∗,k
∗,k
∗,k
(2) The case im ≤ 2k and there exists some l(1 ≤ l ≤ m) such that
8 ≤ il < 2k .
il
From [11; 5.1], xil ∈ J∗,k
. Since i1 + i2 + · · · + im > 2k + 8, i1 +
i2 + · · · + il−1 + il+1 + · · · + im > 2k + 8 − il . Because 2k + 8 − il ≤ 2k ,
2k +8−il
from [11; 5.1] and [12], we have xi1 xi2 · · · xil−1 xil+1 · · · xim ∈ J∗,k
. Then
k
k
2 +8−il il
2 +8
xi1 xi2 · · · xim ∈ J∗,k
J∗,k ⊂ J∗,k
.
(3) The case im ≤ 2k , with no l(1 ≤ l ≤ m) satisfying 8 ≤ il < 2k .
In this case, ij = 2k , or ij ≤ 6(1 ≤ j ≤ m).
(Z2 )k -ACTIONS WITH FIXED POINT SET
621
2k −2 T 2k −1
(i) If im = 2k , then i1 + i2 + · · · + im−1 ≥ 9 and xim ∈ J∗,k
J∗,k .
If there are odd numbers in i1 , · · · , im−1 , then χ(xi1 xi2 · · · xim−1 ) = 0 and
2k −l
2k +8
9 . So x x · · · x
9
xi1 xi2 · · · xim−1 ∈ J∗,k
i1 i2
im ∈ J∗,k J∗,k ⊂ J∗,k . If i1 , · · · , im−1
10 . Then
are all even, i1 + i2 + · · · + im−1 ≥ 10 and xi1 xi2 · · · xim−1 ∈ J∗,k
k
k
10 J 2 −2 ⊂ J 2 +8 .
xi1 xi2 · · · xim ∈ J∗,k
∗,k
∗,k
im
(ii) If im ≤ 6 and there exists some t such that it 6= 2, then xim ∈ J∗,k
and i1 +i2 +· · ·+im−1 > 2k +8−im . From [15],16] and [18], xi1 xi2 · · · xim−1 ∈
2k +8−im
2k +8−im im
2k +8
J∗,k
. Then xi1 xi2 · · · xim ∈ J∗,k
J∗,k ⊂ J∗,k
.
(iii) If i1 = i2 = · · · = im = 2, then 2(m − 4) > 2k . From [11; 5.1] and
4 m−4 ∈ J 8 J 2k ⊂ J 2k +8 .
[12], xm
2 = x2 x2
∗,k ∗,k
∗,k
Case 2 — i1 + i2 + · · · + im = 2k + 8.
ir
If im < 2k , from [11; 5.1], xir ∈ J∗,k
(r = 1, 2, · · · , m) and
k
i1 i2
im
2 +8
xi1 xi2 · · · xim ∈ J∗,k
J∗,k · · · J∗,k
⊂ J∗,k
.
If im ≥ 2k , then s(i1 ,i2 ,··· ,im ) [xi1 xi2 · · · xim ] 6= 0. By Lemma 2.5, the decomposable classes x2 x2k +6 , x4 x2k +4 , x2 2 x2k +4 , x5 x2k +3 , x6 x2k +2 , x2 x4 x2k +2 ,
2k +8
x2 3 x2k +2 , x2 x5 x2k +1 , x8 x2k , x2 x6 x2k , x4 2 x2k , x2 2 x4 x2k , x2 4 x2k are not in J∗,k
.
For some linear combination of the above classes such as
xi1,1 xi2,1 · · · xim1 ,1 + xi1,2 xi2,2 · · · xim2 ,2 + · · · + xi1,j xi2,j · · · ximj ,j ,
we can find
(i1 , i2 , · · · , inj ) ∈ {(i1,1 , i2,1 · · · im1 ,1 ), (i1,2 , i2,2 · · · im2 ,2 ), · · · , (i1,j , i2,j , · · · imj ,j )}
such that (i1 , i2 , · · · , inj ) is not the refinement of other elements in
{(i1,1 , i2,1 · · · im1 ,1 ), (i1,2 , i2,2 · · · im2 ,2 ), · · · (i1,j , i2,j , · · · imj ,j )}.
Then, from [23],
s(i1 ,i2 ,··· ,inj ) [xi1,1 xi2,1 · · · xim1 ,1 +xi1,2 xi2,2 · · · xim2 ,2 +· · ·+xi1,j xi2,j · · · ximj ,j ] 6= 0.
By Lemma 2.5, any linear combination of the above classes are not in
2k +8
J∗,k
.
¤
622
JINGYAN LI AND YANYING WANG
Acknowledgement
We would like to thank the referee for a careful reading of the paper and
many helpful suggestions.
References
1. P. E. Conner, Differentiable periodic maps (second edition), Lecture Notes in
Math, Univ. Berlin, Springer Verlag. 1979.
2. P. E. Conner and E. E. Floyd, Fibering within a cobordism class, Michigan
Math. J., 12 (1965), 33-47.
3. R. E. Stong, On fibering of cobordism classes, Tran. Amer. Math. Soc., 178
(1973), 431-447.
4. F. L. Capobianco, Cobordism classes represented by fiberings with fiber
RP (2k + 1), Michigan Math. J., 24 (1977), 185-192.
5. F. L. Capobianco, Manifolds with involution whose fixed point set has codimension four, Proc. Amer. Math. Soc., 61 (1976), 157-162.
6. K. Iwata, Involutions with fixed point set of constant codimension, Proc.
Amer. Math. Soc., 83 (1981), 829-832.
7. T. Wada, Manifolds with involution whose fixed point set has codimension
Six, Bull. Yamagata Univ. Nature Sci., 10 (1981), 193-250.
8. Z. D. Wu, Manifolds with involution whose fixed point set has codimension
seven, Acta Math. Sin., New Series, 4 (1989), 302-306.
9. S. Kikuchi, Involution with fixed point set of codimension eight, Sci. Rep.
Hirosaki Univ., 33 (1986), 45-52.
10. P. L. Pergher, (Z2 )2 -actions with fixed point set of constant codimension,
Topology Appl., 46 (1992), 55-64.
11. R. J. Shaker, Constant codimension fixed point sets of commuting involutions, Proc. Amer. Math. Soc., 121 (1994), 275-281.
12. R. J. Shaker, Dold manifolds with (Z2 )k -action, Proc. Amer. Math. Soc.,
123 (1995), 955-958.
13. C. Kosniowski and R. E. Stong, (Z2 )k -actions and characteristic numbers, Indiana Univ. Math. J., 28 (1979), 725-743.
(Z2 )k -ACTIONS WITH FIXED POINT SET
14. Y. Y. Wang, Z. D. Wu and K. Ma, (Z2 )k -action with fixed point set of constant codimension 2k + 1, Proc. Amer. Math. Soc., 128 (1999), 1515-1521.
15. Y. Y. Wang, Z. D. Wu and Y. H. Ding, Commuting involutions with fixed
point set of constant codimension, Acta Math. Sin., English Series, 15 (1999),
181-186.
16. Z. Liu and Z. D. Wu, (Z2 )k -actions with fixed point set of constant codimension 2k + 2, Acta Math. Sin., English Series, 21 (2005), 409-412.
17. R. L. W. Brown, Immersions and embeddings up to cobordism, Can. J.
Math., Vol. XXIII (1971), 1102-1115.
18. R. C. Li, K. Ma and Z. D. Wu, (Z2 )k -actions with fixed point set of constant
codimension 2k + 4, J. Math. Res. Exp., 25 (2005), 659-664.
19. X. F. Feng, (Z2 )k -actions with fixed point set of constant codimension 2k + 5,
Master’s thesis (2006), Hebei Normal University.
20. J. Geng, (Z2 )k -actions with fixed point set of constant codimension 2k + 6,
Master’s thesis(2006), Hebei Normal University.
21. J. Y. He, (Z2 )k -actions with fixed point set of constant codimension 2k + 7,
Master’s thesis (2007), Hebei Normal University.
22. Y. Y. Wang, Z. D. Wu and Z. J. Meng, Commuting involutions with fixed
point set of variable codimension, Acta Math. Sin., English Series, 17 (2001),
425-430.
23. Y. Y. Wang, Characteristic numbers and cobordism class of fiberings with
fiber RP (2k), Proc. Amer. Math. Soc., 129 (2000), 899-911.
623