On the signature of four-manifolds with universal covering spin
On the signature of four-manifolds with universal covering spin
复制标题
DOI:
10.1007/bf01444915
复制
发表时间:
1993
影响因子:
1.4
通讯作者:
P. Teichner
中科院分区:
文献类型:
--
作者:
P. Teichner
In this note we study closed oriented 4-manifolds whose universal covering is spin and ask whether there are restrictions on the divisibility of the signature. Since any natural number appears as the signature of a connected sum of r 2, s, without the assumption on the universal covering there cannot exist any restrictions. Certainly, the most famous such restriction was proved by Rohlin in [10], where he showed that the signature a of a smooth 4-dimensional spin manifold is divisible by 16 (compare part (2) of our Main Theorem for a new proof). The Kummer surface K shows that this is the best possible general result. Dividing by a certain free holomorphic involution on K, one obtains the Enriques surface (compare [1]) which by construction has signature 8 and fundamental group 7//2. Furthermore, Hitchin showed in [5] that there exists an antiholomorphic free involution on the Enriques surface. We will refer to the quotient as the Hitchin manifold which then has signature 4 and fundamental group 7//2 x 7//2. Rohlin's theorem admits a nice generalization to nonspin 4-manifolds, compare [4, Theorem 6.3]: a (M)= FoF-2-fl (F) mod 16.Here F is a (not necessarily orientable) surface in M which is dual to W2 M. This implies that there exists a Pin--structure on M\F which induces a Pin--structure on F and hereby a quadratic refinement of the 7//2-intersection form on F. Thus the 77/8-valued Brown-Aft invariant fl (F) can be defined. Starting with a surface F, the above formula suggests that one can construct further examples of 4-manifolds with small signature and universal covering spin. But so far, all attempts using this method failed and therefore we tried to find a different way of attacking the problem. This was motivated by the questions of several people at the Oberwolfach Topology Conference in 1990 who wanted to know whether the signature of all closed oriented 4-manifolds whose universal