首页 | 本学科首页   官方微博 | 高级检索  
     


Von Neumann Categories and Extended L2 Cohomology
Authors:Michael Farber
Affiliation:(1) School of Mathematical Sciences, Tel"dash"Aviv University, Ramat"dash"Aviv, 69978, Israel Email
Abstract:In this paper we suggest a new general formalism for studying the L2 invariants of polyhedra and manifolds. First, we examine generality in which one may apply the construction of the extended Abelian category, which was earlier suggested by the author using the ideas of P. Freyd. This leads to the notions of a finite von Neumann category and a trace on such a category. Given a finite von Neumann category, we study the extended L2 homology and cohomology theories with values in the Abelian extension. Any trace on the initial category produces numerical invariants – the von Neumann dimension and the Novikov–Shubin numbers. Thus, we obtain the local versions of the Novikov–Shubin invariants, localized at different traces. In the lsquoAbelianrsquo case this localization can be made more geometric: we show that any torsion object determines a lsquodivisorrsquo – a closed subspace of the space of the parameters. The divisors of torsion objects together with the information produced by the local Novikov–Shubin invariants may be used to study multiplicities of intersections of algebraic and analytic varieties (we discuss here only simple examples demonstrating this possibility). We compute explicitly the divisors and the von Neumann dimensions of the extended L2 cohomology in the real analytic situation. We also give general formulae for the extended L2 cohomology of a mapping torus. Finally, we show how one can define a De Rham version of the extended cohomology and prove a De Rhamdashtype theorem.
Keywords:L2  /content/v3w3137162255h03/xxlarge8208.gif"   alt="  dash"   align="  MIDDLE"   BORDER="  0"  >cohomology  Novikov  /content/v3w3137162255h03/xxlarge8208.gif"   alt="  dash"   align="  MIDDLE"   BORDER="  0"  >Shubin invariants
本文献已被 SpringerLink 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号