No label defined (Q4952545)

This paper is a survey of the refinements of index theory associated with the local index formulas that arise from heat equation techniques. Local index formulas are expressed in terms of differential forms. Different perspectives on the same index problem may yield different representatives of the associated index class. Thus there are secondary invariants that transgress the index theorem to measure the difference between representative differential forms. Much of this paper is devoted to the eta forms that transgress the families index theorem of a fibration and to the (holomorphic) analytic torsion forms that refine the eta forms for a Kähler fibration. One perspective on the families index theorem involves the index bundle while the other perspective draws information from the geometry of the fibration's total space. Hence this approach to the families index theorem requires the full power of Quillen superconnections in an infinite-dimensional setting. Among the other topics discussed in this article are: the Atiyah-Patodi-Singer index theorem (including the approaches of J. Cheeger and R. Melrose); the families version of the Atiyah-Patodi-Singer theorem; adiabatic limits of secondary invariants; the geometry of determinant line bundles; analytic torsion currents associated with immersions; and the functoriality of analytic torsion forms and currents. This article does not discuss the role of heat equation techniques in noncommutative geometry. It does mention the relation of the surveyed material with Arakelov geometry. Much of the material discussed in this paper first appeared in long papers by the author (some with co-authors). The author has also written other surveys with somewhat different emphases. The present paper would be an excellent resource for a reader who has some familiarity with the field and who wants to learn how an expert views local index theory (with particular attention to its analytic aspects). This paper is clearly written, informative, and self-contained in the sense that it contains definitions, theorems, sketches of proofs, and motivating comments. The organization of the paper clearly communicates the logical structure of the subject. However a reader unfamiliar with the field might be overwhelmed by the quantity of information in this paper. Such a reader might get a gentler introduction to eta forms and (non-holomorphic) analytic torsion forms by reading \textit{S. Rosenberg} [Bull. Am. Math. Soc., New Ser. 34, No. 4, 423-433 (1997; Zbl 0886.58110)] and \textit{J. Lott} [in Connes, A. (ed.) et al., Quantum symmetries/ Symétries quantiques, pp. 947-955 (North-Holland, Amsterdam) (1998; Zbl 0941.58015)]. For the entire collection see [Zbl 0918.00028].