On the determinant and the holonomy of equivariant elliptic operators

On the determinant and the holonomy of equivariant elliptic operators
复制标题

等变椭圆算子的行列式和完整性

DOI:
10.1090/s0002-9939-1995-1260183-9
复制
发表时间:
1995
期刊:
--
影响因子:
--
通讯作者:
K. Tsuboi
K. Tsuboi
中科院分区:
--
文献类型:
--
作者:
K. Tsuboi

文献摘要

被引文献

相似文献

设 M 为闭导向光滑流形,G 为由 M 的微分同胚组成的紧李群,P -+ Z 为具有连接的主 G 丛,D 为 G 等变椭圆算子。然后,椭圆算子的局部常数族及其在 Z 上的行列式线束自然由 D 定义。此外,行列式线束的完整性由 P 中的连接定义。在本文中,我们给出了计算完整性的显式公式(定理 1.4),并给出了上述特殊情况下的 Witten 完整性公式(定理 1.7)的证明。 1. 主要结果 设 M 为闭导向光滑流形,G 为由 M 的微分同胚组成的紧李群,P -* Z 为具有连接的光滑流形 Z 上的主 G 丛,D 为 G 等变椭圆算子。然后,对于任何 g e G,在 g 处评估的 D 指数 Index(D, g) 由 Index(D, g) = tr(glker D) tr(glcokerD) 定义,并且可以通过众所周知的定点公式计算(参见 [1]、[2] 或 [5])。另一方面,在 g 处计算的 D 行列式 det(D, g) 由 det(D, g) = det(glker D)/ det(gIcokerD) 定义。那么下一个命题是附录中引理 1 基本结果的直接结果。命题1.1。令 g e G 为有限阶 p 的任意元素。那么下一个等式成立: det(D, g) = exp p E 1 -e-2?iklp {Index(D) -Index(D, gk)} 其中 Index(D) = Index(D, 1) 是 D 的数字索引。编辑于 1993 年 11 月 2 日收到。1991 年数学学科分类。主要58G26;次要58G10。
Let M be a closed oriented smooth manifold, G a compact Lie group consisting of diffeomorphisms of M, P -+ Z a principal G-bundle with a connection and D a G-equivariant elliptic operator. Then a locally constant family of elliptic operators and its determinant line bundle over Z are naturally defined by D . Moreover the holonomy of the determinant line bundle is defined by the connection in P. In this note, we give an explicit formula to calculate the holonomy (Theorem 1.4) and give a proof of the Witten holonomy formula (Theorem 1.7) in the special case above. 1. MAIN RESULTS Let M be a closed oriented smooth manifold, G a compact Lie group consisting of diffeomorphisms of M, P -* Z a principal G-bundle over a smooth manifold Z with a connection and D a G-equivariant elliptic operator. Then, for any g e G, the index of D evaluated at g, Index(D, g), is defined by Index(D, g) = tr(glker D) tr(glcokerD) and can be calculated by the well-known fixed point formula (cf. [1], [2] or [5]). On the other hand, the determinant of D evaluated at g, det(D, g), is defined by det(D, g) = det(glker D)/ det(gIcokerD). Then the next proposition is an immediate consequence of the elementary result of Lemma 1 in Appendix. Proposition 1.1. Let g e G be any element of finite order p. Then the next equality holds: det(D, g) = exp p E 1 -e-2?iklp {Index(D) -Index(D, gk)} where Index(D) = Index(D, 1) is the numerical index of D. Received by the editors November 2, 1993. 1991 Mathematics Subject Classification. Primary 58G26; Secondary 58G10.