Noncommutative Maslov index and \(\eta \)-forms reconsidered (Q347131): Difference between revisions

Recall that the Maslov triple index \(\tau(L_0, L_1, L_2)\) is an invariant associated to a triple of Lagrangian subspaces of a finite-dimensional symplectic vector space. It is related to the Maslov index \(\mu\) for a pair of paths of Lagrangians \(\mu(L_1(t), L_2(t))\) through the formula \[ \begin{aligned} & \tau(L_0(1), L_1(1), L_2(1))-\tau(L_0(0), L_1(0), L_2(0))\\ &=2(\mu(L_0(t), L_1(t))+\mu(L_1(t), L_2(t))+\mu(L_2(t), L_0(t))). \end{aligned} \] One can also associate the \(\eta\)-invariant to a pair of Lagrangians, which has the following relation with the Maslov triple index \[ \tau(L_0, L_1, L_2)=\eta(L_0, L_1)+\eta(L_1, L_2)+\eta(L_2, L_0). \] The author generalizes the above two formulas to the noncommutative setting in [\textit{C. Wahl}, J. Ramanujan Math. Soc. 22, No. 2, 135--187 (2007; Zbl 1136.58014)] and in [\textit{C. Wahl}, Mem. Am. Math. Soc. 887, 118 p. (2007; Zbl 1144.53099)], i.e., she considers \(C^*\)-modules instead of vector spaces. The author continues the study of the generalization of the Maslov index to the \(C^*\)-algebraic setting in the article under review. First she gives a short survey on the noncommutative analogue of the second formula and also establishes further properties of these invariants. Second, she corrects an error in the generalization of the second formula to the \(C^*\)-algebraic setting. She shows that there are choices involved in the definition of the noncommutative \(\eta\)-forms which cannot be eliminated, and how these corrections and new results affect the main result of [\textit{C. Wahl}, Mem. Am. Math. Soc. 887, 118 p. (2007; Zbl 1144.53099)].