The arithmetic Grothendieck-Riemann-Roch theorem for general projective morphisms (Q476071)

The Grothendieck-Riemann-Roch theorem plays an essential role in the study of algebraic geometry. Roughly speaking, it describes how far the Chern character map between the algebraic \(K\)-theory and the algebraic Chow-theory to be commutative with the push-forward operations w.r.t. proper morphisms. This theorem vastly generalizes the classical Riemann-Roch theorem on compact Riemann surfaces, and hence it can be viewed as an Atiyah-Singer type index theorem. In previous work, the Grothendieck-Riemann-Roch theorem was extended by several mathematicians to an arithmetic setting in the sense of Arakelov. In this situation, the generators of algebraic \(K\)-groups and algebraic Chow groups are equipped with additional Hermitian structures so that the fibers of an arithmetic variety over the infinite places can be treated as the same footing as the finite ones. Historically, \textit{H. Gillet} and \textit{C. Soulé} [Invent. Math. 110, No. 3, 473--543 (1992; Zbl 0777.14008)] proved a degree one version of the arithmetic Grothendieck-Riemann-Roch theorem, they used the theory of Ray-Singer analytic torsion to define a reasonable push-forward on the level of arithmetic \(K\)-groups. Whence \textit{J.-M. Bismut} and \textit{K. Köhler} [J. Algebr. Geom. 1, No. 4, 647--684 (1992; Zbl 0784.32023)] developed the analytic torsion theory to higher degree case, a complete proof of the arithmetic Grothendieck-Riemann-Roch was given by Gillet, Rössler and Soulé [\textit{H. Gillet} et al., Ann. Inst. Fourier 58, No. 6, 2169--2189 (2008; Zbl 1152.14023)]. But, due to technical reason, these two works had to restrict to push-forward by morphisms which are smooth over generic fibers. The aim of the article under review is to break this limit, namely is to generalize the works of Gillet-Soulé and Gillet-Rössler-Soulé to arbitrary projective morphisms, not necessarily smooth over the generic fibers. This forces the authors to enlarge the arithmetic \(K\)-groups and the arithmetic Chow groups to afford corresponding push-forward functorialities. The strategy the authors followed was replacing smooth differential forms in the theory of Gillet-Soulé by currents with possibly non-empty wave front sets. As a consequence, by explaining the relationship between the topological correction term and the generalized analytic torsion classes, the authors obtained all the possible forms of the arithmetic Grothendieck-Riemann-Roch theorem.