On numerical equivalence for algebraic cobordism (Q886985)

The authors define here a quotient group of the algebraic cobordism group \(\Omega_*(-)\) of \textit{M. Levine} and \textit{F. Morel} [Algebraic cobordism. Berlin: Springer (2007; Zbl 1188.14015)] (for quasi-projective schemes of finite type over a field of characteristic zero) that is based on a proposed notion of numerical equivalence for algebraic cobordism cycles. This notion is a generalization and adaptation of the classical equivalent for algebraic cycles, where one would count the number of intersection points of two varieties, considering multiplicities. In this work, this form of counting is in a sense made using not the integers but the Lazard ring \(\mathbb{L}\) of \textit{D. Quillen} [Adv. Math. 7, 29--56 (1971; Zbl 0214.50502)]. The key notion here is that of numerically trivial algebraic cycles, which form an ideal \(\mathcal{N}_*(-)\) of \(\Omega_*(-)\). For smooth projective varieties \(X\) (in section 3), this is achieved by considering the composition \(\Omega_*(X) \otimes \Omega_*(X) \rightarrow \Omega_*(X \times X) \rightarrow \Omega_*(X) \rightarrow \Omega_*(k)\) (where one uses in sequence the external product, the l.c.i. pull-back by the diagonal morphism, and the structure morphism of \(X\)), and its induced map \(\Omega_*(X) \rightarrow \mathrm{Hom}_{\mathbb{L}} (\Omega_* (X), \Omega_*(k))\), whose kernel is defined as the intended \(\mathcal{N}_*(X)\). The group \(\Omega_*^{\mathrm{num}}(X)\) is then the quotient of the algebraic cobordism group by this ideal of numerically trivial cycles. Next, the properties of this new group are studied. The relations with the Chow group of \(X\) are presented, and it is shown that the canonical map \(\Omega_*(X) \rightarrow \mathrm{CH}(X)\) maps numerically trivial cobordism cycles into numerically trivial algebraic cycles and that \(\mathcal{N}_*(X) \otimes_{\mathbb{L}} \mathbb{Z}\) is isomorphic to the group of numerically trivial algebraic cycles of \(X\). Section 4 extends the definition of \(\Omega_*^{\mathrm{num}}(X)\) (made, in section 3, for smooth projective varieties), first to smooth quasi-projective varieties, and then to quasi-projective schemes. This is done by considering desingularizations of smooth quasi-projective varieties and by applying this definition to the irreducible components of quasi-projective schemes. The resulting quotient of the algebraic cobordism group (for quasi-projective schemes) is next proven to satisfy the required properties of an oriented Borel-Moore homology theory, including the localization sequence, the homotopy invariance and the projective bundle formula, all of which are explored in this section, culminating in Theorem 4.23. (As usual, a dual oriented cohomology theory \(\Omega^*_{\mathrm{num}}(X)\) is also obtained for smooth varieties.) In section 5, the authors explore the relations between the notions of numerically trivial, homologically trivial, and étale homologically trivial algebraic cobordism cycles. Homologically trivial cycles are those in the kernel of the canonical map \(\Omega^*(X) \rightarrow {\mathrm{MU}}^{2*}(X_\sigma(\mathbb{C}))\), where \(MU\) denotes Quillen's complex cobordism and \(\sigma : k \rightarrow \mathbb{C}\) is any embedding of \(k\) (the cohomology theories are defined only for smooth varieties \(X\)); and étale homologically trivial cycles are those in the kernel of \(\Omega^*(k;\mathbb{Z}_l) \rightarrow {\widehat{\mathrm{MU}}}^{2*}_{et}(k;\mathbb{Z}_l)\), where \(l\) is a prime, \(k\) is separably closed of characteristic zero, and \({\widehat{ \mathrm{MU}}}^{2*}_{\mathrm{et}}\) is Quick's étale cobordism. Theorems 5.1 and 5.2 show that, under the above mentioned conditions, homologically trivial cycles are numerically trivial, and that the same happens with étale homologically trivial cycles. These results are put in the wider context of the standard conjecture (in various settings, relating algebraic cobordism with complex cobordism, the Chow group, and singular cohomology), culminating in the results of Proposition 5.5. The final section applies the relations between rational smash nilpotence and homologically triviality of cobordism cycles (and Voevodsky's conjecture on these themes) to the new \(\Omega_*^{\mathrm{num}}(X)\). The main conclusions of this approach are found in Theorem 6.20 and its Corollary 6.21.