K-theory and matrix transfers
Grigory Garkusha
TL;DR
The paper develops matrix transfers as a noncommutative analogue of Voevodsky framed transfers to model bivariant algebraic K-theory. It constructs symmetric and big symmetric matrix motives via spectral categories, and proves a network of equivalences linking matrix transfers to Waldhausen K-theory and to classical K-motives, providing a local, non-$\mathbb{A}^1$-localized framework for motivic K-theory. A central outcome is that symmetric matrix motives of varieties are stably equivalent to $K$-motives, thereby recovering Quillen’s $K$-theory through motivic homotopy techniques and allowing a clean passage to $KGL$-modules after inverting the exponential characteristic. The framework yields a robust toolkit for working with $K$-theory motivically, including a cancellation theorem and a concrete model for $\mathsf{kgl}$-modules, situating bivariant K-theory firmly inside enriched motivic homotopy theory. Overall, the work unifies transfers, matrix motives, and motivic K-theory into a coherent spectral-motive formalism with strong functorial and monoidal properties.
Abstract
We introduce and study matrix transfers to achieve elementary models for bivariant $K$-theory. They share lots of common properties with Voevodsky's framed correspondences and lead to symmetric matrix motives of algebraic varieties introduced in this paper. Symmetric matrix motives recover $K$-motives and fit in a closed symmetric monoidal triangulated category of symmetric matrix motives constructed in this paper by using methods of enriched motivic homotopy theory.
