The geometry of vector bundles and derived categories on complex K3 surfaces has developed rapidly since Mukai's seminal work [Muk87]. Many foundational questions have been answered:• the existence of vector bundles and twisted sheaves with prescribed invariants; , and other authors. Our focus in this note is on rational points over non-closed fields of arithmetic interest. We seek to relate the notion of derived equivalence to arithmetic problems over various fields. Our guiding questions are: Question 1. Let X and Y be K3 surfaces, derived equivalent over a field F . Does the existence/density of rational points of X imply the same for Y ? Given α ∈ Br(X), let (X, α) denote the twisted K3 surface associated with α, i.e., if P → X is an étale projective bundle representing α, of relative dimension r − 1 then (X, α) = [P/ SL r ].Question 2. Suppose that (X, α) and (Y, β) are derived equivalent over F . Does the existence of a rational point on the former imply the same for the latter?Date: September 9, 2015. Note that an F -rational point of (X, α) corresponds to an x ∈ X(F ) such that α|x = 0 ∈ Br(F ). After this paper was released, Ascher, Dasaratha, Perry, and Zhou [ADPZ15] found that Question 2 has a negative answer, even over local fields.We shall consider these questions for F finite, p-adic, real, and local with algebraically closed residue field. These will serve as a foundation for studying how the geometry of K3 surfaces interacts with Diophantine questions over local and global fields. For instance, is the Hasse/Brauer-Manin formalism over global fields compatible with (twisted) derived equivalence? See [HVAV11, HVA13, MSTVA14] for concrete applications to rational points problems.In this paper, we first review general properties of derived equivalence over arbitrary base fields. We then offer examples which illuminate some of the challenges in applying derived category techniques. The case of finite and real fields is presented first-here the picture is well developed. Local fields of equicharacteristic zero are also fairly well understood, at least for K3 surfaces with semistable or other mild reduction. The analogous questions in mixed characteristic remain largely open, but comparison with the geometric case suggests a number of avenues for future investigation.
Acknowledgments:We are grateful to Jean-Louis Colliot-Thélène,