The Hasse principle and the Brauer-Manin obstruction for curves (Q706180)

Let \(K\) be a number field, \(\mathbb A_K\) the adèles of \(K\) and \({\mathcal X}\) a smooth projective variety over \(K\) violating the Hasse Principle (HP), i.e. \({\mathcal X}(K)=\emptyset\) but \({\mathcal X(\mathbb A_K)}\neq\emptyset\). Global reciprocity applied to the Brauer-Grothendieck group \(\text{Br}({\mathcal X})\) defines a certain subset \({\mathcal X}(\mathbb A_K)^{\text{Br}}\subset{\mathcal X}(\mathbb A_K)\) which contains the diagonal image of \({\mathcal X}(K)\). When \({\mathcal X}(\mathbb A_K)^{\text{Br}}=\emptyset\) we say the violation of the HP is due to the Brauer-Manin obstruction (BM). For a surface \({\mathcal X}\), BM is not the only obstruction to HP, but if \({\mathcal X}\) is a curve it is an open question whether BM is the only obstruction to HP. For a smooth proper curve \({\mathcal C}\) with finite Tate-Shafarevich group Ш\(({\mathcal J})\) (where \({\mathcal J}\) is the Jacobian of \({\mathcal C}\)), BM is the only obstruction to HP if: \({\mathcal C}\) has genus 1 (Lemma 1); \({\mathcal C}\) has any genus and it has no \(K\)-rational divisor class of degree 1 (Lemma 2); \({\mathcal C}\) has any genus, \({\mathcal J}\) is finite and \({\mathcal C}\) has a rational divisor class of degree 1 (Corollary 1). The aim of the paper is methodological: the author develops general techniques which allow computation of many violations of HP due to BM, with a style of proof more widely applicable than the isolated examples previously known. Since the techniques are amenable to generating large numbers of examples, they may be a first step towards gaining some statistical insight as to the rarity of any violations of HP that might not be due to BM. The author concludes that the examples do not provide clear evidence on whether BM is the only obstruction to HP for smooth projective curves. They show that ``flat'' (using finite fields to obtain congruence conditions) and ``deep'' (working with local power series) information can be usefully combined (Example 7), but even this example can be resolved with purely flat information. He finds mild evidence to suggest, at least for genus 2 and rank 1, that flat information of the BM obstruction may be the only obstruction to HP for smooth projective curves.