On nonary cubic forms.

From MaRDI portal





It was shown by the reviewer [Proc. Lond. Math. Soc., III. Ser. 47, 225-- 257 (1983; Zbl 0494.10012)] that any nonsingular cubic form in 10 or more variables, defined over \(\mathbb Q\), has a nontrivial rational zero. This is now sharpened to cover forms in 9 variables, providing that they satisfy all appropriate local conditions. (For 10 or more variables these conditions hold automatically.) This monumental work provides perhaps the most significant example of a Hasse principle after the Hasse-Minkowski Theorem itself. The method builds on that employed by the reviewer, using the Hardy- Littlewood method with the Kloosterman refinement, and estimating the cubic exponential sums via the Poisson summation formula, together with Deligne's bounds for complete sums. Very roughly speaking, if \(f(\mathfrak x)\) is the cubic form in question, the reviewer's method would estimate the number of integer solutions of \(f(\mathfrak x)=0\) in a box of side \(P\), as \(cP^{n-3}+O(P^{(3n-3)/4+\varepsilon})\). The present work removes the factor \(P^{\varepsilon}\) in the error term, and indeed saves an additional power of a logarithm. The underlying idea here originates in the author's work [Mathematika 11, 39--49 (1964; Zbl 0123.25802)], and was developed further by him in his investigations on sums of two squares and three cubes [J. Reine Angew. Math. 328, 161--207 (1981; Zbl 0463.10036)]. In order to carry through his programme Hooley has to be much more careful than in the reviewer's work. Four points deserve special mention. Firstly, in the Kloosterman refinement it is necessary to average both by varying the endpoints between arcs of the Farey dissection, and by varying the order of the dissection. Otherwise a crucial logarithm would be lost. Secondly, the integrals arising from the Poisson summation formula are estimated more exactly. This is facilitated by using the fact that the Hessian of the cubic form does not vanish identically on the corresponding cubic hypersurface. (For the reviewer's work a slightly easier, weaker, result sufficed.) Thirdly, the complete exponential sums that occur are estimated individually, not just for prime moduli, as was done by the reviewer, but also for prime square moduli. A general bound of this type has been given by the reviewer [Can. Math. Bull. 28, 394--396 (1985; Zbl 0529.10036)]. Fourthly, in both works one needs to consider integer poi We construct a basis \(\alpha_ 1,\ldots,\alpha_{24}\) of \(\Lambda\) consisting of elements of \(R\), such that: \[ \forall (\alpha \in R)\quad \alpha =\sum n_ i\alpha_ i\quad \vee \quad (\forall n_ i\geq 0)\quad \bigvee \quad (\forall n_ i\leq 0). \] Furthermore it is shown that this special basis is not unique.




Cited in
(41)








This page was built for publication: On nonary cubic forms.

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3781824)