Cubic Diophantine inequalities (Q5936146)

In this paper, it is proved that for any real numbers \(\lambda_1\), \(\lambda_2,\ldots,\lambda_7\) with \(\lambda_i\geq 1\) \((1\leq i\leq 7)\), the Diophantine inequality \[ |\lambda_1x_1^3+\lambda_2x_2^3+\cdots+\lambda_7x_7^3|<1 \] has a non-trivial solution in integers \(x_1,\ldots,x_7\) satisfying \[ 0<\sum_{i=1}^7\lambda_i|x_i|^3\ll (\lambda_1\lambda_2\ldots\lambda_7)^{89814}. \] With respect to the latter bound for the solution, this result refines the consequence stemming immediately from the work of \textit{R. C. Baker, J. Brüdern} and \textit{T. Wooley} [Cubic Diophantine inequalities, Mathematika 42, 264-277 (1995; Zbl 0837.11016)]. The argument of this paper is mostly constructed by a variant of the Hardy-Littlewood circle method, which is familiar in this area. The crucial estimate for minor arc integral is established by skillful subdivision of the minor arcs and various mean value estimates for smooth Weyl sums obtained in recent developments of Vaughan's iterative method. Amongst others, ``breaking classical convexity'' device of \textit{T. Wooley }[Breaking classical convexity in Waring's problem: Sums of cubes and quasi-diagonal behaviour, Invent. Math. 122, 421-451 (1995; Zbl 0851.11055)] plays an important role. Now suppose that \(\lambda_1\) is the largest amongst \(\lambda_i\)'s without loss of generality, and write \(\Lambda=\prod_{i=1}^7\lambda_i\). Then unless \(\lambda_i/\lambda_1=a_i/b_i\) with natural numbers \(a_i\) and \(b_i\) satisfying \(b_i<\Lambda^{365.4}\) for every \(i\) with \(2\leq i\leq 7\), the aforementioned theorem is established indeed with the exponent 1613 in place of 89814 by the argument outlined above. In the case excluded here, the theorem follows directly from the author's recent result on diagonal cubic equations with seven variables [Diagonal cubic equations, Acta Arith. 81, 199-227 (1997; Zbl 0888.11036)].