Diophantine approximation with one prime and three squares of primes (Q5894266): Difference between revisions

The main theorem of this paper asserts that if \(\lambda_1\), \(\lambda_2\), \(\lambda_3\), \(\lambda_4\) are nonzero real numbers, not all of the same sign, and at least one of the ratios \(\lambda_1/\lambda_j\) (\(j=2\), 3, 4) is irrational, then for any real numbers \(\eta\) and \(\varepsilon\) with \(\varepsilon>0\), there exist infinitely many ordered quadruples of the primes \((p_1,p_2,p_3,p_4)\) such that \[ \bigl|\lambda_1p_1+\lambda_2p_2^2+\lambda_3p_3^2+\lambda_4p_4^2+\eta\bigr| <\bigl(\max_{1\leq j\leq 4}p_j\bigr)^{-1/28+\varepsilon}. \] The authors point out that this conclusion is indeed valid unless all the ratios \(\lambda_j/\lambda_k\) \((1\leq j<k\leq 4)\) are rational, since it follows from the work of \textit{G. Harman} [Mathematika 51, 83--96 (2004; Zbl 1107.11043)] in the remaining cases where at least one of the ratios \(\lambda_j/\lambda_k\) (\(2\leq j<k\leq 4\)) is irrational. The proof is based on the Davenport-Heilbronn circle method.