No label defined (Q4249709)

The problem of the solubility of the equation \(a_1p_1+a_2p_2+a_3p_3=b\) (where \(a_1\), \(a_2\) and \(a_3\) are non-zero integers and the \(p\)'s are primes) and of an upper bound for the smallest solution, when it exists, has been raised by A. Baker and qualitatively settled by \textit{M.-C. Liu} [Math. Z. 194, 573-583 (1987; Zbl 0602.10040)] and by \textit{M.-C. Liu} and \textit{K. M. Tsang} [Théorie des Nombres, Québec 1987, 595-624 (1989; Zbl 0682.10043)]. The authors consider the case where the primes are required to lie in given arithmetic progressions modulo a fixed integer \(k\), provided that the parameters satisfy some natural conditions. Set \(K=\max\{2,\) \(|a_1|\), \(|a_2|\), \(|a_3|\), \(k\}\). The main result is that if all the \(a\)'s are positive, then the given equation is soluble as soon as \(b>K^{\nu}\), where \(\nu\) is a computable absolute constant, whereas if the \(a\)'s are not all of the same sign, the equation has a solution satisfying \(\max\{p_1\), \(p_2\), \(p_3\}\leq 3|b|+K^B\), where \(B\) is another computable absolute constant. The proof is an application of the circle method and follows the one in the paper by Liu and Tsang quoted above, the main new difficulty being the treatment of the exponential sums restricted to arithmetic progressions, which takes the bulk of the paper. For the entire collection see [Zbl 0910.00037].