Solving exponential diophantine equations using lattice basis reduction algorithms (Q579297): Difference between revisions

In this paper it are solved completely: the diophantine inequality \[ | p_ 1^{x_ 1}-p_ 2^{x_ 2}| \quad <\quad \min (p_ 1^{x_ 1},p_ 2^{x_ 2})^{\vartheta} \] under the assumptions: (a) \(p_ 1\) and \(p_ 2\) are primes such that \(p_ 1<p_ 2<200\) and \(x_ 1,x_ 2\in {\mathbb{Z}}\), \(x_ 1\geq 2\), \(x_ 2\geq 2\) and either \(\vartheta =\) or (1) \(\vartheta =9/10\) and \(\min (p_ 1^{x_ 1},p_ 2^{x_ 2})>10^{15};\) (b) \(2\leq p_ 1<p_ 2\leq 50\) are non-powers and (1) holds; the diophantine inequality \(0<x-y<y^{1/2}\), where \(x,y\in S=\{2^{x_ 1}... 13^{x_ 6}:\) \(x_ i\geq 0\), \(x_ i\in {\mathbb{Z}}\), \(i=1,...,6\};\) the diophantine equation \(x\pm y=wz\), where \(x=p_ 0^{x_ 0}\), \(y=p_ 1^{x_ 1}\), \(z=p^ u\), \((p,p_ 0,p_ 1)=(2,3,5)\), (3,2,5) or (5,2,3), \(x_ 0,x_ 1,u\) are non-negative integers, \(w\in {\mathbb{Z}}\), \(| w| \leq 10^ 6\), and p \(\nmid w;\) and the diophantine equation \(x+y=z\), where x,y,z\(\in S\), with \((x,y)=1\) and \(x\leq y.\) One can divide the method of the proofs into three steps. In the first step explicit upper bounds are proved for the solutions using \textit{M. Waldschmidt}'s [Acta Arith. 37, 257-283 (1980; Zbl 0357.10017)] lower bound for linear forms in logarithms of algebraic numbers, and \textit{A. J. van der Poorten}'s [Transcend. Theory, Proc. Conf., Cambridge 1976, 29-57 (1977; Zbl 0367.10034)] analogous p-adic theorem. In the second step these large upper bounds are reduced considerably using computational diophantine approximation techniques. The tools here are continued fractions and the basis reduction algorithm of \textit{A. K. Lenstra, H. W. Lenstra} jun. and \textit{L. Lovász} [Math. Ann. 261, 515- 534 (1982; Zbl 0488.12001)]. In the final step the solutions are found by enumerating the remaining cases.