The exponent of convergence for 2-dimensional Jacobi-Perron type algorithms (Q5939038)

There are several algorithms to generalize the continued fraction algorithm to higher, in special, \(2\)-dimensions. They produce a sequence of rational numbers approximating several given numbers simultaneously. The Jacobi-Perron algorithm is one of the most famous ones. It is practical to study the efficiency of algorithms. This paper proposes a systematic method for several good algorithms. Let \(X=\{(x,y): 1\geq x\geq y\geq 0\}\). Define the exponent of convergence for an algorithm at \((x,y)\) by \(\delta(x,y)=\sup\{\delta: \exists n(x,y)\in\mathbb Z\) such that \(\forall n\geq n(x,y)\), \(|x-p_n/q_n|\leq 1/q_n^\delta\), \(|y-p_n/q_n|\leq 1/q_n^\delta, \}\). Then, for the Jacobi-Perron algorithm, Podsypanin's modified Jacobi-Perron algorithm, Güting's algorithm and the acceleration of Brun's algorithm, there is a constant \(d>1\) such that \(\delta(x,y)=d\) for almost every point \((x,y)\in X\).