An extended multistep Shanks transformation and convergence acceleration algorithm with their convergence and stability analysis (Q2435050)

The authors study the Lotka-Volterra equation of the form \[ \frac{d}{dt}\left(\prod_{i=0}^{q-1} a_{k+i} \right)=\left(\prod_{i=0}^{q-1} a_{k+i} \right) (a_{k+q}-a_{k-1}), \] with the following difference equation as its time discretization \[ \begin{multlined}\left(\prod_{m=0}^{M_k-1} \frac{1+a^{(n+mp+p+1)}_{k-mq-1}}{1+a^{(n+mp+1)}_{k-mq-1}} \right)(1+a_{k-1}^{(n+1)})\prod_{i=0}^{q-1} a_{k+i}^{(n)}\\ =\left(\prod_{m=0}^{M_k} \frac{1+a^{(n+mp+p)}_{k-mq}}{1+a^{(n+mp)}_{k-mq}} \right)(1+a_{k+q}^{(n)})\prod_{i=0}^{q-1} a_{k+i}^{(n)}, \end{multlined}\tag{1} \] where \(p=0,1,\dots,\) and \(M_k=\lfloor k/q \rfloor+1\). The bilinear form of the discrete equation (1) is obtained and then its molecule solution is constructed. A new sequence transformation is obtained and it is shown that there exists a two-dimensional difference equation, which shares the same bilinear form of Eq. (1). A convergence acceleration algorithm for implementing this sequence transformation is found. The article gives examples illustrating the efficiency of the algorithms.