1: \begin{abstract}

2: {Pseudo-arclength continuation is a well-established method for

3: generating a numerical curve approximating the solution of an underdetermined system of nonlinear equations. It is an inherently sequential predictor-corrector method 

4: in which new approximate solutions are extrapolated from previously 

5: converged results and then iteratively refined.

6: Convergence of the iterative corrections is guaranteed only

7: for sufficiently small prediction steps.

8: In high-dimensional systems, corrector steps are extremely

9: costly to compute and the prediction step-length 

10: must be adapted carefully to avoid failed steps or 

11: unnecessarily slow progress.

12: We describe a parallel method for adapting the step-length employing several

13: predictor-corrector sequences of different step lengths computed 

14: concurrently. In addition, the algorithm permits

15: intermediate results of unconverged correction sequences to seed new 

16: predictions. This strategy results in an aggressive optimization of

17: the step length at the cost of redundancy in the concurrent computation.

18: We present two examples of convoluted solution curves of 

19: high-dimensional systems showing that speed-up by a factor of two

20: can be attained on a multi-core CPU while a factor of three is attainable on

21: a small cluster.}

22: \end{abstract}

23: