The method of controlling chaos using small parameter perturbations wh
ich was first proposed by Ott, Grebogi and Yorke (Phys. Rev. Lett., 64
, 1196-1199) suffers from two problems: the 'basin of attraction' for
the fixed point which will be stabilized maybe small if the maximum pe
rmissible parameter perturbation is small, resulting in long chaotic t
ransients before control is achieved, and noise can result in control
being lost. We address both these problems by constructing an extended
basin of attraction in which several iterations using the maximum par
ameter perturbation may be made before attempting to place an iterate
on the stable manifold of the fixed point using an appropriately chose
n perturbation of the parameter. This has the effect of reducing trans
ient times by a factor of approximately \lambda(u)\/(\lambda(u)\ -1) w
here lambda(u) is the unstable eigenvalue of the saddle fixed point, a
s well as reducing the effects of noise. The method is also applied to
other related control methods and it is shown that the same extended
basin of attraction is obtained. The method is illustrated with a nume
rical example. (C) 1997 Elsevier Science Ltd.