At Mon, 8 Oct 2007 22:38:19 -0700,
Philip Du Toit wrote:
> I use gsl_odeiv_control to implement adaptive step size control.
> I understand that this function reduces the step size by a scaling
> factor to ensure that an error tolerance is met.
>
> Is there any checking to be sure that underflow after many rescalings
> does not cause the stepsize to become zero?
Hello,
There isn't any checking for this, but it in practice it hasn't been
reported as a problem. If you have an example program that shows it,
please could you send it as a bug report. Thanks.
> Is there anyway to ensure that the adaptive control never reduces the
> stepsize below a prescribed lower bound regardless of the error?
It should be possible for you to check the value of h returned by
gsl_odeiv_evolve_apply -- it is only a suggested value so it is
possible to override it on subsequent calls.
--
best regards,
Brian Gough
(GSL Maintainer)
Network Theory Ltd,
Publishing the GSL Manual -
http://www.network-theory.co.uk/gsl/manual/
