Re: [Bug-gsl] ode-initval/bsimp is always high order

[Brian Gough]

At Mon, 20 Sep 2010 13:18:28 -0700 (PDT),
Tom Quinn wrote:
> 
> The bsimp_alloc() routine uses a very small EPSILON to set the order of 
> the integrator, and there seems to be no interface to change this.
> 
> The result is that even for small steps and large error tolerances, a 
> minimum of 140 function evaluations are taken per timestep.  For my 
> application, this is more than an order of magnitude more than are 
> necessary.

Thanks for the bug report.  I agree it's a problem.
I've logged it in the bug tracker
at https://savannah.gnu.org/bugs/index.php?31109

I'm not sure what is the best way to make configurable in the current
framework. To work around it you could copy bsimp.c into your own
program and modify it to create an integrator with a different name.

-- 
Brian Gough