|
From: | Tuomo Keskitalo |
Subject: | Re: [Bug-gsl] Re: [bug #30947] Please, include fixed step size control object for ode suite |
Date: | Sat, 04 Sep 2010 14:01:41 +0300 |
User-agent: | Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.1.11) Gecko/20100713 Thunderbird/3.0.6 |
Hi, On 09/03/2010 12:46 PM, Sergey B Kirpichev wrote:
On 9/3/10, Tuomo Keskitalo<address@hidden> wrote:Please see function gsl_odeiv2_evolve_apply in evolve.c. If step_apply returns with a failure, then a smaller step size is attempted. evolve_apply tries hard to make a step forward, even if it has to use a smaller step size than the user or previous step suggested.Please explain, why this smaller step was attempted. Where the step size is decremented?
In ode-initval2/evolve.c starting from line 186. Failure of a step can be due to e.g. local error of the step being larger than desired error level (too large step size), approaching a singularity or discontinuity, or domain error in the user function. Whatever the reason, evolve_apply tries to advance with a smaller step size.
I really think the best way to add evolution with a constant step size to GSL ode-initval framework is to make a separate evolve_apply routine for it. This new routine should probably return an error code if the local error of the step exceeds the desired error level.
BR, Tuomo -- address@hidden http://iki.fi/tuomo.keskitalo
[Prev in Thread] | Current Thread | [Next in Thread] |