HarryKane
05-27-2008, 12:04 PM
Hello,
I have to integrate a system of second order ODEs over a loooong interval, but I could not integrate the full intervall at once, because of the limit MAXSTP in odeint.h.
I allready had this problem with the old version, and so I choped the interval in serveral substeps. As I read the 3rd ed. I was happy to hear that there is now a possibility to get information about the steps (or equal spaced interpolated points) taken in this small intervall, so that nobody has to subdivide the full intervall. I hoped, that I would be now able to integrate the long intervall, but I got disappointed :(.
I understand that there should to be a limit, that ends an intergration, if something goes wrong (a singularity or something).
I thought about to reset the counter nstep after a step was successfull so that it the intergration only breaks after 50k bad steps ... or is this a stupid idea?
Cheers,
HarryKane
I have to integrate a system of second order ODEs over a loooong interval, but I could not integrate the full intervall at once, because of the limit MAXSTP in odeint.h.
I allready had this problem with the old version, and so I choped the interval in serveral substeps. As I read the 3rd ed. I was happy to hear that there is now a possibility to get information about the steps (or equal spaced interpolated points) taken in this small intervall, so that nobody has to subdivide the full intervall. I hoped, that I would be now able to integrate the long intervall, but I got disappointed :(.
I understand that there should to be a limit, that ends an intergration, if something goes wrong (a singularity or something).
I thought about to reset the counter nstep after a step was successfull so that it the intergration only breaks after 50k bad steps ... or is this a stupid idea?
Cheers,
HarryKane