|
From: | Michael Koehn |
Subject: | [Help-gsl] question on parallelizing rkf45 |
Date: | Tue, 30 Mar 2010 18:24:11 +0200 |
Dear GSL users,I came across the GSL ODE integrators and am very happy about that. However there is a question I can not answer, and I would be very thankful if you could quickly help me out. I am doing a Method of Lines analysis of a 2+1-dimensional PDE. I put all the ODEs for each grid point into the solution vector. I would really like to paralellize the integration of these 160.000 ODEs onto different cores. However, I am unsure on how to implement this using OpenMP. I could take for instance ten equally sized bunches of ODEs and give one of these to each core. However the GSL routines update the solution vector in place and the problem is that the function "func" includes the finite difference approximations for the spatial derivatives. Thus, as far as I can tell, everything will probably get messed up, since it can occur that the finite differencing refers to neighboring grid points that might be already updated by some parallel process or not.
Do you have any idea on this? Thank you very much for your time! Best regards, Michael
[Prev in Thread] | Current Thread | [Next in Thread] |