Hello Bruno, Apologies for the delayed response.
On Tuesday, November 1, 2016 at 6:06:16 PM UTC+5:30, Bruno Turcksin wrote: > Have you tried to increase the number of Newton iterations? The Newton > solver is pretty basic (it doesn't do any line search) so it might be > the reason it doesn't work. If you send me a short program that > reproduces the problem, I can try to make it work. However, I am > pretty busy in November, so it may take a month or so before I can > work on it. > Looking at the error message I get, the non convergence is probably not due to the Newton iterations. I am using an iterative solver(Iterative Inverse class of dealii) for computing the ((M-tau*J)^-1)*M*y term (the term dealii probably requires for Newton iterations). This is in contrast to step-52 (which uses UMFPACK) and the non convergence is in this iterative solution step. In fact, the solution converges(and matches with an equivalent ODE45 implementation) for many time steps and no convergence occurs when a steep temporal change in the solution(characteristic of the stiff system) is encountered. Any advice on how to make it converge? Sincere thanks for your offer to help, but being a beginner, I would like to learn it myself rather than outsource my problem. A very popular package for ODE solving is sundials > http://computation.llnl.gov/projects/sundials. That's what I would > use. > > As I had some previous exposure to the GSL ODE solver, I tried implementing that but sadly ran into another problem. The GSL solver requires a vectorization of the Jacobian of the RHS of the ODE system. The Jacobian in our case, if I am not wrong, will be (M^-1)*(df/du), the product of a matrix inverse and a matrix which I am unsure how to calculate. Any advice here will also be appreciated. Thank you for all the help so far! Regards, Vaibhav -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to dealii+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
