Hi,
I am trying to solve a very large stiff system with the CVODE_BDF() solver. When I apply the Jacobian-free solver by setting linear_solver =: GMRES, it magically solved my problem with about 20 sec, which I believe is too good to be real.
Upon checking the output, I noticed that the results do not satisfy physical understanding. In some smaller system, I can get reasonable output by decreasing the relative tolerance. But in the big system I really want to solve, when I decreasing reltol, I always get the error info below:
It seems adjusting dtmin and dtmax do not make any difference. I wonder what are the possible workarounds to this error information?
Thanks in advance!