> Not from the high-level interface. > If you use the NewtonSolver class directly, the solve function returns > the number of iterations.
Right. This functionality would be really useful for me to have in Python as well, so I added a suggestion enhancement. Cheers, Nico On Wed, May 15, 2013 at 8:14 PM, Garth N. Wells <[email protected]> wrote: > On 15 May 2013 16:18, Nico Schlömer <[email protected]> wrote: > > Also: Is there a way to extract the number of performed Newton > iterations? > > > > Not from the high-level interface. > > If you use the NewtonSolver class directly, the solve function returns > the number of iterations. > > Garth > > > > --Nico > > > > > > On Wed, May 15, 2013 at 5:02 PM, Nico Schlömer <[email protected] > > > > wrote: > >> > >> Hi, > >> > >> two brief question about the nonlinear solver interface, > >> > >> solve(F == 0, u) > >> > >> * How to specify an intial guess? Is the value used that's found in u? > >> * In the output, > >> > >> Newton iteration 1: r (abs) = 3.402e-01 (tol = 1.000e-10) r (rel) > = > >> 1.040e+00 (tol = 1.000e-09) > >> > >> is it correct that r(rel) = F(x)/F(x0)? > >> > >> --Nico > > > > > > > > _______________________________________________ > > fenics-support mailing list > > [email protected] > > http://fenicsproject.org/mailman/listinfo/fenics-support > > >
_______________________________________________ fenics-support mailing list [email protected] http://fenicsproject.org/mailman/listinfo/fenics-support
