On Wed, Jul 20, 2016 at 5:45 PM, Raymond Smith <smit...@mit.edu> wrote: > That makes sense. So, here > https://gist.github.com/raybsmith/b0b6ee7c90efdcc35d6a0658319f1a01 > I've changed it so that the error is calculated as > sum( (\phi-phi*)^2 * mesh.cellVolumes ) > and depending on the value I choose for the ratio of dx_{i+1} / dx_i, I get > convergence for exponential spacing ranging from 2nd order (at that ratio = > 1) and decreasing order as that ratio decreases from unity.
Thanks for that. I'm not entirely sure, but the integral as written on line 44, https://gist.github.com/raybsmith/b0b6ee7c90efdcc35d6a0658319f1a01#file-fipy_accuracy-py-L44 may still only be first order accurate for non-uniform grids. I'll try and investigate this though and see if it matters. -- Daniel Wheeler _______________________________________________ fipy mailing list fipy@nist.gov http://www.ctcms.nist.gov/fipy [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
