On Fri, 16 Apr 2010, Andrea Hawkins wrote: > I believe that the exact_error_estimator.estimate_error function has a > bug in it, or at least I don't understand the current idea of the > function. It has an outer loop for the variables and an inner loop for > the elements. For each variable it seems to do as expected. However, > the error_per_cell is essentially getting overwritten for each new > variable.
Yes - it looks like I introduced a bug during an old refactoring of this method, then never noticed because I'd only been using this estimator on single-variable benchmarks. Thanks for catching it! It should now be fixed (= -> +=, plus a slightly more complicated fix for the computation of parent cell errors). Please check it out? Thanks, --- Roy ------------------------------------------------------------------------------ Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev _______________________________________________ Libmesh-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/libmesh-users
