Wolfgang Bangerth
Wolfgang Bangerth
*Background:* We are trying to convert a number of places in the ASPECT project (https://github.com/geodynamics/aspect/) to use KINSOL instead of hand-rolled nonlinear solvers. ASPECT is a code for large-scale simulation...
I've been reading my way through the documentation of the pipeline classes and noticed two things: * All of the examples shown always use serial stages. It would be nice...
This addresses the second issue of #17180. This one should be uncontroversial: Instead of returning a bool via a reference argument, it returns it by value.
@stefanozampini I'm (finally!) putting the finishing touches on step-86 (see #15540) and I'm coming by the `PETScWrappers::TimeStepper::distribute` callback. I know why you called it that way in #15066: Because all...
It would be nice if one could use KINSOL for scalar root finding problems `f(x)=0` rather than for vector-valued ones where one then has to use a vector size of...
As pointed out in https://github.com/geodynamics/aspect/issues/5727, we run into an error when trying to output higher order cells in 1d because `ReferenceCell::vtk_lexicographic_to_node_index()` is not implemented for the 1d case. I *suspect*...
Do this for every release, just before we make the release, then move the release milestone to the following release.
I came across this paper https://dl.acm.org/doi/full/10.1145/3503925: "On Memory Traffic and Optimisations for Low-order Finite Element Assembly Algorithms on Multi-core CPUs". @kronbichler and @peterrum might be interested in it as well....
All of the C++ container classes seem to have received `empty()` functions that make writing `x.size() == 0` simpler. We might want to think of adding those to our classes...
On my shiny new platform on which I've installed all prereqs with candi, the ARPACK tests fail: ``` > ./step-36_parpack.debug ./step-36_parpack.debug: error while loading shared libraries: libarpack.so.2: cannot open shared...