Wolfgang Bangerth

step-86: Use reasonable time stepping for a heat equation solver.

31

step-26 uses a home-grown implicit Euler time stepper. We should really show how one can do that better, using the PETSc TS or SUNDIALS ARKode wrappers. The current state of...

LinearAlgebra::{E,T}petraWrappers::Vector::compress() doesn't do anything.

5

At the time of writing, the `LinearAlgebra::EpetraWrappers::Vector` class does not implement its own `compress()` function, but because it derives from `VectorSpaceVector` inherits that base class's implementation of `compress()` -- which...

Check vector concepts for all of our solvers.

1

#15637 creates a concept for what constitutes a vector-space-vector. We should annotate all of our solver classes with this requirement. The files in question are specifically these: ``` > egrep...

Clustering cells by similarity.

2

@GrahamBenHarper gave a talk in our seminar today that made me think about how we can do our cell similarity things better. What he is doing is this, in essence:...

A TaskResult class.

1

Here's something I've wanted for a long time already, and current discussions in #16182 suggest that this is the right time: #16182 adds a `Lazy` class that allows us to...

Deprecate the contents of the LocalIntegrators namespace.

3

I think it's time. This patch deprecates all classes in namespace `LocalIntegrators`. These implement concrete integration routines. It also deprecates step-16b, one of two programs that use them. When we...

Epetra_CrsGraph: Overflow issues when trying to add too many nonzeros.

5

This is part of the explorations in https://github.com/dealii/dealii/issues/13428 where we're tracking down what happens if one tries to allocate sparsity patterns with more than 2B entries on one process. This...

Improve matrix-free performance marginally.

1

@tjhei This is one for you, pointed out to me by @marcfehling . In `stokes_matrix_free.cc`, we apply the block preconditioner in the following way (abridged): ``` void BlockSchurGMGPreconditioner:: vmult (dealii::LinearAlgebra::distributed::BlockVector...

Compute a better alpha factor in the implementation of the SPD-ness of the Newton method.

6

@YiminJin pointed out in an email to @MFraters that we have a mistake in the paper that describes the Newton method. Specifically, in the derivation of eq. (18), we introduce...