make PetscInfo() variadic

SETERRQ[1-9]+ begone

Merge branch 'jacobf/2021-12-09/attribute-format-mat' into 'main'

Feature: Attribute Format - Mat

See merge request petsc/petsc!4634

remove trailing newline from SETERRQ

Merge branch 'tisaac/feature-dubiner-basis' into 'master'

PetscDT: PKD (Dubiner) basis

See merge request petsc/petsc!2691

Add PetscDTPDKEvalJet(), PetscDTJacobiEvalJet(), PetscDTIndexToGradedOrder(), and PetscDTGradedOrderToIndex()

The Proriol-Koordwinder-Dubiner basis (PKD) is orthonormal on the
reference simplex, so

Add PetscDTPDKEvalJet(), PetscDTJacobiEvalJet(), PetscDTIndexToGradedOrder(), and PetscDTGradedOrderToIndex()

The Proriol-Koordwinder-Dubiner basis (PKD) is orthonormal on the
reference simplex, so we should make it available as a stable basis of
evaluation.

The recursive definition allows for a straightforward and stable
calculation of all derivatives, not just first and second.

`src/dm/dt/tests/ex9.c` tests that the basis is orthonormal
and compares the derivatives against the derivatives computed by
tensor-product Legendre polynomials.

I thought about how to make an interface that would handle that,
and I decided returning the jet in the same kind of graded
representation that makes sense for multivariate polynomial spaces was a
good choice.

To facilitate this, I also added indexing functions for figuring out
where a tuple (either of exponents in a monomial or orders in a partial
derivatives) is in this ordering.

Along the way I made PetscDTStroudConicalQuadrature() handle arbitrary
dimensions instead of just 1--3.

We have a variety of dimensional orderings in PetscDT, related to how we
order points of evaluation, components of the jet, and basis functions.

I thought that:
- downstream users concerned with performance might prefer points of
evaluation as the fastest varying index to facilitate vectorization
of numerical integration, and
- having basis functions as one of the outer dimensions (first or last)
would simplify computing the matrix-matrix product of
a [basis function x evaluation point] matrix with a Vandermonde
inverse matrix

So I ordered the output dimensions of PetscDTPKDEvalJet() as
[basis function, jet component (derivative), evaluation point]. I am
seeking comment as to whether this is good, and whether it might be a
good convention to adopt elsewhere in PetscDT.

undo extern

Fix complex type issue in dm/dt/tests/ex9.c

squashme: fix scaling in PetscDTStroudConicalQuadrature()

squashme: loosen tolerance for single precision

squashme: cast to double before printing

squashme: remove unused static functions

show more ...

Merge branch 'jed/promote-examples-tests-tutorials' [petsc/petsc!2610]

* jed/promote-examples-tests-tutorials:
Promote examples/{tests,tutorials}/ to {tests,tutorials}/

Promote examples/{tests,tutorials}/ to {tests,tutorials}/

This shortens paths and improves consistency between test target names
and paths to the source and output files. Most of the work was
compl

Promote examples/{tests,tutorials}/ to {tests,tutorials}/

This shortens paths and improves consistency between test target names
and paths to the source and output files. Most of the work was
completed by this script, followed by mild cleanup of nonconforming
cases.

for makefile in `git ls-files 'src/*makefile'`; do
if rg -q 'DIRS.*\bexamples\b' $makefile; then
base=$(dirname $makefile)
dirs=$(cd $base/examples && ls -d tests tutorials 2>/dev/null | xargs echo)
perl -pi -e "s#^(DIRS.*)\bexamples\b#\1${dirs}#" $makefile
git rm $base/examples/makefile
for t in $dirs; do
git mv $base/examples/$t $base/
perl -pi -e "s#^(LOCDIR[[:space:]]*=).*#\1 $base/$t/#" $base/$t
done
fi
done

git grep -l -E -z 'examples/(tutorials|tests)' | xargs -0 perl -pi -e 's#examples/(tutorials|tests)#\1#g'
git checkout @ \
src/docs/website/documentation/changes/ \
src/benchmarks/results/

show more ...