From ae8739d0671acf34b4c7e317b02a21e4544bd972 Mon Sep 17 00:00:00 2001 From: Wolfgang Bangerth Date: Wed, 14 Aug 2024 10:08:11 -0600 Subject: [PATCH] Add a note to step-86. --- examples/step-86/doc/results.dox | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/examples/step-86/doc/results.dox b/examples/step-86/doc/results.dox index d578a0c659..b7b57f6f00 100644 --- a/examples/step-86/doc/results.dox +++ b/examples/step-86/doc/results.dox @@ -234,7 +234,14 @@ ourselves to blame for that by setting @code set match final time = false @endcode -in the input file. +in the input file. If hitting the end time exactly is important to us, +setting the flag to `true` resolves this issue. + +We can even reason why PETSc eventually chooses a time step of around +0.025: The boundary values undergo a complete cosine cycle within 0.5 +time units; we should expect that it takes around ten or twenty time +steps to resolve each period of a cycle to reasonable accuracy, and +this leads to the time step choice PETSc finds. Not all combinations of methods, time step adaptation algorithms, and other parameters are valid, but the main messages from the experiment -- 2.39.5