Feature/default order 4 ark #345
Conversation
example_utilities.hpp requires `stoi` and similar methods. These are declared in `<string>`, but `<string>` is not included. This means this file will break any TU it's included in. Resolves #330 Co-authored-by: Cody Balos <balos1@llnl.gov>
Fixes #335 Co-authored-by: David Gardner <gardner48@llnl.gov>
Fixes #333. Some other typedefs already had specialized (towards idas or cvodes) prefixed names, it won't harm to generalize this. --------- Co-authored-by: Cody Balos <balos1@llnl.gov>
Add scripts for comparing benchmark and example timings against release timings and a notebook using Thicket to visualize data --------- Co-authored-by: David J. Gardner <gardner48@llnl.gov> Co-authored-by: Cody J. Balos <balos1@llnl.gov>
Steven-Roberts
requested review from
balos1,
gardner48,
cswoodward and
drreynolds
as code owners
There was a problem hiding this comment.
The improvements from switching out these default integrators ranges from dramatic to marginal (most problems improve dramatically).
My one question is: how much testing with various RK table options did you do before settling on these specific default tables?
Also, you'll need to edit the CHANGELOG.md file, as well as the "recent changes" section in ARKODE's Introduction.rst file. In those, I recommend adding a sentence to tell users how to switch back to the previous default.
I specifically chose these method due to there optimized properties and so that the number of explicit stages or implicit solves would be the same as the old defaults. Thus, the per step cost will likely not change significantly, but the accuracy or embedding quality will be better. This constraint doesn't leave many options, and I only tested the methods in this PR with the old defaults. If you have candidate methods you think might be better defaults, I can run some comparisons. |
This is fine (at least for now). I'm working on updates to my regression-testing repository for ARKODE, that would hopefully allow us to do more thorough testing of different methods. That said, given the improvement here, I think that we can approve this PR, and if needed we can update the default method again if we find something better before the next release. |
|
This PR will need a corresponding update of https://github.com/sundials-codes/answers to update the output files used in the GitHub actions CI. |
Fix missing soversion on some SUNLinSol and SUNNonlinSol targets --------- Signed-off-by: Julien Schueller <schueller@phimeca.com> Co-authored-by: Cody Balos <balos1@llnl.gov>
1. Adds missing implicit tables to the ARKODE constants docs 2. Makes formatting of constants consistent in ARKODE docs 3. Groups implicit tables by order --------- Co-authored-by: David Gardner <gardner48@llnl.gov>
Update ARKStepSetTableNum to recognize ARK2-3-1-2 as a valid ARK pair.
Only adding the `Sofroniou-Spaletta-5-3-4` method but not changing the defaults like #345 --------- Co-authored-by: Daniel R. Reynolds <reynolds@smu.edu> Co-authored-by: David Gardner <gardner48@llnl.gov>
Only adding the `Sofroniou-Spaletta-5-3-4` method but not changing the defaults like #345 --------- Co-authored-by: Daniel R. Reynolds <reynolds@smu.edu> Co-authored-by: David Gardner <gardner48@llnl.gov>
Only adding the `Sofroniou-Spaletta-5-3-4` method but not changing the defaults like #345 --------- Co-authored-by: Daniel R. Reynolds <reynolds@smu.edu> Co-authored-by: David Gardner <gardner48@llnl.gov>
gardner48
dismissed
drreynolds’s stale review
The merge-base changed after approval.
Changes the default ERK and DIRK integrators to improve robustness of method and embedding