Closed K20shores closed 1 week ago
How do you decide the tolerance values? It is interesting to see they vary between each solver type and rate constant type.
it was super duper scientific. I set a value of 1e-8
and when that failed for a particular test I subtracted one from the exponent until it passed.
A more intelligent way would be to play with the solver tolerances to choose values that minimize the error
How do you decide the tolerance values? It is interesting to see they vary between each solver type and rate constant type.
it was super duper scientific. I set a value of
1e-8
and when that failed for a particular test I subtracted one from the exponent until it passed.A more intelligent way would be to play with the solver tolerances to choose values that minimize the error
Haha I see. Yeah that makes sense but would require a good amount of work to investigate that. I guess we don't need to worry about it for now
All modified and coverable lines are covered by tests :white_check_mark:
Please upload report for BASE (
main@854ca2d
). Learn more about missing BASE report.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
Closes #499
Backward euler passes all tests, but the robertson tests (the solver fails to converge and seems to get stuck, will be investigated by #498). I don't know how to test backward euler against HIRES, Oregonator, E5, but hopefully that can be fixed with #575