You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We need to stop assuming ordering of unknowns/ps in tests. I tried to eliminate this in the main library long ago, but there are apparently still some tests that have issues in this regard.
We should test that ps and us are subsets of system ps and us in general, and not test set equality (since MTK may now add extra stuff). When using structural_simplify we need to test vs. us and observed since us can be moved to observed.
Some tests around conservation laws may be incorrect, see Aayush's Slack comment that:
https://github.com/SciML/Catalyst.jl/blob/master/test/upstream/mtk_structure_indexing.jl#L405
This test is incorrect. prob9 has the initial conditions Gamma[1] => 10.0, X1 => 10.0, X2 => 2.0 which are inconsistent since X1 + X2 ~ Gamma[1] (edited)
[9:29](https://julialang.slack.com/archives/C0894545874/p1745501399505049)
The same holds for Gamma[2] => 20.0, Y1 => 3.0, Y2 => 4.0
The text was updated successfully, but these errors were encountered:
The output thing is something we need to deal with more generally whenever structural_simplify has been involved. However, there are some implications for plotting I'd like to have ironed out first ideally, and I'd like to see what actually is going to happen in MTKv10 as well, as it feels like that is changing things up drastically.
ps
andus
are subsets of systemps
andus
in general, and not test set equality (since MTK may now add extra stuff). When usingstructural_simplify
we need to test vs.us
andobserved
sinceus
can be moved to observed.The text was updated successfully, but these errors were encountered: