Fix broken tests #1070
Fix broken tests #1070
Conversation
| 0.1 + hill(X,5.0,K,3), 0 --> X | ||
| 1.0, X --> 0 | ||
| end | ||
| @unpack X, v, K, n, d = bistable_switch | ||
| @unpack X, K = bistable_switch | ||
| u0_guess = [X => 1.0] | ||
| p_start = [v => 5.0, K => 2.5, n => 3, d => 1.0] | ||
| p_start = [K => 2.5] |
There was a problem hiding this comment.
Do multiple parameters that are symbolic no longer work with the extension? Why do you need to change the parameters to hard-coded values?
There was a problem hiding this comment.
No, still work (see other tests).
The problem is that I wrote a test which evaluates some internal MTK functions and check that they return the same stuff. However, this kind of depended on ordering, and the test fails due to the parameter ordering not being right. I changed the this test so that there was only a single parameter (so that parameter ordering is not really a worry).
Probably the test should just be removed instead, but for now I just tried to make a minimal change.
|
There is still some weird stuff with SI. I have been looking at it for a while and untangled it a bit, but still not fully sure how to finally fix it. SI being broken is probably what holds up the docs as well. |
|
Sounds good. Feel free to merge this then when you want. |
|
@TorkelE I'm going to merge this as I don't want to hold off much longer on getting PRs merged and making a release with the array fixes. |
Fixes the tests that have been broken by recent updates.
The long running for identifiability analysis has to do with us using array variables for representing unknown quantities, and there seems to be a problem with this (likely related to recent updates). Might be easiest to simple rewrite code to not use that, will have a look.