Add no parameter tests #241
Conversation
test/steady_state_problems.jl
Outdated
| for factor in [1e0, 1e1, 1e2] | ||
| u0 = factor*rand(length(no_param_network.states)) | ||
| sol_ss = solve(SteadyStateProblem(no_param_network,u0),SSRootfind()) | ||
| @test abs.(sol_ss.u[1]-0.6/3.2) < 10000000*eps() |
There was a problem hiding this comment.
specify a tolerance for the rootfind and make sure it is roughly achieved?
There was a problem hiding this comment.
That's a good point, didn't even think of something like that. Will modify accordingly.
test/solve_ODEs.jl
Outdated
| u0 = factor*rand(length(no_param_network.states)) | ||
| prob = ODEProblem(no_param_network,u0,(0.,1000.)) | ||
| sol = solve(prob,Rosenbrock23()) | ||
| @test abs.(sol.u[end][1] - 1.5/2) < 1000000*eps() |
There was a problem hiding this comment.
It'd just say something like <1e-8
|
@TorkelE Apparently the signature for I think it is from this PR to ModelingToolkit: |
|
@TorkelE You need to merge master into your branch I think. |
|
Good point |
|
So I think what's going wrong is that Travis doesn't use a version of MTK that includes SciML/ModelingToolkit.jl#543 (but by no means to knowledgable about how this works, so might be more complicated). I do not get this error on my machine, but when I roll back my local MTK version to pre SciML/ModelingToolkit.jl#543 I get the same error. Also, Travis seems to run MTK 3.14.2, and checking that, I do not think it includes SciML/ModelingToolkit.jl#543. If that's the case, we can just wait until the next release of MTK and then this should be fine. |
|
@ChrisRackauckas ok to make a new MT release? (Not sure the state of master there currently.) |
|
It's in the process now. |
|
@TorkelE Looks like it is having SS issues again. I'll try restarting. |
|
@TorkelE Finally passing... Good to merge? |
Adds a simple test each for ODE, SDE, Jump, and SteadyState Problems, testing that they work fro systems without parameters.