Mathematica source code missing in KomaMRI repo

[beorostica]

We are performing simulation tests with results obtained from Mathematica simulations. It is a good idea to have the Mathematica source code stored in the KomaMRI repo.

[cncastillo]

This should be replaced by DiffEq.jl.

[cncastillo]

@Stockless can you check if this has been already been solved? I believe it has, as we are not importing any results to test against anymore (besides JEMRIS for imaging results).

[Stockless]

In KomaMRICore/test/runtest.jl:270 says that there is a Mathematica-simulated Bloch equation result being used in that test. What should I do about it?

[cncastillo]

Can you link the code?

[cncastillo]

Ok, just looked at it. These hardcoded results:

    #Mathematica-simulated Bloch equation result
    res1 = [0.153592+0.46505im,
            0.208571+0.437734im,
            0.259184+0.40408im,
            0.304722+0.364744im,
            0.344571+0.320455im,
            0.378217+0.272008im]
    res2 = [-0.0153894+0.142582im,
            0.00257641+0.14196im,
            0.020146+0.13912im,
            0.037051+0.134149im,
            0.0530392+0.12717im,
            0.0678774+0.11833im]

Should be replaced by something similar to the tests in Motion, which uses OrdinaryDiffEqTsit5.jl. This would be easier if #478 was solved, but it should be generated using DiffEq.

I found a slight problem @pvillacorta , can you check if that is the case elsewhere?

• GPUFunctions tests: Line 416 shouldn't be "commented" (@test true)

[Stockless]

Taking the Motion test as reference, do I also have to create a testset for every simulation method or keep using only the Bloch method?

for sim_method in [KomaMRICore.Bloch(), KomaMRICore.BlochSimple(), KomaMRICore.BlochDict()]
@testset "$(typeof(sim_method))" begin
...

[cncastillo]

No, just look at what is inside, to see how to use DiffEq to get a solution.

[Stockless]

created a starting point for this test update in a new branch, commit