Closed ziyiyin97 closed 2 years ago
BTW there isn't any problem if
options = Options(return_array=true)
F = judiModeling(model, srcGeometry, recGeometry; options=options)
d_obs = F*q
d_obs = F.options.return_array ? reshape(d_obs, F.rInterpolation, F.model; with_batch=true) : d_obs
A bit tricky because the shape of dm
is uncertain here https://github.com/slimgroup/JUDI.jl/blob/6c539b1cb002f54bd4d7c59bcbf55cab27d36f5d/src/rrules.jl#L25 any thought?
You either work with a vector or a PhysicalParameter not an array.
Thanks. Precisely speaking we need to work with a Matrix
in size (prod(n), 1)
because https://github.com/slimgroup/JUDI.jl/blob/5e3b99f8c90298516170685e0f22129650e2f822/src/TimeModeling/Types/ModelStructure.jl#L94 -- any reason why PhysicalParameter as an AbstractVector has a size in length of 2, i.e. why not (prod(n),)
?
Should be (prod(A.n),)
MFE is here
As you can see in the first example, these two got different sizes
thus when working with
v0 = sqrt.(1f0./m0)
, there is a shape mismatch error