For a model, simulate(model, dr) returns an xarray with labels corresponding to variables, time index and simulation index.
It would make sense for process.simulate() to also return an xarray. Among the advantages:
no confusion about the dimensions ordering (it's never clear whether it's a N.T or a T.N array
we could carry variable names
As a side note, for consistency, 1d iid processes should return a 1.N.T array.
For a model,
simulate(model, dr)
returns an xarray with labels corresponding to variables, time index and simulation index. It would make sense forprocess.simulate()
to also return an xarray. Among the advantages:N.T
or aT.N
array1.N.T
array.