Problem running example

1

hi all,
walking through the example here: https://github.com/PumasAI/Pumas.jl, I’m seeing the following error right after res = fit(model,data,param,Pumas.FOCEI()):

03%20PM
Screen Shot 2019-07-30 at 3.19.03 PM.png956×170 40.3 KB

any ideas?

2

Hi Ron,

You say you are getting this error just by running the example in the README, right?

It looks like you’re running this in Juno. Can you verify the issue in a fresh session? That is, close all Juno/Atom windows down, and run the code in the exact order that is in the README.md once more. Please copy it over again instead of using a file you already have on your computer, just to be sure that nothing was modified.

Best,
Patrick

3

hi Ron,

Looks like there is a small typo in the readme, but also something that we need to check internally.

You need to change the specification of omega in the param to be like this below

param = (
  tvcl = 4.0,
  tvv  = 70,
  pmoncl = -0.7,
  Ω = Diagonal([0.09,0.09]),
  σ_prop = 0.04
  )

and since Diagonal is part of the LinearAlgebra ecosystem, please change the list of libraries being called to

using Pumas, Plots, LinearAlgebra

this should get you going.

Technically, we should be able to pass in just an array [0.09, 0.09] instead of Diagonal([0.09,0.09]). We are checking on that and will get back to you.

We will also update the readme to reflect this.

4

I encountered a similar problem trying to run the Introduction Jupyter notebook from the PumasTutorials package.

julia> Pkg.status("Pumas")
    Status `~/.juliapro/JuliaPro_v1.1.1.1/environments/v1.1/Project.toml`
  [4f2c3c20] Pumas v0.1.0

julia> Pkg.status("PumasTutorials")
    Status `~/.juliapro/JuliaPro_v1.1.1.1/environments/v1.1/Project.toml`
  [b7b41870] PumasTutorials v0.0.1 #master (https://github.com/PumasAI/PumasTutorials.jl)

The code block after using Pumas calls the @param macro but that macro either doesn’t exist or is not exported.

@param begin
  θ ∈ VectorDomain(12)
end

UndefVarError: @param not defined
5

It appears that the call to @param must be made within a call to @model

6

Thanks Patrick and Vijay, the Diagonal() omega fixed it. Agree, would probably be best if the simobs() and fit() functions would take the same parameter types.
Will be testing more next few weeks, excited to see this progress.

7

Yes Doug. The @model macro is the place where all the model components are defined. Please refer to the documentation here - https://docs.pumas.ai/dev/basics/models/

8

Thanks, Ron. I think it is a quick fix. Will post it here when done!

9

Right now that’s a hard constraint, and it’s difficult to fix because of how the parser works. However, it’s not impossible to add incremental construction in the future.