Issue with decode the mode first

[senna1128]

Hi,

I'm using Julia 1.5.2. I try to run multiple models in parallel using your julia code provided here https://github.com/JuliaSmoothOptimizers/CUTEst.jl. I also pass decode=false.

But I got an error like

ERROR: On worker 7:
CUTEst: lib not found; decode problem first

This seems CUTEstModel must decode problem first. If I use standard for loop without running in parallel, it works well. So my cutest package should be fine.

Is there any way to fix it? Also I cannot run

fetch_sif_problems()

The error is

ERROR: UndefVarError: fetch_sif_problems not defined

Are they related? Not using fetch_sif_problems does not affect my usage of cutest at the moment, but I'm not sure if it matters in the future.

Best,
Sen

[abelsiqueira]

Hi @senna1128, thanks for the report.

You indeed need to decode the problem first using sifdecoder. You could decode it at a different time and then call CUTEstModel, but at some point the sifdecoder has to be called. This is related: #145 (comment)

fetch_sif_problems was removed, it's not needed anymore, but we forgot to update the README.

[abelsiqueira]

By the way, I forgot we had this information on the README: https://github.com/JuliaSmoothOptimizers/CUTEst.jl#run-multiple-models-in-parallel

[dpo]

That's very clever. I also forgot that we could do that!

[senna1128]

Hi @abelsiqueira,

Thank you very much for the quick reply. I tried to run the code

addprocs(2)
@everywhere using CUTEst
@everywhere function opt(name, decode)
    nlp = CUTEstModel(name;decode=decode)
    retval = obj(nlp,nlp.meta.x0)
    #retval = nlp.meta.nvar
    finalize(nlp)
    retval
end

probs = sort(CUTEst.select(contype="unc",max_var=5))
map(x->opt(x,true), probs) # Ensure all the problems have been decoded
pmap(x->opt(x,false), probs)

provided in #145 comment. I could get the result for the commandmap(x->opt(x,true), probs) # Ensure all the problems have been decoded . It returned 95 objective values. While it still reported the same error in pmap(x->opt(x,false), probs). Do you think that is my installation issue? I'm using Julia 1.5.2 via JuliaPro.

[abelsiqueira]

I've had the same error. I will investigate this today.

[abelsiqueira]

Took me a while to get to it, sorry about that. I think the issue is fixed by #242.

[dpo]

@senna1128 could you confirm if your isse has been fixed?

[dpo]

Please feel free to open a new issue if problems persist.