Open odow opened 1 year ago
I think this is really interesting and helpful. I will definitely add this. Thanks for the suggestion!
Hi Oscar , How are you? I am coming to the last stages of the benchmarking and I have two new additional real-world problems that the MSPFormat reader cannot read successfully.
Problem_16.tar.gz Problem_17.tar.gz
I am also slowly addressing each issues on the MSPLib GitHub repo (specifically this JSON Schema and the MetaData). Thanks again for the recommendations.
Do you have a log of the errors?
(p.s. Are you coming to Paris with Nils?)
For Problem #16:
For Problem #17:
Vincent Leclere invited me to give a talk but I could not as I have my defence on the 13th October. It would be nice though. Nils is going only for his talk, if I am not mistaken.
If you have time, Luxembourg is a 2-hour train ride and I I can show you around!
Hmm. So these errors are because storage_5
appears as a state variable, but it does not appear in every stage. We can work-around this in the SDDP reader, but more generally, we should ask: does the vector of state variables need to be constant across the nodes in a graph?
The files contain Dict{String, Any}("MIN" => 9.549789656601526)
and Dict{String, Any}("MAX" => 0.0)
. What are these nodes?
Hmmm. I am also not sure. Let me check this up with Nils.
We should add a JSON schema that can be used to validate lattice and problem files.
As examples, see:
JSON-schema is defined by:
There is a Python library for validating: