Create function to plot

[abelsiqueira]

Create a function to create graph plots and attach information to the graph.
Check https://github.com/JuliaGraphs/NetworkLayout.jl or https://github.com/JuliaGraphs/GraphPlot.jl

The following functions for plots will be added:

• time series for a single flow
• graph with flow values
• balance (stacked bar charts) for assets
• time series for all flows?

[clizbe]

I started playing with this, but if someone else wants to tackle it while I'm gone, go ahead.

[clizbe]

I've made the first graph and am working on the second.
I would like to scale the thickness of the connecting lines by the total capacity (initial+investment) and maybe the same for the size (or color) of the nodes - or maybe just labels.

But I can't figure out how to get this data. It means adding a Dict and a DenseAxis Array:

[clizbe]

I found a function "merge" that can add Dicts based on their keys, but it doesn't recognize a DenseAxisArray:

[abelsiqueira]

Indeed, there is no way to nicely sum these because they are different structures but also because they are not indexed on the same sets.

parameters.flows_unit_capacity is indexed on F and solution.flows_investment is indexed on Fi.

So you have to iterate on the flows and aggregate these based on what you want.

If you want all flows_unit_capacity, and if they are investable, add the solution, then you can do.

total = copy(parameters.flows_unit_capacity)
for f in sets.flows
    if parameters.flows_investable[f]
        total[f] += solution.flows_investment[f]
    end
end

And a += b is the same as a = a + b.

[clizbe]

Yes that's exactly what I need, thanks!
But also I think this is something we'll want in the output, so it might make sense to put this is post-processing, or otherwise change the structure of the data. 🤔

[clizbe]

Is there a way to ask the graph where it's connections are FROM and TO? Because now I have this data in a Dict{Tuple{String, String}, Float64} but the gplot parameter edgelinewidth is expecting just a Vector that somehow matches the order of the flows in the graph.

[clizbe]

Google seems to suggest we use a MetaGraph or MetaGraphsNext which would allow node names and weights?
https://discourse.julialang.org/t/identifying-nodes-in-a-graph/41498

[datejada]

Is there a way to ask the graph where it's connections are FROM and TO? Because now I have this data in a Dict{Tuple{String, String}, Float64} but the gplot parameter edgelinewidth is expecting just a Vector that somehow matches the order of the flows in the graph.

From the graph structure, something like this will give you the id and the name:

for e ∈ edges(graph)
   from_asset_id = e.src
   from_asset_name = sets.assets[e.src]
   to_asset_id = e.dst
   to_asset_name = sets.assets[e.dst]
end

If you are using the F set, then something like this should work too:

for f ∈ F
   from_asset_name = f[1]
   to_asset_name = f[2]
end

Depending on the information in your function, one option will be more straightforward.

@abelsiqueira, any comment, suggestion or thoughts? Thanks!

[clizbe]

With the new data format, I would like to show a directed graph with:

• Node names with capacity values (done)
• Export flows with capacity (normal directed graph)
• Import flows with capacity (reverse directed graph)

I know how to calculate the export/import caps for flow A->B, but how can I get the reverse direction arrows onto the same graph to label with the import capacity? The graph just thinks there's 1 arrow that can have a positive or negative flow - but I need to show two arrows with their own capacities.

[clizbe]

I could do a non-directed graph and make the labeled capacity be a different color for export/import (or display on a different part of the line), but I think having double arrows is the better solution.

[abelsiqueira]

You will need to create a separate graph for that, since the one we have is indeed only from A to B (unless explicitly declared from B to A in the csv as well).

Here is how to construct one such graph using the existing graph:

graph
dgraph = DiGraph(nv(graph)) # Creates a graph with the same number of vertices
for e in edges(graph)
    add_edge!(dgraph, e.src, e.dst)
    add_edge!(dgraph, e.dst, e.src)
end

The thing is, dgraph is NOT a MetaGraph. So using dgraph[u, v].some_field will not work, so you have to get those properties from graph. Let me know if you need more info.