Coupling graph

This example illustrates the notion of coupling graph.

from __future__ import annotations

from gemseo import generate_coupling_graph
from gemseo.utils.discipline import DummyDiscipline

Create the disciplines

First, we create dummy disciplines that do nothing:

dummy_disciplines = [
    DummyDiscipline(name=name, input_names=input_names, output_names=output_names)
    for (name, input_names, output_names) in (
        ("A", ["a"], ["b"]),
        ("B", ["c"], ["a", "n"]),
        ("C", ["b", "d"], ["c", "e"]),
        ("D", ["f"], ["d", "g"]),
        ("E", ["e"], ["f", "h", "o"]),
        ("F", ["g", "j"], ["i"]),
        ("G", ["i", "h"], ["k", "l"]),
        ("H", ["k", "m"], ["j"]),
        ("I", ["l"], ["m", "w"]),
        ("J", ["n", "o"], ["p", "q"]),
        ("K", ["y"], ["x"]),
        ("L", ["w", "x"], ["y", "z"]),
        ("M", ["p", "s"], ["r"]),
        ("N", ["r"], ["t", "u"]),
        ("O", ["q", "t"], ["s", "v"]),
        ("P", ["u", "v", "z"], ["z"]),
    )
]

Generate the coupling graph

The coupling graph represents each discipline by a node and each coupling variable by an edge. By default, the generate_coupling_graph() function saves the graphical representation of the coupling graph as a PDF file. Here, we prefer to display it in this web page by setting file_path to "" (the same would work in a notebook).

generate_coupling_graph(dummy_disciplines, file_path="")

WARNING - 16:24:52: Two disciplines, among which P, compute the same outputs: {'z'}

We can also draw the condensed coupling graph, where each groups of strongly coupled disciplines is represented by a node:

generate_coupling_graph(dummy_disciplines, file_path="", full=False)

WARNING - 16:24:52: Two disciplines, among which P, compute the same outputs: {'z'}