[TODO] Provide bridges to graph visualization libraries

[phcerdan]

Candidates:

• [ ] igraph (GPL)
• [x] networx
• [x] matplotlib?
• [x] graphviz??

[phcerdan]

First approach has been using networkx built-in capabilities of drawing. It uses pygraphviz for graphviz layouts. There is another physical layout with a GPLv2 license (forcelayout?), but I don't think it is useful for the case here.

The challenging part is to find an appealing layout for the different cases:

• [x] at generation N Using expressions.atoms and expressions.generation. Done using wolfram_model.vis.nx.grapH_at_generation. See tests: https://github.com/phcerdan/wolfram_model/blob/99856fa6af09ffcf357a5cba6303cde8abb92e58/python/tests/vis/test_vis.py#L24-L32

• [ ] evolution over generations It would be nice to keep fixed the position of nodes over future generation, where the automatic algorithm (dot for example) might otherwise reorder it.

• [ ] causal graphs We will have to use expression.creator_event and expression.destructor_event