Adding (B)AGGs to the catalog

[rahulsavani]

This feature request is for...

Other

Requested enahancement or feature

We have decided that all games used in the tests should be in the catalog going forward, even if they are only interesting in the context of tests.

Gambit supports: EFGs, NFGs, AGGs and BAGGs.

In issue #859, I am adding some regression tests for (B)AGGs. Thus we need (B)AGGs in the catalog.

Later, it will be very cool to "draw" them, but this will NOT be with draw_tree, but something new (though very likely also tikz based). Later, we will also be adding more interesting examples of (B)AGGs. For now, we just need any way to have them in the catalog.

[rahulsavani]

@wyz2368 @tturocy : Thoughts welcome.

[edwardchalstrey1]

Started working on this on catalog/937 which also includes changes from #859

[wyz2368]

Thanks Ed. I will talk to the student at UNF who works on the translation from natural language game description to (B)AGGs. Maybe he can help to collect some (B)AGGs for the catelog first. The first sources we should cover are the original papers. Then we can explore more papers studying (B)AGGs.

[tturocy]

To summarise briefly the conversation from a call:

• We definitely want (B)AGGs in the catalog eventually. "Eventually" likely means post-17.0, however, for the following reasons.
• The (B)AGG format does not have a way to specify a game title or description, other than in the comments. So either we need to change the format, or we start hacking reading this out of comments. The latter is not a great idea; the former is something we really should think about before we do.
• We do have a principle that we want to avoid having games in tests/test_games unnecessarily. And possibly in the fullness of time we would have none. However the spirit of this initiative is driven by the observation that we have wound up with an uncurated mess of games in test_games. Many of those tests could be accomplished by "standard" games from the literature, or, failing that, games we construct for the purpose which fill gaps in the literature. Such games really should be in the catalog to make them discoverable.
• Nevertheless at least in the interim we may have games that we want to keep around purely for testing mechanical or implementation details, and for these keeping a small number of games in tests/test_games would be appropriate.
• Because for 17.0 we are principally concerned that we don't want to be introducing regressions into our (B)AGG support, keeping the needed (B)AGG files for testing in tests/test_games would be OK.

So yes we would like to progress this, both in collecting (B)AGG files and also how to visualise them. We probably won't however add them to the catalog as a priority for 17.0 given everything else we have on our plate.

[wyz2368]

@tturocy What is the difficulty for changing the format here or what are the key considerations? Can we simply update the format by adding the meta data? For example,

#AGG
# META {
# META   "title": "Market entry game",
# META   "description": "An action-graph representation of a two-firm market entry game.",
# META   "source": "Example adapted from Author (Year), Section 3."
# META }

<Previous AGG body starts here>

[tturocy]

@wyz2368 First, it's not "our" format - the AGG creators made it (and in fact the #AGG/#BAGG magic at the top is something that got added for Gambit and isn't in the spec proper).

Second, if we are going to make changes, we should think about them carefully and not rush into them. We already have a lot going on. (B)AGG support is important but it is something we will be adding over many months, there is time to think.