In anticipation of 5.0.0 I wonder if the ResultSet is perhaps worth simplifying a lot. Currently it does a huge amount of calculations (some includes a fair bit of complexity to do all the stuff on disc etc...) that are very specific to the IPD.
I think there are 2 possibilities:
- Keep the
ResultSet and aim to have it generic so as to do the same amount of calculations for any/all games.
- Make the generic output of the
Tournament.play() be a payoff matrix and include the option to write the interactions. For the case of the IPD this would lead to no loss of functionality as we would "simply" move the current ResultSet calculations to a specific analysis.ipd spot that we could point at the interactions.
I very much prefer option 2 and wonder if that would streamline things so that we could use the Tournament class in the Moran process? (but this is a minor point)
In anticipation of
5.0.0I wonder if theResultSetis perhaps worth simplifying a lot. Currently it does a huge amount of calculations (some includes a fair bit of complexity to do all the stuff on disc etc...) that are very specific to the IPD.I think there are 2 possibilities:
ResultSetand aim to have it generic so as to do the same amount of calculations for any/all games.Tournament.play()be a payoff matrix and include the option to write the interactions. For the case of the IPD this would lead to no loss of functionality as we would "simply" move the currentResultSetcalculations to a specificanalysis.ipdspot that we could point at the interactions.I very much prefer option 2 and wonder if that would streamline things so that we could use the
Tournamentclass in the Moran process? (but this is a minor point)