Implemented White strategy, k72r from Axelrod's Second #1154
Conversation
There was a problem hiding this comment.
More of a question then a request but just want to be sure (could indicate an incorrect ranking in axelrod_fortran or an off by 1 error).
Note that changing https://github.com/Axelrod-Python/axelrod-fortran/blob/master/src/axelrod_fortran/strategies.py#L216 to be False so that a value of R of 0 is passed to the Fortran code changes the behaviour of the fortran function.
axelrod/strategies/axelrod_second.py
Outdated
| class White(Player): | ||
| """ | ||
| Strategy submitted to Axelrod's second tournament by Edward C Whtie (K72R) | ||
| and came in twelfth in that tournament. |
There was a problem hiding this comment.
Looking at https://github.com/Axelrod-Python/axelrod-fortran/blob/master/src/axelrod_fortran/strategies.py#L214 it looks like it ranked 13th?
There was a problem hiding this comment.
It should say thirteenth, and actually the last few would be off. I'll fix them.
There was a problem hiding this comment.
I got thrown off at Kluepfel, which had already been implemented since I've started doing these in order.
There was a problem hiding this comment.
changes the behaviour of the fortran function
Changes how? https://github.com/Axelrod-Python/TourExec/blob/v0.3.0/src/strategies/k72r.f#L12 must be rounding down the right-hand side, otherwise the 100% cooperation in the first ten moves (in the fingerprints) wouldn't make sense.
There was a problem hiding this comment.
If you pass a 0 (which is what we do when marking a function as stochastic: False) then the function seems to always defect.
It's not an issue, just noting it. I think that essentially that stochastic: key is whether or not the original Fortran code used the random number anywhere, even if as you've noticed here, it ended up not being relevant.
Conflicts: axelrod/strategies/axelrod_second.py
|
Thanks! (my bad for missing those rankings in the docstrings previously. 👍) |
|
Nudge @marcharper and @meatballs (in case you've missed this): I have checked the logic and this looks good to go 👍 |
axelrod/strategies/axelrod_second.py
Outdated
|
|
||
| class White(Player): | ||
| """ | ||
| Strategy submitted to Axelrod's second tournament by Edward C Whtie (K72R) |
axelrod/strategies/axelrod_second.py
Outdated
| Strategy submitted to Axelrod's second tournament by Edward C Whtie (K72R) | ||
| and came in thirteenth in that tournament. | ||
|
|
||
| If the opponent Cooperated last turn or in the first ten turns, then |
There was a problem hiding this comment.
"""
* If the opponent cooperated last turn or in the first ten turns, then
cooperate.
* Otherwise defect if and only if:
floor(log(turn)) * opponent.defections >= turn
"""
|
Looks fine to me, please fix the typo in the name at least. |
|
Made those changes. |
|
Great. Thanks 👍 |
Fingerprints below.
Notes on refactor from https://github.com/Axelrod-Python/TourExec/blob/v0.3.0/src/strategies/k72r.f :
N, M, and JCOUNT are all integers, so everything gets rounded. So the probability statement (line 12) actually only ever gets run with 0% or 100% (evidenced by the basic-strategy fingerprints). And actually unless the if-statement in line 11 gets run, then the probability statement always Cooperates, since JCOUNT can be at most M-1.