Add BurnBothEnds strategy

Author: marcharper

axelrod/strategies/_strategies.py

@@ -252,6 +252,7 @@
     Alexei,
     AntiTitForTat,
     Bully,
+    BurnBothEnds,
     ContriteTitForTat,
     DynamicTwoTitsForTat,
     EugineNier,
@@ -314,6 +315,7 @@
     BackStabber,
     BetterAndBetter,
     Bully,
+    BurnBothEnds,
     BushMosteller,
     Calculator,
     Capri,

axelrod/strategies/titfortat.py

@@ -919,3 +919,35 @@ def strategy(self, opponent: Player) -> Action:
 
         self.act_random = True
         return opponent.history[-1]
+
+
+class BurnBothEnds(Player):
+    """
+    Plays like TitForTat except it cooperates after opponent cooperation with probability 9/10.
+
+    Names:
+
+     - BurnBothEnds (BBE): Marinoff [Marinoff1992]_
+    """
+
+    name = "Burn Both Ends"
+    classifier = {
+        "memory_depth": 1,
+        "stochastic": True,
+        "long_run_time": False,
+        "inspects_source": False,
+        "manipulates_source": False,
+        "manipulates_state": False,
+    }
+
+    def strategy(self, opponent):
+        # First move
+        if not opponent.history:
+            # Make sure we cooperate first turn
+            return C
+        # BBE modification 
+        if opponent.history[-1] == C:
+            # Cooperate with 0.9
+            return self._random.random_choice(0.9)
+        # Else TFT. Opponent played D, so play D in return.
+        return D

axelrod/tests/strategies/test_titfortat.py

@@ -1316,3 +1316,28 @@ def test_deterministic_classification(self):
         for p in [0, 1]:
             player = axl.RandomTitForTat(p=p)
             self.assertFalse(axl.Classifiers["stochastic"](player))
+
+
+class TestBurnBothEnds(TestPlayer):
+    name = "Burn Both Ends"
+    player = axl.BurnBothEnds
+    expected_classifier = {
+        "memory_depth": 1,
+        "stochastic": True,
+        "makes_use_of": set(),
+        "inspects_source": False,
+        "manipulates_source": False,
+        "manipulates_state": False,
+    }
+
+    def test_vs_cooperator(self):
+        actions = [(C, C), (C, C), (C, C), (D, C), (C, C)]
+        self.versus_test(axl.Cooperator(), expected_actions=actions, seed=1)
+
+    def test_vs_cooperator2(self):
+        actions = [(C, C), (C, C), (C, C), (C, C), (C, C)]
+        self.versus_test(axl.Cooperator(), expected_actions=actions, seed=2)
+
+    def test_vs_defector(self):
+        actions = [(C, D), (D, D), (D, D), (D, D), (D, D)]
+        self.versus_test(axl.Defector(), expected_actions=actions)

docs/how-to/classify_strategies.rst

@@ -57,7 +57,7 @@ strategies::
     ... }
     >>> strategies = axl.filtered_strategies(filterset)
     >>> len(strategies)
-    87
+    88
 
 Or, to find out how many strategies only use 1 turn worth of memory to
 make a decision::
@@ -67,7 +67,7 @@ make a decision::
     ... }
     >>> strategies = axl.filtered_strategies(filterset)
     >>> len(strategies)
-    32
+    33
 
 Multiple filters can be specified within the filterset dictionary. To specify a
 range of memory_depth values, we can use the 'min_memory_depth' and
@@ -79,7 +79,7 @@ range of memory_depth values, we can use the 'min_memory_depth' and
     ... }
     >>> strategies = axl.filtered_strategies(filterset)
     >>> len(strategies)
-    56
+    57
 
 We can also identify strategies that make use of particular properties of the
 tournament. For example, here is the number of strategies that  make use of the

docs/index.rst

@@ -53,7 +53,7 @@ Count the number of available players::
 
     >>> import axelrod as axl
     >>> len(axl.strategies)
-    239
+    240
 
 Create matches between two players::
 

docs/reference/bibliography.rst

@@ -43,9 +43,8 @@ documentation.
 .. [Li2014] Li, J. and Kendall, G. (2014). The Effect of Memory Size on the Evolutionary Stability of Strategies in Iterated Prisoner's Dilemma. IEEE Transactions on Evolutionary Computation, 18(6) 819-826
 .. [LiS2014] Li, Siwei. (2014). Strategies in the Stochastic Iterated Prisoner's Dilemma. Available at: http://math.uchicago.edu/~may/REU2014/REUPapers/Li,Siwei.pdf
 .. [Luis2008] Luis R. Izquierdo and Segismundo S. Izquierdo (2008). Dynamics of the Bush-Mosteller Learning Algorithm in 2x2 Games, Reinforcement Learning, Cornelius Weber,  Mark Elshaw and Norbert Michael Mayer  (Ed.), InTech, DOI: 10.5772/5282. Available from: https://www.intechopen.com/books/reinforcement_learning/dynamics_of_the_bush-mosteller_learning_algorithm_in_2x2_games
-.. [Mathieu2015] Mathieu, P. and Delahaye, J. (2015). New Winning Strategies
-  for the Iterated Prisoner's Dilemma. Proceedings of the 2015
-  International Conference on Autonomous Agents and Multiagent Systems.
+.. [Marinoff1992] Marinoff, Louis. (1992). Maximizing expected utilities in the prisoner's dilemma. Journal of Conflict Resolution 36.1: 183-216.
+.. [Mathieu2015] Mathieu, P. and Delahaye, J. (2015). New Winning Strategies for the Iterated Prisoner's Dilemma. Proceedings of the 2015 International Conference on Autonomous Agents and Multiagent Systems.
 .. [Mittal2009] Mittal, S., & Deb, K. (2009). Optimal strategies of the iterated prisoner’s dilemma problem for multiple conflicting objectives. IEEE Transactions on Evolutionary Computation, 13(3), 554–565. https://doi.org/10.1109/TEVC.2008.2009459
 .. [Murase2020] Murase, Y., & Baek, S.K. (2020). Five Rules for Friendly Rivalry in Direct Reciprocity. Scientific Reports 10:16904 https://doi.org/10.1038/s41598-020-73855-x
 .. [Nachbar1992] Nachbar J., Evolution in the finitely repeated prisoner’s dilemma, Journal of Economic Behavior & Organization, 19(3): 307-326, 1992.