1. Robert Axelrod
2. 1980a: 14+1 strategies
3. 1980b: 64+1 strategies

class TitForTat(Player):
    """A player starts by cooperating and then mimics previous move by opponent."""

    name = 'Tit For Tat'
    classifier = {
        'memory_depth': 1,  # Four-Vector = (1.,0.,1.,0.)
        'stochastic': False,
        'inspects_source': False,
        'manipulates_source': False,
        'manipulates_state': False
    }

    @staticmethod
    def strategy(opponent):
        return 'D' if opponent.history[-1:] == ['D'] else 'C'
					

class TestTitForTat(TestPlayer):

    name = "Tit For Tat"
    player = axelrod.TitForTat
    expected_classifier = {
        'memory_depth': 1,
        'stochastic': False,
        'inspects_source': False,
        'manipulates_source': False,
        'manipulates_state': False
    }

    def test_strategy(self):
        """Starts by cooperating."""
        self.first_play_test(C)

    def test_effect_of_strategy(self):
        """Repeats last action of opponent history."""
        self.markov_test([C, D, C, D])
        self.responses_test([C] * 4, [C, C, C, C], [C])
        self.responses_test([C] * 5, [C, C, C, C, D], [D])
					

Demo

Outcomes

• Convex programming applied to the PD
• Meta study of tournaments
• Machine learning strategies

Press and Dyson (2012): "Iterated Prisoner's Dilemma contains strategies that dominate any evolutionary opponent."

Optimise: $\frac{x^TQx+xc^t}{x^TQ'x+xc'^T}$

Lee, Harper and Dyer (2015): "The Art of War: Beyond Memory-one Strategies in Population Games"

http://axelrod.readthedocs.io/