Add Alexei strategy #997

souravsingh · 2017-04-30T07:42:59Z

Strategy is adapted from http://lesswrong.com/lw/7f2/prisoners_dilemma_tournament_results/

drvinceknight · 2017-04-30T10:49:45Z

Tests are failing on a syntax error.

meatballs · 2017-05-01T19:10:51Z

This is failing at axelrod/tests/strategies/test_player.py", line 361

AssertionError: 'D' != 'C'

souravsingh · 2017-05-01T19:14:39Z

@meatballs I am not sure if the strategy written is fully correct.

souravsingh · 2017-05-01T20:53:00Z

Can someone check the strategy written? It plays like Tit-for-Tat with a defection at the end.

drvinceknight · 2017-05-01T21:13:39Z

The strategy you have written does not do what you describe (there's no reference to the number of turns in it which is what is needed to be able to defect on the last move). Take a look at strategy transformers for an way to do this easily: http://axelrod.readthedocs.io/en/latest/tutorials/advanced/strategy_transformers.html

In particular use the Final Transformer: take a look at backstabber.

Nikoleta-v3

Hello @souravsingh. Your tests are currently failing at various places. Here are some comments to help you

Nikoleta-v3 · 2017-05-04T12:26:05Z

axelrod/strategies/titfortat.py

+    classifier = {
+        'memory_depth': 1,
+        'stochastic': False,
+        'makes_use_of': set(),


your strategy makes use of length.

Nikoleta-v3 · 2017-05-04T12:26:38Z

axelrod/tests/strategies/test_titfortat.py

+        self.versus_test(axelrod.Alternator(), expected_actions=actions)
+
+        actions = [(C, C), (C, C), (C, C), (C, C), (C, C)]
+        self.versus_test(axelrod.Cooperator(), expected_actions=actions)


You are getting a failure against Cooperator.

Nikoleta-v3 · 2017-05-04T12:27:07Z

axelrod/tests/strategies/test_titfortat.py

+    documentation.
+    """
+
+    name = "Alexei"


Now that your are using @FinalTransformer the name of your strategy is different.
AssertionError: 'Alexei: D' != 'Alexei'

Take a look at the backstabber tests https://github.com/Axelrod-Python/Axelrod/blob/master/axelrod/tests/strategies/test_backstabber.py.

Nikoleta-v3 · 2017-05-05T10:14:40Z

@souravsingh thank you for the changes, though you did not get everything correct.

now that you have included that your strategy is making use of length you should alter your test
Looking at how the @FinalTransformer works the name of the strategy should be Given Name: the action you specified. Thus 'Alexei: D'.
when you play against strategies could you make sure you write down the expected actions by hand first. For example your strategy should always defect on the last round but that's not the case in your tests.

I would recommend you run your tests before pushing and if you are getting an error you can't solve feel free to throw it in here!

Nikoleta-v3 · 2017-05-08T14:04:01Z

axelrod/tests/strategies/test_titfortat.py

+        self.versus_test(opponent, expected_actions=actions)
+
+        opponent = axelrod.MockPlayer(actions=[C, C, D, D, C, D])
+        actions = [(C, C), (C, C), (C, D), (D, D), (D, C), (C, D)]


Try to keep in mind how your strategy plays. Again I would do the expected actions by hand before writing the test. Your strategy has a rule, Defect on the last round

souravsingh · 2017-05-08T14:56:14Z

@Nikoleta-v3 All tests pass except for one which is unrelated to the strategy.

drvinceknight · 2017-05-08T15:00:04Z

All tests pass except for one which is unrelated to the strategy.

@souravsingh that failing test might be related to the strategy in some way (I'm not sure how) but all tests pass on the library so that failure is being cause by this addition.

drvinceknight · 2017-05-08T15:01:17Z

axelrod/strategies/titfortat.py

+    
+    Names:
+
+    - From http://lesswrong.com/lw/7f2/prisoners_dilemma_tournament_results/    


Can you reference this as for other strategies please.

This is the only strategy implemented from the site.

Yes that's fine but can you move the site to the bibliography and make the Names section fit the guidelines: http://axelrod.readthedocs.io/en/stable/tutorials/contributing/strategy/docstrings.html

Take a look at how Prison is written in the bibliography. Let me know if you'd like a hand.

Oh, I had forgotten the docstrings conventions for Axelrod.

(I think you actually wrote that file I linked to :) 🤣 👍 )

souravsingh · 2017-05-08T15:02:53Z

Was there an update made to test/unit/test_tournament.py before version 2.11 was released? If so, I just have to do a rebase and it should be fixed.

drvinceknight · 2017-05-08T15:06:00Z

Was there an update made to test/unit/test_tournament.py before version 2.11 was released? If so, I just have to do a rebase and it should be fixed.

I don't believe so but rebasing anyway sounds like a good idea. Then you can run the full test locally python -m unittest discover from the master branch and this branch.

drvinceknight · 2017-05-08T15:14:39Z

I've just confirmed that all tests on master pass (locally).

drvinceknight · 2017-05-08T15:26:53Z

And I've just confirmed the failure locally. It's failing on the test for probalistic ending tournaments with probability 1. So it's something to do with the strategy not behaving as expected when it does not know the length of the game. Could point to something not right with the Transformer. I'm investigating.

drvinceknight · 2017-05-08T15:30:03Z

This brings up the failure:

python -m unittest axelrod.tests.unit.test_tournament.TestProbEndingSpatialTournament

drvinceknight · 2017-05-08T15:43:42Z

The error seems to be something to do with the Final transformer not working when there sequence passed has length 1.

drvinceknight · 2017-05-08T15:44:25Z

axelrod/strategies/titfortat.py

        # Otherwise play previous move
        return self.history[-1]
+
+@FinalTransformer(D, name_prefix=None)


FinalTransformer takes a sequence. Change this to (D, ) (although that doesn't fix what's going wrong).

drvinceknight · 2017-05-08T15:59:10Z

axelrod/strategies/titfortat.py

+
+    name = 'Alexei'
+    classifier = {
+        'memory_depth': 1,


This strategy has memory depth: float('inf') as it needs to keep count of how many turns it has played to be able to know when to defect on the last round.

Fixing this will fix the other test failure as it was being picked up as a basic strategy (which it is not).

drvinceknight · 2017-05-08T15:59:24Z

axelrod/strategies/titfortat.py

        # Otherwise play previous move
        return self.history[-1]
+
+@FinalTransformer((D, ) name_prefix=None)


(D,) (no space)

drvinceknight · 2017-05-08T16:00:07Z

The error seems to be something to do with the Final transformer not working when there sequence passed has length 1.

For the record: this was not the issue, the transformer works exactly as expected.

drvinceknight · 2017-05-08T16:20:04Z

axelrod/strategies/titfortat.py

        # Otherwise play previous move
        return self.history[-1]
+
+@FinalTransformer((D,) name_prefix=None)


You now have a syntax error, type exactly this:

@FinalTransformer((D,), name_prefix=None)

There's information on running the tests (that would have picked this up immediately) here: http://axelrod.readthedocs.io/en/stable/tutorials/contributing/running_tests.html

souravsingh · 2017-05-08T19:46:53Z

@drvinceknight I have made the fixes. Can you review and suggest any more changes?

drvinceknight

I'll review properly as soon as I get a moment @souravsingh.

drvinceknight · 2017-05-09T07:26:52Z

docs/reference/bibliography.rst

 .. [Frean1994] Frean, Marcus R. “The Prisoner's Dilemma without Synchrony.” Proceedings: Biological Sciences, vol. 257, no. 1348, 1994, pp. 75–79. www.jstor.org/stable/50253.
 .. [Hilde2013] Hilbe, C., Nowak, M.A. and Traulsen, A. (2013). Adaptive dynamics of extortion and compliance, PLoS ONE, 8(11), p. e77886. doi: 10.1371/journal.pone.0077886.
 .. [Kraines1989] Kraines, D. & Kraines, V. Theor Decis (1989) 26: 47.  doi:10.1007/BF00134056
+.. [LessWrong2011] Zoo of Strategies (2011) LessWrong. Available at: http://lesswrong.com/lw/7f2/prisoners_dilemma_tournament_results/ 


All on one line please. I think this doesn't play properly with sphinx otherwise.

I think I have written in one line. Unless I am misunderstanding something.

You have indeed. I'll review properly when I get a moment.

drvinceknight

Looking good, some minor adjustments to the tests from me.

drvinceknight · 2017-05-10T07:14:31Z

axelrod/tests/strategies/test_titfortat.py

+        actions = [(C, D), (D, D), (D, D), (D, D), (D, D)]
+        self.versus_test(axelrod.Defector(), expected_actions=actions)
+
+        actions = [(C, C), (C, D), (D, C), (C, D), (D, C)]


Could you add one more round to this actions, the point of having match_attributes={"length":-1} is to show how the strategy acts when playing without knowledge of the when the last round is so the last actions would be:

actions = [(C, C), (C, D), (D, C), (C, D), (D, C), (C, D)]

Then can you repeat the exact same test but without match_attributes which should then have:

actions = [(C, C), (C, D), (D, C), (C, D), (D, C), (D D)]

Which shows that the "last round" instruction is working. (So just move your other alternator test here and add the extra round).

drvinceknight · 2017-05-10T07:16:50Z

axelrod/tests/strategies/test_titfortat.py

+        self.versus_test(axelrod.Alternator(), expected_actions=actions,
+                         match_attributes={"length": -1})
+
+        actions = [(C, D), (D, D), (D, C), (C, C), (D, D)]


I think all the test case will be covered by the cases above, we can simplify this and remove everything below.

drvinceknight · 2017-05-10T11:37:24Z

axelrod/tests/strategies/test_titfortat.py

+        self.versus_test(axelrod.Alternator(), expected_actions=actions)
+
+
+        actions = [(C, D), (D, D), (D, C), (C, C), (D, D)]


As long as coverage remains the same, I think we can remove this test.

drvinceknight · 2017-05-10T11:37:28Z

axelrod/tests/strategies/test_titfortat.py

+        self.versus_test(axelrod.Random(), expected_actions=actions,
+                         seed=0)
+
+        actions = [(C, C), (C, D), (D, D), (D, C)]


As long as coverage remains the same, I think we can remove this test.

drvinceknight · 2017-05-10T11:37:32Z

axelrod/tests/strategies/test_titfortat.py

+        self.versus_test(axelrod.Random(), expected_actions=actions,
+                         seed=1)
+
+        opponent = axelrod.MockPlayer(actions=[C, D])


As long as coverage remains the same, I think we can remove this test.

drvinceknight · 2017-05-10T11:37:38Z

axelrod/tests/strategies/test_titfortat.py

+        actions = [(C, C), (C, D), (D, C), (D, D)]
+        self.versus_test(opponent, expected_actions=actions)
+
+        opponent = axelrod.MockPlayer(actions=[C, C, D, D, C, D])


As long as coverage remains the same, I think we can remove this test.

drvinceknight · 2017-05-10T11:38:00Z

axelrod/tests/strategies/test_titfortat.py

+        self.first_play_test(C)
+        self.second_play_test(rCC=C, rCD=D, rDC=C, rDD=D)
+
+        actions = [(C, C), (C, D), (D, C), (C, D), (D, C)]


We can remove this test because you moved it.

drvinceknight · 2017-05-10T12:18:29Z

This looks great to me. Thanks @souravsingh 👍

souravsingh added 3 commits April 30, 2017 13:02

Update _strategies.py

0b71b6e

Update titfortat.py

fa70b52

Update test_titfortat.py

0be3624

souravsingh added 2 commits April 30, 2017 20:35

Fix error

4c471b9

Update classification_of_strategies.rst

640712e

souravsingh added 2 commits May 3, 2017 05:06

Add Finaltransformer in Alexei strategy

de8f570

Update titfortat.py

c977452

Nikoleta-v3 requested changes May 4, 2017

View reviewed changes

souravsingh added 3 commits May 4, 2017 23:07

Make changes to strategy

a0b9f63

Update test_titfortat.py

b70c5c2

Update test_titfortat.py

693ea45

souravsingh added 6 commits May 7, 2017 01:43

Fix name of strategy

e83de3a

Fix classifier parameter

1aa8b4f

Update titfortat.py

fd1fa66

Update test_titfortat.py

fd55bf6

Update classification_of_strategies.rst

ea8bb11

Fix test

d3207d6

Nikoleta-v3 reviewed May 8, 2017

View reviewed changes

Update test_titfortat.py

5c1d688

drvinceknight requested changes May 8, 2017

View reviewed changes

Fix issue with FinalTransformer

62bb722

drvinceknight requested changes May 8, 2017

View reviewed changes

souravsingh added 2 commits May 8, 2017 21:44

Update titfortat.py

7d1a8cf

Update the memory_depth parameter

f8db68d

drvinceknight requested changes May 8, 2017

View reviewed changes

souravsingh added 5 commits May 8, 2017 22:04

Fix missing comma between parameters

d789c38

Change name of strategy

b10c09d

Fix doctest

bb55e6c

Add reference to LessWrong for Alexei

e3fe33b

Update titfortat.py

f0f863a

drvinceknight requested changes May 9, 2017

View reviewed changes

drvinceknight requested changes May 10, 2017

View reviewed changes

Add test cases

02a5801

drvinceknight requested changes May 10, 2017

View reviewed changes

Remove a few tests

54c54fc

drvinceknight approved these changes May 10, 2017

View reviewed changes

drvinceknight added the ready-to-merge label May 10, 2017

meatballs merged commit cb3bccc into Axelrod-Python:master May 10, 2017


		Names:

		- From http://lesswrong.com/lw/7f2/prisoners_dilemma_tournament_results/

		self.versus_test(axelrod.Alternator(), expected_actions=actions)


		actions = [(C, D), (D, D), (D, C), (C, C), (D, D)]

Add Alexei strategy #997

Add Alexei strategy #997

Uh oh!

Conversation

souravsingh commented Apr 30, 2017

Uh oh!

drvinceknight commented Apr 30, 2017

Uh oh!

meatballs commented May 1, 2017

Uh oh!

souravsingh commented May 1, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

souravsingh commented May 1, 2017

Uh oh!

drvinceknight commented May 1, 2017

Uh oh!

Nikoleta-v3 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Nikoleta-v3 commented May 5, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

souravsingh commented May 8, 2017

Uh oh!

drvinceknight commented May 8, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

souravsingh commented May 8, 2017

Uh oh!

drvinceknight commented May 8, 2017

Uh oh!

drvinceknight commented May 8, 2017

Uh oh!

drvinceknight commented May 8, 2017

Uh oh!

drvinceknight commented May 8, 2017

Uh oh!

drvinceknight commented May 8, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

drvinceknight commented May 8, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

souravsingh commented May 8, 2017

Uh oh!

drvinceknight left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

drvinceknight left a comment

souravsingh commented May 1, 2017 •

edited

Loading