New Strategy From the PRISON (http://www.lifl.fr/IPD/ipd.frame.html) #724

AdamPohl · 2016-09-25T21:32:11Z

Implemented Worse and Worse strategy as suggested in #379 and Knowledgeable Worse and Worse which I started during the PyconUK Axelrod sprint. This pr is the same as #720 but now hopefully the appveyer test should pass

AdamPohl · 2016-09-25T22:21:21Z

I have absolutely no idea as to why the tests are failing, could it be because I am using sys.version_info in my test function? 😞 😕 😖

drvinceknight · 2016-09-26T05:25:13Z

The error messages indicate that you haven't quite coded your strategy right for the case when match lengths are not known by the players.

There should not be a need to use sys.version either.

I will look through this today and fix it.

drvinceknight · 2016-09-26T09:44:37Z

I've opened AdamPohl#1 to your branch @Huaraz2.

However, I'm not actually sure this is the correct strategy. I couldn't find this strategy at http://www.lifl.fr/IPD/ipd.frame.html and looking through what I could find in the literature I can't seem to find any details (please point me towards them though).

AdamPohl · 2016-09-26T21:07:02Z

Thanks for the pr @drvinceknight, I found this strategy in the classics.str file in the prison package that you can get from the downloads page in the link I have attached to this pr. As far as I understood it, each mactch in the tournament would last for 1000 turns, so I read the strategy as: it defects with probability of current_turn / expected_length. I now do not think this is the case and therefore will rename this one to WorseAndWorseRandom and make WorseAndWorse for real.

drvinceknight

Thanks for clarifying the source. I've got a couple of change requests but this is looking like it's on the right road. 👍

drvinceknight · 2016-09-27T06:52:32Z

axelrod/strategies/worse_and_worse.py

+class WorseAndWorse (Player):
+    """
+    Defects with probability of 'current turn / 1000'. Therefore
+    it is more and more likely to defect as the round goes on.


Could you add:

Source code available at the download tab of [PRISON1998].

Would this line go at the end of line 10 or do you want it as part of the names section?

In between line 10 and the names section. So:

Defects with probability ... is more likely ... Source code available ... Names...

Thanks for the clarification, I have now made this change.

drvinceknight · 2016-09-27T06:54:05Z

axelrod/strategies/worse_and_worse.py

+
+    def strategy(self, opponent):
+        current_round = len(self.history) + 1
+        if random.randint(0, 1000) < current_round:


Can you rewrite this so that it's like the other strategy? IE so that is uses random_choice:

probability = 1 - float(current_round) / 1000 return random_choice(probability)

When you do that you can remove the import random at the top of this file.

drvinceknight · 2016-09-27T06:55:10Z

axelrod/strategies/worse_and_worse.py

+        return C
+
+
+class WorseAndWorseRandom (Player):


Not a fan of this name: it makes it sound like WorseAndWorse is not random. As this is a version of WorseAndWorse that has some knowledge I suggest you call it KnowledgeableWorseAndWorse.

drvinceknight · 2016-09-27T06:56:04Z

axelrod/strategies/worse_and_worse.py

+    """
+    This strategy is based of 'Worse And Worse' but will defect with probability
+    of 'current turn / total no. of turns'.
+


This still needs a names section:

Names: - Knowledgeable Worse and Worse: Original name by Adam Phol

drvinceknight · 2016-09-27T06:57:18Z

axelrod/tests/unit/test_worse_and_worse.py

+        opponent = axelrod.Cooperator()
+        player = axelrod.WorseAndWorse()
+        match = axelrod.Match((opponent, player), turns=5)
+        self.assertEqual(match.play(), [('C', 'C'),


Once you've made the expected change, this might change but either way it would be good if you picked a seed that didn't just have WorseAndWorse cooperate (ie choose one that gives a mix of C and D). Change the strategy first though.

drvinceknight · 2016-09-27T06:57:51Z

docs/reference/all_strategies.rst

@@ -120,6 +123,6 @@ Here are the docstrings of all the strategies in the library.
 .. automodule:: axelrod.strategies.titfortat
   :members:
   :undoc-members:
-.. automodule:: axelrod.strategies.gradualkiller


Do not delete gradualkiller.

I didn't, if you look further up you will see that I moved it so that the list was in alphabetical order again.

Ah great: my mistake :) 👍

drvinceknight · 2016-09-27T19:18:49Z

axelrod/strategies/worse_and_worse.py

@@ -10,11 +8,12 @@ class WorseAndWorse (Player):
    it is more and more likely to defect as the round goes on.

    Names:
-        - worse_and_worse: [PRISON1998]
+        - worse_and_worse: Source code available at the download tab of
+                            [PRISON1998].


Does this indent render ok? (you can run make html to check from the docs directory, you might have to pip install a sphinx theme...)

I think the rst syntax should be:

- worse_and_worse: Source code available at the download tab of [PRISON1998].

drvinceknight

Nearly there as far as I am concerned.

drvinceknight · 2016-09-28T08:05:07Z

axelrod/strategies/worse_and_worse.py

+
+class KnowledgeableWorseAndWorse (Player):
+    """
+    This strategy is based of 'Worse And Worse' but will defect with probability


Typo: should be 'is based on'

Thanks, I have now made these changes

drvinceknight · 2016-09-28T08:10:37Z

axelrod/strategies/worse_and_worse.py

+
+
+    Names:
+        - worse_and_worse: [PRISON1998].


You need the _ at the end of the [PRISON1998]. So this should be [PRISON1998]_ same for the one above (line 10).

Also can you change it to be:

- Worse and Worse: [PRISON1998]_

drvinceknight · 2016-09-28T11:18:17Z

axelrod/strategies/_strategies.py

@@ -195,12 +195,14 @@
    Willing,
    WinShiftLoseStay,
    WinStayLoseShift,
+    WorseAndWorse,
+    KnowledgeableWorseAndWorse,


Can you put this in alphabetical order please.

drvinceknight

This now looks good to me. We have a two core reviewer policy so it needs to be checked through by someone else. Thanks for your work on it :)

AdamPohl · 2016-09-28T13:40:46Z

Thanks Vince, it's been fun 👍

AdamPohl and others added 9 commits September 26, 2016 23:22

Adding WorseAndWorse stratergy

45368a7

Adding WorseAndWorse stratergy attempt 2

5b6ebfb

sorting out try/except setup as per Vince's request.

593ae56

Fixed python 2.7 issue YAAAY!!! 😈

e0a4fda

Make tests more verbose.

384d49c

Correct strategy.

9c397ab

Modify test function to set all seeds.

008656c

Have added actual Worse and worse strategy now.

bf9e9b9

fixing docs

0209f71

AdamPohl force-pushed the WorseAndWorse branch from 62f4662 to 0209f71 Compare September 26, 2016 22:22

drvinceknight requested changes Sep 27, 2016

View reviewed changes

Have made the changes that @drvinceknight suggested.

53f68e8

drvinceknight requested changes Sep 27, 2016

View reviewed changes

Made the change @drvinceknight sugested.

4ec2ba9

drvinceknight requested changes Sep 28, 2016

View reviewed changes

Made changes @drvinceknight has suggested

4bc708d

drvinceknight requested changes Sep 28, 2016

View reviewed changes

Made change @drvinceknight has suggested

3ddaa0e

drvinceknight approved these changes Sep 28, 2016

View reviewed changes

drvinceknight added the ready-to-merge label Sep 28, 2016

meatballs merged commit cb4d840 into Axelrod-Python:master Sep 28, 2016

AdamPohl deleted the WorseAndWorse branch September 28, 2016 14:40

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

New Strategy From the PRISON (http://www.lifl.fr/IPD/ipd.frame.html) #724

New Strategy From the PRISON (http://www.lifl.fr/IPD/ipd.frame.html) #724

AdamPohl commented Sep 25, 2016 •

edited

Loading

AdamPohl commented Sep 25, 2016 •

edited

Loading

drvinceknight commented Sep 26, 2016

drvinceknight commented Sep 26, 2016 •

edited

Loading

AdamPohl commented Sep 26, 2016

drvinceknight left a comment

drvinceknight Sep 27, 2016

AdamPohl Sep 27, 2016

drvinceknight Sep 27, 2016

AdamPohl Sep 28, 2016

drvinceknight Sep 27, 2016

drvinceknight Sep 27, 2016

drvinceknight Sep 27, 2016 •

edited

Loading

drvinceknight Sep 27, 2016

drvinceknight Sep 27, 2016

AdamPohl Sep 27, 2016

drvinceknight Sep 27, 2016

drvinceknight Sep 27, 2016 •

edited

Loading

drvinceknight left a comment

drvinceknight Sep 28, 2016

AdamPohl Sep 28, 2016 •

edited

Loading

drvinceknight Sep 28, 2016 •

edited

Loading

drvinceknight Sep 28, 2016

AdamPohl Sep 28, 2016

drvinceknight left a comment

AdamPohl commented Sep 28, 2016

New Strategy From the PRISON (http://www.lifl.fr/IPD/ipd.frame.html) #724

New Strategy From the PRISON (http://www.lifl.fr/IPD/ipd.frame.html) #724

Conversation

AdamPohl commented Sep 25, 2016 • edited Loading

AdamPohl commented Sep 25, 2016 • edited Loading

drvinceknight commented Sep 26, 2016

drvinceknight commented Sep 26, 2016 • edited Loading

AdamPohl commented Sep 26, 2016

drvinceknight left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

drvinceknight Sep 27, 2016 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

drvinceknight Sep 27, 2016 • edited Loading

Choose a reason for hiding this comment

drvinceknight left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

AdamPohl Sep 28, 2016 • edited Loading

Choose a reason for hiding this comment

drvinceknight Sep 28, 2016 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

drvinceknight left a comment

Choose a reason for hiding this comment

AdamPohl commented Sep 28, 2016

AdamPohl commented Sep 25, 2016 •

edited

Loading

AdamPohl commented Sep 25, 2016 •

edited

Loading

drvinceknight commented Sep 26, 2016 •

edited

Loading

drvinceknight Sep 27, 2016 •

edited

Loading

drvinceknight Sep 27, 2016 •

edited

Loading

AdamPohl Sep 28, 2016 •

edited

Loading

drvinceknight Sep 28, 2016 •

edited

Loading