TidemanAndChieruzzi strategy #1118

dmanc · 2017-08-13T09:38:08Z

This is the implementation of Tideman and Chieruzzi's strategy from Axelrod's first tournament.
Resolves #1105.

drvinceknight

Thanks for this: a quick scan looks good! I'll review properly tomorrow.

The tests on travis are failing because of the dosctring formatting.

drvinceknight · 2017-08-13T16:18:17Z

axelrod/strategies/axelrod_first.py

+    1. Every run of defections played by the opponent increases the number of
+    defections that this strategy retaliates with by 1.
+    2. The opponent is given a ‘fresh start’ if:
+        - it is 10 points behind this strategy


Need a space between this line and 2. The opponent is

drvinceknight · 2017-08-13T16:18:48Z

axelrod/strategies/axelrod_first.py

+        - and it has been at least 20 rounds since the last ‘fresh start’
+        - and there are more than 10 rounds remaining in the tournament
+        - and the total number of defections differs from a 50-50 random sample
+        by at least 3.0 standard deviations.


This needs to be at the same level of indentation as the previous line:

+ - and the total number of defections differs from a 50-50 random sample by at least 3.0 standard deviations.

drvinceknight

This is looking fantastic and my requests are minor doc changes (you've understandably grabbed what was in the issue but I perhaps should have written that a tiny bit better) and a few more tests please.

drvinceknight · 2017-08-14T07:59:52Z

axelrod/strategies/axelrod_first.py

+
+class TidemanAndChieruzzi(Player):
+    """
+    This strategy begins by playing Tit For Tat and then things get slightly


Change and then things get slightly complicated to and then follows the following rules

drvinceknight · 2017-08-14T08:14:33Z

axelrod/strategies/axelrod_first.py

+        - it is 10 points behind this strategy
+        - and it has not just started a run of defections
+        - and it has been at least 20 rounds since the last ‘fresh start’
+        - and there are more than 10 rounds remaining in the tournament


in the match

drvinceknight · 2017-08-14T08:19:42Z

axelrod/tests/strategies/test_axelrod_first.py

+                                'retaliation_length': 4,
+                                'retaliation_remaining': 1})
+
+


To be a slight pain (this strategy really is excellently written), could we add a couple more tests please:

Check the standard deviation condition (test similar actions where one is and one isn't within bounds)

Check the self.fresh_start attribute

In the description of the strategy

and the total number of defections differs from a 50-50 random sample by at least 3.0 standard deviations.

I implemented this as if self.defections <= lower or self.defections >= upper:
But I am now question what "total defections" means, would it be self.defections + opponent.defections?

I agree that the wording is not completely clear but I am 99.9% sure that the way you have interpreted/implemented it is correct.

It makes sense to check the behaviour of the opponent.

drvinceknight · 2017-08-15T06:10:57Z

This looks great. Thanks 👍

One final request from me (sorry I should have realised this earlier). Can you modify https://github.com/Axelrod-Python/Axelrod/blob/master/docs/reference/overview_of_strategies.rst to show that this is now implemented please.

drvinceknight · 2017-08-15T10:00:09Z

Thanks this looks good to me 👍

marcharper · 2017-08-16T00:52:14Z

axelrod/strategies/axelrod_first.py

+
+        if valid_fresh_start:
+            valid_points = self.current_score - self.opponent_score >= 10
+            valid_rounds = self.match_attributes['length'] - current_round >= 10


What will happen if length is not supplied?

I did not consider this but I just tried the strategy in a probabilistic ending tournament and self.match_attributes['length'] evaluated to inf. When would the length not be supplied?

You have dealt with this the correct way. When a length is not supplied the match tells the players that it's infinite (we have a few players that do this). 👍

Could we add a test where we don't tell the match length to the players?

Similar to the test for Alexei in this file: https://github.com/Axelrod-Python/Axelrod/blob/master/axelrod/tests/strategies/test_titfortat.py

Except go for match_attributes={"length": float('inf')}, that's essentially equivalent to -1 but more readable (~~I'll probably open a PR to that test file and change it~~ #1120).

Wouldn't the behavior be different if I use -1?
valid_rounds = self.match_attributes['length'] - current_round >= 10 would be true if length was float('inf') and false if length was -1.

Yup you're correct. float('inf') is the correct choice here: in essence if match length is not known then that corresponds to an infinite match length from the point of view of the strategy.

In this case: are we within ten turns of the end? Probably not: we don't know when the end is. So it's essentially going to assume there is always more than 10 rounds remaining.

Alright thanks, i'll make a quick change to the test.

drvinceknight

Just the one extra request for a test with match_attributes={"length": float('inf')} please (ideally showing a change in behaviour).

drvinceknight · 2017-08-16T08:06:30Z

axelrod/tests/strategies/test_axelrod_first.py

+        # Similar to the last test except the length is not given.
+        opponent = axelrod.Cycler('DDCDD')
+        self.versus_test(opponent, expected_actions=actions,
+                         match_attributes={'length': float('inf')})


Is there a way to devise a test where the play plays differently when the match length is not given? Or even if the actions are the same, are the attrs different?

Yes, I can change this test so with 15 rounds of play this strategy will give a fresh start on the last two rounds when the length is not given. Otherwise if the length was 15 it would defect on the last two rounds.

drvinceknight · 2017-08-16T10:50:26Z

Perfect! Thanks 👍

TidemanAndChieruzzi strategy

5169a25

drvinceknight requested changes Aug 13, 2017

View reviewed changes

Fix indentation

0edb3cf

drvinceknight requested changes Aug 14, 2017

View reviewed changes

Fix documentation and add some test

f2bb878

dmanc added 2 commits August 15, 2017 03:56

Modify overview_of_strategies.rst

ad4663b

Fix typo

c62be11

drvinceknight approved these changes Aug 15, 2017

View reviewed changes

drvinceknight added the ready-to-merge label Aug 15, 2017

marcharper reviewed Aug 16, 2017

View reviewed changes

marcharper approved these changes Aug 16, 2017

View reviewed changes

drvinceknight removed the ready-to-merge label Aug 16, 2017

drvinceknight requested changes Aug 16, 2017

View reviewed changes

dmanc added 2 commits August 16, 2017 01:32

Add test for length

6ecd99e

Fix test

949ddfc

drvinceknight requested changes Aug 16, 2017

View reviewed changes

Test length conditions

447d0c5

drvinceknight approved these changes Aug 16, 2017

View reviewed changes

drvinceknight merged commit dd0bb42 into Axelrod-Python:master Aug 16, 2017

dmanc deleted the TidemanAndChieruzzi branch August 17, 2017 03:33

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

TidemanAndChieruzzi strategy #1118

TidemanAndChieruzzi strategy #1118

dmanc commented Aug 13, 2017

drvinceknight left a comment

drvinceknight Aug 13, 2017

drvinceknight Aug 13, 2017

drvinceknight left a comment

drvinceknight Aug 14, 2017

drvinceknight Aug 14, 2017

drvinceknight Aug 14, 2017

dmanc Aug 14, 2017

drvinceknight Aug 14, 2017

drvinceknight commented Aug 15, 2017

drvinceknight commented Aug 15, 2017

marcharper Aug 16, 2017

dmanc Aug 16, 2017

drvinceknight Aug 16, 2017

drvinceknight Aug 16, 2017

drvinceknight Aug 16, 2017 •

edited

dmanc Aug 16, 2017

drvinceknight Aug 16, 2017 •

edited

dmanc Aug 16, 2017

drvinceknight left a comment

drvinceknight Aug 16, 2017

dmanc Aug 16, 2017

drvinceknight Aug 16, 2017

drvinceknight commented Aug 16, 2017

TidemanAndChieruzzi strategy #1118

TidemanAndChieruzzi strategy #1118

Conversation

dmanc commented Aug 13, 2017

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

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

drvinceknight commented Aug 15, 2017

drvinceknight commented Aug 15, 2017

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 Aug 16, 2017 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

drvinceknight Aug 16, 2017 • edited

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

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

drvinceknight commented Aug 16, 2017

drvinceknight Aug 16, 2017 •

edited

drvinceknight Aug 16, 2017 •

edited