A Python and game theory competition - Page 4 - Computer Technical Help

Anyone actually submitted anything yet?

I've come up with something simple that I think is good for a t-shirt at least, but feels nowhere near as good as I'd have if I actually sat down and did this thing, ha. I've probably spent 10x more time on the sim than I have on actually trying to win the contest.

EDIT: I didn't test against this field before I submitted it but it crushes the field from hehasrisen's last post, haha

Quote

08-11-2013 , 12:19 AM

#77

Hehasrisen

centurion

Join Date: Sep 2011 Posts: 195

Not yet. I did a good amount of work on it in the beginning and then kind of fizzled out unfortunately. I gotta pick back up the pace and just put something together that is submittable.

Damn, that good that you can crush fields you don't even test against haha

Quote

08-11-2013 , 12:22 AM

#78

Xhad

Carpal \'Tunnel

Join Date: Jul 2005 Posts: 10,962

Ironically I discovered a bug and after fixing it my bot does worse. It is still a theoretical favorite against that lineup, but barely. =x

Quote

08-15-2013 , 09:21 PM

#79

Xhad

Carpal \'Tunnel

Join Date: Jul 2005 Posts: 10,962

bug alert:

https://github.com/ChadAMiller/hungergames/issues/36

To be clear, this bug only manifests if you also have a bug in your custom Player code, but thought I'd give a heads-up.

Quote

08-17-2013 , 09:07 AM

#80

Xhad

Carpal \'Tunnel

Join Date: Jul 2005 Posts: 10,962

Oh snap

People have pretty much described my current algorithm in the thread. Guess I'd better try even harder to finish the upgrade I've been pondering.

I'm not the downvote, so I'm guessing at least one other person had the same idea.

https://brilliant.org/discussions/th...t-a3f515668e92

Quote

08-17-2013 , 09:29 PM

#81

maxtower

Carpal \'Tunnel

Join Date: Sep 2005 Posts: 7,996

I am submitted I hope there are enough cooperators for me to win a tshirt.

Quote

08-17-2013 , 09:32 PM

#82

Xhad

Carpal \'Tunnel

Join Date: Jul 2005 Posts: 10,962

I think there's no way there aren't. Everything counterintuitive I've discovered messing with this thing has indicated that people will, if anything, undervalue hunting with low-rep players.

I predict that greedy strategies will die earliest, then collusive strategies that fail to adjust to changing game conditions, followed by some kind of equilibrium that continues to the end of the contest.

Quote

08-18-2013 , 09:10 AM

#83

Sholar

Carpal \'Tunnel

Join Date: Jul 2007 Posts: 6,357

Quote:

Originally Posted by Xhad

Reputation tit-for-tat progressing to individual tit-for-tat? I would guess a lot of people would have this idea...that plus modeling the universe (assuming most people are lazy and likely to be playing a character that you've already dreamed up like "bounded hunters", "fair hunters", etc.) seems like the best approach. But this is a bit of work...

But isn't it better for you for this strategy (rep-tit-for-that) to be widely known and implemented?

Quote

08-18-2013 , 11:52 AM

#84

Xhad

Carpal \'Tunnel

Join Date: Jul 2005 Posts: 10,962

I think the ideal environment for TfT is players who hunt a lot in general but who behave incorrectly against people who aren't you. I can't hope for those if they're also playing my strategy.

I will say that TfT seems like a good fallback if not the best fallback for exploitative strategies capable of noticing when their strategy is failing. Even if I update my bot today, it will have a "doomswitch" that turns it into a multiway TfT bot under certain circumstances.

BTW, those of you who haven't submitted anything have about 8 hours to reverse-engineer a TfT bot from all these coy hints.

Quote

08-18-2013 , 02:18 PM

#85

Xhad

Carpal \'Tunnel

Join Date: Jul 2005 Posts: 10,962

Feh. I must be making an algebra mistake somewhere because my plan B is giving impossible answers. Maybe I'll document all my failed attempts in the hope of winning one of the discretionary prizes, ha

Quote

08-18-2013 , 05:08 PM

#86

Hehasrisen

centurion

Join Date: Sep 2011 Posts: 195

Damn, i don't have time to implement my strategy even though it's not too complicated, and I spent a lot of time thinking about this... The weather is too nice here, it's summer time!

Quote

08-18-2013 , 05:12 PM

#87

noobpoker

grinder

Join Date: Nov 2012 Posts: 574

you have ~7 hours, if you have a good idea it should probably take 1-3 hrs to actually write it out and test

Quote

08-18-2013 , 05:19 PM

#88

Hehasrisen

centurion

Join Date: Sep 2011 Posts: 195

it says i have 2hrs and 40 minutes?

Quote

08-18-2013 , 05:20 PM

#89

Hehasrisen

centurion

Join Date: Sep 2011 Posts: 195

I may just submit a hunter that hunts like 99% of the time and get a T-shirt

Quote

08-18-2013 , 05:34 PM

#90

noobpoker

grinder

Join Date: Nov 2012 Posts: 574

oh wow, didn't realize that it was 2 hrs 30 min from now lol...thought it would be midnight

Quote

08-18-2013 , 05:44 PM

#91

Xhad

Carpal \'Tunnel

Join Date: Jul 2005 Posts: 10,962

You know that quip about "sorry I wrote such a long letter, I didn't have time to write a short one?"

I really should have given myself time to make the documentation less than 2,000 words.

Quote

08-18-2013 , 06:17 PM

#92

noobpoker

grinder

Join Date: Nov 2012 Posts: 574

i have a quick q

so i made an algorithm but i want to make sure it correctly reads through the player_reputations array, how can i do this? every time i tried this by making my own player_reputations array filled with random reputations it gives an error

Quote

08-18-2013 , 06:23 PM

#93

Xhad

Carpal \'Tunnel

Join Date: Jul 2005 Posts: 10,962

that question's a bit too vague for me to even try to answer. There's a bunch of sample code in my simulation at https://github.com/ChadAMiller/hungergames

Quote

08-18-2013 , 07:43 PM

#94

Hehasrisen

centurion

Join Date: Sep 2011 Posts: 195

I just typed up real quick and sloppy what I was going to submit and the reason behind it. Then I said but I didn't realize the competition ended so early this Sunday, so I am submitting this... haha

It's T-shirt time

Quote

08-18-2013 , 07:52 PM

#95

noobpoker

grinder

Join Date: Nov 2012 Posts: 574

Quote:

Originally Posted by Hehasrisen

hah i didnt even bother submitting, got headache trying to think about it, gl you guys!

xhad, just from browsing your github, it would be a crime for you to not win anything for all you have contributed thus far

Quote

08-18-2013 , 07:54 PM

#96

Xhad

Carpal \'Tunnel

Join Date: Jul 2005 Posts: 10,962

Funny, I did something similar:

Quote:

Originally Posted by My Documentation

Toward that end, I ended up writing this really nifty implementation of a Mole-like strategy with a more complicated model accepting arbitrary reputations from opponents reduced to a set of linear equations in numpy with the fatal flaw that it has a bug and doesn't actually work.

Quote

08-18-2013 , 08:03 PM

#97

Hehasrisen

centurion

Join Date: Sep 2011 Posts: 195

Quote:

Originally Posted by noobpoker

xhad, just from browsing your github, it would be a crime for you to not win anything for all you have contributed thus far

Quote

08-18-2013 , 08:04 PM

#98

Hehasrisen

centurion

Join Date: Sep 2011 Posts: 195

Shoot they just extended it for 7 more hrs. Now what is my excuse ha

Quote

08-18-2013 , 08:05 PM

#99

noobpoker

grinder

Join Date: Nov 2012 Posts: 574

LOL they extended it 7 hours

time to make this work!!!

edit: meh, breaking bad ftw gg gl everyone

Last edited by noobpoker; 08-18-2013 at 08:26 PM.

Quote

08-18-2013 , 08:05 PM

#100

Xhad

Carpal \'Tunnel

Join Date: Jul 2005 Posts: 10,962

Quote:

Originally Posted by Hehasrisen

Shoot they just extended it for 7 more hrs. Now what is my excuse ha

argh, agreed on both counts

prob gonna pretend I didn't see this post

Quote

Page 4 of 6

First

1 2 3 4 5 6

Last

Post Reply Subscribe

...

Page 4 of 6

First

1 2 3 4 5 6

Last