If you can read this, either the style sheet didn't load or you have an older browser that doesn't support style sheets. Try clearing your browser cache and refreshing the page.

(Popular Science)   Proof that scientists should never be left alone without adult supervision comes a detailed study on how ants play chess. Fark: the ants found a new solution to a very old chess problem   (popsci.com) divider line 17
    More: Weird, chess problem, the conversation, scientists, University of Nottingham, pheromones  
•       •       •

3484 clicks; posted to Geek » on 04 Feb 2014 at 8:06 AM (34 weeks ago)   |  Favorite    |   share:  Share on Twitter share via Email Share on Facebook   more»



17 Comments   (+0 »)
   
View Voting Results: Smartest and Funniest
 
2014-02-04 08:17:46 AM
"Ants Playing Chess" would be a good name for a band.
 
2014-02-04 08:31:25 AM
Well, simulated ants.
 
2014-02-04 08:38:38 AM

JasonOfOrillia: Well, simulated ants.


And even calling it simulated ants is a bit of a stretch. They used the Ant Colony Optimization algorithm, which is a graph-theory technique for finding paths through graphs. Ant food finding behaviour did inspire its design, so in that sense yeah, simulated ant colony...
 
2014-02-04 08:57:59 AM

entropic_existence: JasonOfOrillia: Well, simulated ants.

And even calling it simulated ants is a bit of a stretch. They used the Ant Colony Optimization algorithm, which is a graph-theory technique for finding paths through graphs. Ant food finding behaviour did inspire its design, so in that sense yeah, simulated ant colony...


Still, it's an interesting application of what we learned from how ants go about their daily lives can be applied to something completely unrelated.  And while using that particular technique to solve an ages old chess problem might not seem to be very practical, it does demonstrate the power of the technique, and perhaps might inspire others to use it optimize more mundane tasks.
 
2014-02-04 09:19:22 AM
Simulated annealing called, it wants it's problem-solving turf back.
 
2014-02-04 09:23:36 AM
They flipped the board over and called the other guy a pussy?
 
2014-02-04 09:27:13 AM

dittybopper: entropic_existence: JasonOfOrillia: Well, simulated ants.

And even calling it simulated ants is a bit of a stretch. They used the Ant Colony Optimization algorithm, which is a graph-theory technique for finding paths through graphs. Ant food finding behaviour did inspire its design, so in that sense yeah, simulated ant colony...

Still, it's an interesting application of what we learned from how ants go about their daily lives can be applied to something completely unrelated.  And while using that particular technique to solve an ages old chess problem might not seem to be very practical, it does demonstrate the power of the technique, and perhaps might inspire others to use it optimize more mundane tasks.


SCIENCE!
 
2014-02-04 09:40:01 AM
I'm still trying to beat the chicken at the fair.

That chicken cheats damn it!
 
2014-02-04 09:52:21 AM
www.pjsmprints.com

Ponder Stibbons is not impressed.
 
2014-02-04 10:08:08 AM

dittybopper: Still, it's an interesting application of what we learned from how ants go about their daily lives can be applied to something completely unrelated.  And while using that particular technique to solve an ages old chess problem might not seem to be very practical, it does demonstrate the power of the technique, and perhaps might inspire others to use it optimize more mundane tasks.


Oh definitely, don't get me wrong. I was just piling on with what appears to be some shoddy reading comprehension on <b>subby</b>'s part, or just a bad headline, suggesting scientists were sitting around studying ants playing chess.

I'm a big fan of optimization algorithms routed in understanding biology. Genetic algorithms, evolutionary algorithms, etc. All are pretty awesome. But I'm also a bioinformatician, so my day to day research is in using computational methods to study biology.

lake_huron: Simulated annealing called, it wants it's problem-solving turf back.


Still has its place in a wide range of applications. It can even be modelled using graph theory with the graph being the search space.
 
2014-02-04 10:22:37 AM
Thants.
 
2014-02-04 10:36:19 AM
I, for one, welcome our new insect overlords...
 
2014-02-04 11:56:57 AM
Remove all the pieces from a chess board except for one knight. Then try to move the knight across all 64 squares of the board, touching each once.

How does the ant colony optimization algorithm solve this? I could see it if you could go over the same spot multiple times, then the shortest path is the correct solution, but as it stands there are no sub-optimal paths. There are only correct paths (100% optimal) and incorrect paths (0% optimal).

Did they remove the "only once" stipulation for the ants?
 
2014-02-04 12:07:52 PM
Anyone who has found the easter egg in Spybot S&D has played it.
http://www.howtogeek.com/79827/the-spybot-search-and-destroy-game/
 
2014-02-04 01:21:20 PM

impaler: Remove all the pieces from a chess board except for one knight. Then try to move the knight across all 64 squares of the board, touching each once.

How does the ant colony optimization algorithm solve this? I could see it if you could go over the same spot multiple times, then the shortest path is the correct solution, but as it stands there are no sub-optimal paths. There are only correct paths (100% optimal) and incorrect paths (0% optimal).

Did they remove the "only once" stipulation for the ants?


Yes, that's the whole point of the algorithm, which you can use to solve similar problems like the Travelling Salesman Problem. The only stipulation you give the algorithm is that all moves must be legal knight moves, and that you have to stay within the board (since nothing except the graph actually exists anyway, you don't need to explicitly stipulate that last rule). The ACO weights edges of the graph on specified criteria, in this case whether or not that edge was travelled in a route that completed a tour. By running millions of searchers it doesn't matter that the vast majority never complete a tour since they are just moving randomly anyway. You build up weights on edges corresponding to knight's tour solutions and reproduce the paths.  It is best in this case to think of specific moves from certain squares as being sub-optimal.
 
2014-02-04 01:32:49 PM

Egoy3k: [www.pjsmprints.com image 217x250]

Ponder Stibbons is not impressed.


Just checking to make sure this was covered.
 
2014-02-04 02:27:11 PM
i189.photobucket.com
 
Displayed 17 of 17 comments

View Voting Results: Smartest and Funniest


This thread is closed to new comments.

Continue Farking
Submit a Link »






Report