A few nights ago I saw the movie "The Social Network . Watch it deserves.
At one point, a parallel algorithm to classify girls on the basis of direct clashes hot-or-not . The algorithm is described by the actor who plays Eduardo Saverin in about one femtosecond, and typed on a window in the classic cinematic tradition of Harvard ( the scene here).
I asked what this algorithm and I found googol algorithm ELO. For the curious I explain.
This algorithm is used to rank chess players based on their score , and this score is used to modify according to the clash with players more or less strong. The formula is the basis (written on the window in the film) is:
E_A where is the result you expect the player to score with R_A that collides with a player B with a score R_B. Technical details: in chess a winner-take, who loses and who draws zero 0.5. Let borderline cases:
- If Player A has a score R_A = 1500, playing against a player rated R_B = 0 (total suck) should certainly win (actually tend E_A to one).
- Playing against a player R_B = 1500 (ie equal strength) E_A is exactly 0.5 (tie).
- Finally, if R_B infinity (played excessively loud) E_A tends to zero (failure).
E_A then the score is expected from A collides B.
After a confrontation in which the new score should be in the ranking is given by:
R_A where is the old score, K is a suitably chosen constant that represents the increase in score (in chess for K = 16 teachers and K = 32 for Pippon) s_a the score actually obtained. Example: if A score R_A = 1500 = 1300 R_B and play against us expects the outcome of the match is E_A = 0.76 (using the first formula). But that idiot to get lost and then s_a A = 0. His new score will be (using the second formula):
R'_A = 1500 +16 * (0-0.76) = 1488.
Easy, no?
0 comments:
Post a Comment