Browse files

Centred image asset an readme.

  • Loading branch information...
1 parent 4876bdd commit 96e85c1ee1a98c5a7f8d44fd25dae0eb4d733d9d @amcnamara committed Feb 22, 2012
Showing with 1 addition and 1 deletion.
  1. +1 −1
@@ -14,7 +14,7 @@ The strategy is based on the following formula, where:
n_i is the number of words that are a subset of a given sub-branch (character masking) from a given guess
c is a constant for fine-tuning the sub-group size weighting
-<img src="">
+<img src="" style="display: block; margin: 0px auto;">
The idea is to pick a guess character that will return the most possible information (based on returned character maskings, and the possible subset of words that each of those groups can eliminate). There are two ways to attack this problem: the first is to get as even a distribution as possible across the possible sub-maskings (for t=100 and d=4, n\_(1...4)=25,25,25,25 is far more valuable than n\_(1...4)=97,1,1,1), and second is to give preferrential weighting to groups which are more granular (higher d). The summation in the formula above will tend towards zero as the sub-groups tend towards an even and constant distribution, and the weighting on the right will tend towards c as the granularity increases; the guess for which this function is minimal will be the optimal pick.

0 comments on commit 96e85c1

Please sign in to comment.