D16492: Rewrite current bargame algorithm to use a generalized version of nim agorithm

[Johnny Jazeix]

jjazeix added a comment.


  In D16492#351080 <https://phabricator.kde.org/D16492#351080>, @AkshayCHD wrote:
  
  > As per the documents and my opinion i think i should just check among the end moves if comp. can win. Ending moves would be random in initial levels and focused on winning in the last levels. Extending the algorithm to the entire pile would be an overhead as we cannot assume the user to play optimally in every move that would be.
  
  
  Yes, the "difficulty" is to find the generic algorithm for "easy", "medium" and "hard" modes (which are bad names btw ;))
  At first level, we can keep a total random, on levels 2 and 3, less random and on level 4, we need to only let the first move of the child decide if he wins or not.

REPOSITORY
  R2 GCompris

REVISION DETAIL
  https://phabricator.kde.org/D16492

To: AkshayCHD, #gcompris_improvements
Cc: jjazeix, kde-edu, #gcompris_improvements, harrymecwan, ganeshredcobra, nityanandkumar, echarruau, rahulyadav, narvaez, scagarwal, apol, timotheegiet, hkaelberer, bcoudoin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kde-edu/attachments/20181030/5b136773/attachment.html>