Jump to content

  • Log In with Google      Sign In   
  • Create Account


#Actualpatishi

Posted 13 July 2013 - 10:07 AM

thank you for the reply.    so in general..you are saying that for exhaustive search i can immediately use the score stored from the TT and not check bounds,depth..nothing like that?  

I also don't care so much if the program wins immediately of after 5 moves..as long as it wins!  in fact i already noticed this kind of behaviour in my older engines, i guessed this is because minimax just goes for the first line he sees if the score stays the same (e.g picking the first move).      it actually quite ammusing , like a human player that wants to prove that he can win in other ways too smile.png   twisting  with you mind LOL

For simplicity , right now i would rather go for the simplest approach possible (as few parameters as possible and a less complicated code),  that's why i asked if i can skip the boundaries / depth checking part if i am doing a complete search.

 

So if i want to use the killer moves in an exhaustive search i would still need to use a depth parameter in the alpha beta function (let's say 42..43..to make it get to the end of the tree) ??

 

 

EDIT: By the way, when i am calculating the depth i am starting from  a high number and going down (e.g 8..7..6) until i get to 0.   when saving to the killers table i am using the depth parameter, so it will always be the same ply.  (e.g if the depth was 7 ,it will be 7 also in the killers table.  and only when i am again in that same ply i check for the killer)


#4patishi

Posted 13 July 2013 - 10:05 AM

thank you for the reply.    so in general..you are saying that for exhaustive search i can immediately use the score stored from the TT and not check bounds,depth..nothing like that?  

I also don't care so much if the program wins immediately of after 5 moves..as long as it wins!  in fact i already noticed this kind of behaviour in my older engines, i guessed this is because minimax just goes for the first line he sees if the score stays the same (e.g picking the first move).      it actually quite ammusing , like a human player that wants to prove that he can win in other ways too smile.png   twisting  with you mind LOL

For simplicity , right now i would rather go for the simplest approach possible (as few parameters as possible and a less complicated code),  that's why i asked if i can skip the boundaries / depth checking part if i am doing a complete search.

 

So if i want to use the killer moves in an exhaustive search i would still need to use a depth parameter in the alpha beta function (let's say 42..43..to make it get to the end of the tree) ??

 

 

EDIT: By the way, when i am calculating the depth i am starting from  a high number and going down (e.g 8..7..6) until i get to 0.   when saving to the killers table i am saving by depth number, so it will always be the same ply.


#3patishi

Posted 13 July 2013 - 07:37 AM

thank you for the reply.    so in general..you are saying that for exhaustive search i can immediately use the score stored from the TT and not check bounds,depth..nothing like that?  

I also don't care so much if the program wins immediately of after 5 moves..as long as it wins!  in fact i already noticed this kind of behaviour in my older engines, i guessed this is because minimax just goes for the first line he sees if the score stays the same (e.g picking the first move).      it actually quite ammusing , like a human player that wants to prove that he can win in other ways too smile.png   twisting  with you mind LOL

For simplicity , right now i would rather go for the simplest approach possible (as few parameters as possible and a less complicated code),  that's why i asked if i can skip the boundaries / depth checking part if i am doing a complete search.

 

So if i want to use the killer moves in an exhaustive search i would still need to use a depth parameter in the alpha beta function (let's say 42..43..to make it get to the end of the tree) ??


#2patishi

Posted 13 July 2013 - 07:36 AM

thank you for the reply.    so in general..you are saying that for exhaustive search i can immediately use the score stored from the TT and not check bounds,depth..nothing like that?  

I also don't care so much if the program wins immediately of after 5 moves..as long as it wins!  in fact i already noticed this kind of behaviour in my older engines, i guessed this is because minimax just goes for the first line he sees if the score stays the same (e.g picking the first move).      it actually quite ammusing , like a human player that wants to prove that he can wins in other ways too :)

For simplicity , right now i would rather go for the simplest approach possible (as few parameters as possible and a less complicated code),  that's why i asked if i can skip the boundaries / depth checking part if i am doing a complete search.

 

So if i want to use the killer moves in an exhaustive search i would still need to use a depth parameter in the alpha beta function (let's say 42..43..to make it get to the end of the tree) ??


#1patishi

Posted 13 July 2013 - 07:34 AM

thank you for the reply.    so in general..you are saying that for exhaustive search i can immediately use the score stored from the TT and not check bounds,depth..nothing like that?  

I also don't care so much if the program wins immediately of after 5 moves..as long as it wins!  in fact i already noticed this kind of behaviour in my older engines, i guessed this is because minimax just goes for the first line he sees if the score stays the same (e.g picking the first move).    

For simplicity , right now i would rather go for the simplest approach possible (as few parameters as possible and a less complicated code),  that's why i asked if i can skip the boundaries / depth checking part if i am doing a complete search.

 

So if i want to use the killer moves in an exhaustive search i would still need to use a depth parameter in the alpha beta function (let's say 42..43..to make it get to the end of the tree) ??


PARTNERS