07-28-2022, 02:54 PM
In the LanguageReference is stated that if in the (ai ) ludeme a bestAgent is specified, 'this algorithm will be used when the “Ludii AI” option is selected'. ok.
So when I use bestAgent 'Alpha-Beta' with trainedHeuristics ( influence ) then the 'Ludii AI' will have built in ('trained') this style of play already, is that right ?
But what happenes if I choose 'Alpha-Beta' for the second player too?
Will P2 perform an unbiased alphabeta-search or will it also use the heuristics stated in the ai-bestAgent-ludeme ?
My goal would be to test games with Alpha-Beta: heuristics (influence) or (material) against Alpha-Beta without heuristics . Is this possible ?
(to see which one performs better in a series of games.)
I tried to figure this out watching games of Chess (combining the above heuristics with the important pieceWeights seemed to work)
or English Draughts, but it was not clear if the heuristics had any effect to the style of play.
So when I use bestAgent 'Alpha-Beta' with trainedHeuristics ( influence ) then the 'Ludii AI' will have built in ('trained') this style of play already, is that right ?
Code:
(ai (bestAgent "Alpha-Beta")
trainedHeuristics:(heuristics {(influence)})
)
But what happenes if I choose 'Alpha-Beta' for the second player too?
Will P2 perform an unbiased alphabeta-search or will it also use the heuristics stated in the ai-bestAgent-ludeme ?
My goal would be to test games with Alpha-Beta: heuristics (influence) or (material) against Alpha-Beta without heuristics . Is this possible ?
(to see which one performs better in a series of games.)
I tried to figure this out watching games of Chess (combining the above heuristics with the important pieceWeights seemed to work)
or English Draughts, but it was not clear if the heuristics had any effect to the style of play.