01-07-2021, 01:05 PM
(This post was last modified: 01-08-2021, 12:48 AM by dale walton.
Edit Reason: Might have a solution -would it work?
)
Yes, to keep track of whether a piece might be able to move there over the course of a series of subsequential movements and then selecting move to sites from among them at the end, rather than by selecting them as a series of independent decisions.
It would make the game play easier for both the humans and the AI to be able to move by turn rather than by specifying an exact path of steps.
Otherwise I would need an array or dynamic mapping to store the info. i.e. to store a dynamic list of sites and values associated to them. But from your response it looks like I need to think of another method. - Or for now stick with sequential "hops".
I noticed on some multi-step moves that the AI uses N seconds per move, making play slow, but a game player in a real life game not on Ludii normally gets time per turn. Also the look-ahead must be degraded when possible turn outcomes are not consolidated to a set to choose from before selection, especially if there are a lot of overlapping possibilities, as they artificially expand the branching factor.
Edit: OK, Maybe I can (move Select the starting position of the move, then use hidden neutral pieces as place markers (like your show moves spots) and then (move Select one of them and remove them all as a separate move to complete the turn.
It would make the game play easier for both the humans and the AI to be able to move by turn rather than by specifying an exact path of steps.
Otherwise I would need an array or dynamic mapping to store the info. i.e. to store a dynamic list of sites and values associated to them. But from your response it looks like I need to think of another method. - Or for now stick with sequential "hops".
I noticed on some multi-step moves that the AI uses N seconds per move, making play slow, but a game player in a real life game not on Ludii normally gets time per turn. Also the look-ahead must be degraded when possible turn outcomes are not consolidated to a set to choose from before selection, especially if there are a lot of overlapping possibilities, as they artificially expand the branching factor.
Edit: OK, Maybe I can (move Select the starting position of the move, then use hidden neutral pieces as place markers (like your show moves spots) and then (move Select one of them and remove them all as a separate move to complete the turn.