Minimax tree where a player can move multiple times in a row

Question

Having played around with turn-based games using Minimax and Alpha-Beta Pruining, how would one handle games where the same player can have multiple consecutive moves if some condition is met?

Answer 1

If player A is invoking minimax then:

On Player A's first turn of a run: maximise

On Player A's second turn of a run: maximise again.

This relies of objective evaluations of the board being based from the perspective of A. So regardless of whose turn it is on a state, objective(state) is with reference to A's gain from that state.

You will need a way to determine given a sate X, was there a turn made in the parent state such that the current turn for player Z is the second in a row. However you could also simply evaluate whose turn it is, and if it's A's then perform maximisation as normal.

Answer 2

Isn't it most natural to consider a sequence of moves by one player as one action in the game tree?

Consider a game like Chinese Checkers. You would not consider each jump as a full move. You will consider a full move as the sequence of jumps for one pearl from it's start position to it's end position.

Answer 3

You mean like checkers?

Minimax tree should have all the possible moves. If some condition is met, than i has to be also added like other possibilities. And all intermediate states have to be ignored