1
$\begingroup$

In one of the asnwers in game theory, I came across the answer that finding a Nash Equilibrium in a bimatrix game has a complexity level of PPAD-complete.

However I couldn't find whether finding the minimax(or maximin for that matter) for a game is a lower level of complexity. Correct me if im wrong, but if this were true, it would make life easier to find the Nash Equilibria for zero sum games computationally 'easier' (since then maximin=minimax=NE in zero sum games)?

The problem is, during my google searches, I found many algorithms that help solve m x n matrix games, however I have not found any algorithms that aim to find the minimax. Does anyone know of these as well?

  • 0
    Minimax is just maximin with negative reward. So unless the PPAD-complete argument uses that the pay-offs are nonnegative, also minimax is PPAD-complete. – 2017-01-10

1 Answers 1

3

The problem of finding maxmin or minmax strategies is (presumably) easier than PPAD. They can be found, and thus zero-sum games can be solved, by Linear Programming, which can be done (provably) in polynomial time, by the Ellipsoid method, or more practically by Interior Point methods (or non-provably but practically efficiently by the Simplex method).