Alpha-Beta NegaMax Search

• A variation on NegaMax search that avoids provably useless subtrees in the search space.
• Alpha is the best possible outcome (initially -1)
• Beta is the worst that can happen (initially 1)
• Modifying NegaMax (successor generation loop):
  • Initial values for Alpha and Beta are taken from the parent node (-1, 1 for the starting node).
  • For each legal move m that results in a recursive call:
    • Generate a successor node via recursion
      • Negate and swap the Alpha-Beta values
    • Let alpha be the maximum of:
      • Current alpha value
      • Most recently returned value
    • If Alpha ≥ Beta, exit the loop.