The Absolute Optimality Tolerance is a positive value r, indicating to the branch-and-bound solver that it should only search for integer solutions with objective values at least r units better than the best integer solution found so far. In many integer programming models, there are huge numbers of branches with roughly equivalent potential. This tolerance can help to keep the branch-and-bound solver from being distracted by branches that can’t offer a solution significantly better than the incumbent solution.

In general, you shouldn’t have to set this tolerance. Occasionally, particularly on poorly formulated models, you might need to increase this tolerance slightly to improve performance. In most cases, you should experiment with the relative optimality tolerance rather than the absolute optimality tolerance in order to improve performance.

The default value for the absolute optimality tolerance is 0.