PISA Results

PISA (Problem Instance Simulated Annealing) uses simulated annealing to find problem instances where one scheduling algorithm performs much worse than another.

Each cell in the heatmap shows the worst-case makespan ratio found by PISA for row algorithm vs. column algorithm. A value of 3.0 means PISA found an instance where the row algorithm's makespan is 3x the column algorithm's makespan.

Key finding: For every algorithm, PISA finds instances where it performs at least 2x worse than another algorithm. No single algorithm dominates.