Efficient breadth-first manipulation of binary decision diagrams

We propose new techniques for efficient breadth-first iterative manipulation of ROBDDs. Breadth-first iterative ROBDD manipulation can potentially reduce the total elapsed time by multiple orders of magnitude compared to the conventional depth-first recursive algorithms when the memory requirement exceeds the available physical memory. However, the breadth-first manipulation algorithms proposed so far have had a large enough overhead associated with them to make them impractical. Our techniques are geared towards minimizing the overhead without sacrificing the speed up potential. Experimental results indicate considerable success in that regard.