Abstract

Sometimes analyzing a poset of keys doesn’t give us a complete picture of what’s happening in a given sorting network. Here, we showed an example of bitonic merging to prove that sometimes analyzing the 0/1-cases is better than analyzing the poset of keys. In our example we considered two different ways to construct exactly the same key poset: the CL lists in Figs. 10.16 and 10.18. It turned out that they both have the same key poset. However, the CL in Fig. 10.16 generated 21 cases whereas the CL in Fig. 10.18 generated 31 cases. This due to the fact Sortnet operates in the reverse direction—it first generates the set of 0/1-cases for a particular situation and then generates the poset of the keys from the set of 0/1-cases using the following rule:

Links and resources

Tags