in reply to Randomly biased, random numbers.
Hmm, I had a similar problem a couple of years ago, trying to test worst case scenarios, although in a totally different context. I don't claim at all to be an expert on this kind of things, but my feeling is that if you want to test worst case scenarios, then you have to bite the bullet and accept that your input data is not going to be random, and there is nothing wrong with that since you are specifically looking for not random situations. Then, of course, the problem is that you don't necessarily know for sure what is the true worst case scenario, sometimes it is actually very difficult to find out. For example, Shell sort and comb sort are known to be fairly efficient sorting algorithms, but they are very rarely used because nobody seems to really know what their worst case scenario could be and it is therefore difficult to assess their worst case complexity.
|