Re^2: Generic De Bruijn Sequence

[LanX]

Is this a guess or do you have a formal proof?

Took me a while to understand why prefer "ones succeeds", and AFAIK only if the alphabet is a power of two.

[tybalt89]

It's a guess.

[QM]

Does "prefer ones" succeed regardless of the alphabet size?

My solution is, I think, a "prefer last" solution. Testing a few basic, low valued, parameters appears to give the correct answer.

Perhaps if someone has a good head for this, they could suggest some parameter values that break under "prefer last" (or indeed, any other method). I don't have the time to tinker, and besides, I'm not that good at poking around the edges.

[Anonymous Monk]

I don't know what this "prefer last" is exactly supposed to mean. But there is such a thing as lexicographically minimal de Bruijn sequence. It becomes lexicographically maximal when you reverse the alphabet. So I'm wondering if you might obtain the desired output by simply remapping the symbols of MinimalDeBruijn($alphabet_size).