[r-t] Monster extent of triples

[Ian McCulloch]

On Tue, 7 Oct 2014, Robert Bennett wrote:

>
>
> 	I did say relatively. My point is that the more totally different
> ones that exist, the less likely it is (IMHO) that there will be much
> of the starting structure left over. 
>
> 	At this point, really need more examples based on different starting
> compositions.

If I understand the algorithm correctly, the reversal procedure removes a 
discontinuity at the first position (if there was one) and sometimes 
leaves a discontinuity at the second position (but sometimes it won't, and 
it might even remove a discontinuity).

So given an input that is a random permutation of the extent (so almost 
all changes would be discontinuous) after many iterations it should 
eventually remove all the discontinuities.  The question is, how many 
iterations does it take?  Not having tried it (nor read the RW article - 
it hasn't arrived in OZ yet) it's hard to make a sensible guess, but I 
wouldn't be surprised if it wasn't very many (eg 10x the number of 
original discontinuities).  I also wouldn't be surprised if there are very 
very many such compositions.  Even running a few more iterations starting 
from the 'monster' should produce another different composition.

Cheers,
Ian