41010Re: Statistical analysis of (mostly megaminx) scramblers
- May 1 3:49 PMI have a bit of statistical background. What you have done so far
seems promising, and is well presented, but with how the values jump
lower sometimes when the scramble-length goes up, I think that it's an
indication that one of your inital premise is not as sound as it could
It does discount the variable length-scramble approach though, or at
least I'm convinced.
I wouldn't be suprised to see a few minor jumps in the negative
direction, but I'm concerned with how far it jumps and how often.
I think that the problem is with your original assumption (quoted
below). It even sounds like something a non-cuber, math-geek from
xkcd would come up with - to look at sticker distributions. And
although it would yield rough results that are useful to some extent,
I believe comparing for instance how often every corner piece lands in
every location (CP itself) might yield more evened out results.
Depending on how the programs are written, this could be very hard to
change up. You could have separate charts for CP,EP,CO,EO. (Although
I'm not sure of how to designate orientations on Megaminx.)
Minor point: why is the line for 25-turns missing from the 3x3 chart?
That's the most important one!
--- In firstname.lastname@example.org, "Daniel Hayes"
> The test I conducted was the simplest I could imagine, I applieda
> scrambles to the puzzle and kept track of how often each color landed
> in each position. The assumption: Given a scramble algorithm
> generator that generates an n move scrambling algorithms, if we take
> large number of solved puzzles and apply those algorithms, eachtimes.
> position should end up with each color a roughly equal number of
- << Previous post in topic Next post in topic >>