Message #3725

Subject: Re: Methods for 5+ dimensions
Date: Sun, 18 Jun 2017 17:25:53 +0000

That sounds like exactly what I did. I think a commutator (or possibly alg-based) last cell is the best and easiest way to do things.

Speaking of ZZ^4 (can we use that notation?) I’m actually working on that at the moment, as I felt it would greatly improve the last cell, and also F2L recognition. Once I have completed a solve with it I’ll post my log file and thoughts here. So far my idea is to treat the front, back, top and kata cells as ZZ^3 F and B, and orient the centres to those. I doubt the 3-colour edges could be oriented in a similar way, although I wouldn’t rule it out as being impossible, and if done, I feel that would vastly reduce the time spent on the last cell.

In general though, I feel like some other last cell method could exist, which would be easier than the (n-1)D RKT solve, but faster than the commutator solve, and would prove very useful as the dimensions stack up, although I do feel like it’s quite a way away right now. It seems we are only at the tip of the iceberg of higher dimensional speed methods.

~ Luna