denis_berthier wrote:logel wrote:From some manual case studies I know that k is at least 5.
I bet you'll find a larger k on a larger collection.
It took really much more time than expected. The reduction of planes is a severe problem. Chasing bugs too.
The answer is 6. (not 42!)
So T&E(S6) solves all known sudoku represented by 817681 puzzles of the current hardest list, all SER 10.3 or more.
S6 stands for pattern with max 6 base lines, i.e. with max 6 true nodes in the pattern.
The S6-pattern used are additionally restricted to max three planes, so its in fact < S6.
If the restriction is on two planes, only 169 are left unsolved.
But I have absolutely no clue how you might compare this to T&E(B7). Looks like apples and oranges.
Re-reading your statements in this thread, I still don't understand why you object to a universal pattern definition.
I put up a new thread proposing such a definition: universal-elimination-pattern
There you also find more details of my results.
I appreciate very much your work on nrczt-chains and braids, but these pattern don't loose their value when embedded in a larger scope.
Any comparison of size or complexity needs a common basis.
Another mystery is how you justify that T&E(Bx) build a strict hierarchy for x=2,3,4,... in whatever sense of simplicity.
The size/complexity of pattern on T&E level one does not limit the size/complexity of the primary elimination.
Some observation (more to do) rather show that smaller patterns on level one increase the complexity of the primary pattern in some cases.
I mistrust the simplest-first strategy to be valid globally.