denis_berthier wrote:.
Whatever the real proportion, this also shows that there remains numerous cases (15-20%) where the final value in more than one of the tridagon cells is not one of the 3 values of the pattern.
It also shows that searching for tridagons only in solution grids that have a "solution-tridagon" is condemned to miss many interesting cases.
.
In my sample test on a chunk of 600 000 solution grids, I had 10 expected grids matching with mith's file. For 2 of them, I had no match due to the lack of target (tridagon pattern in the solution grid).
I think that all records in mith's file are T&E(3), and I always read that all T&E(3) had this pattern. Looking at SKFR path, I understood why and I have seen that the PM had one of these tridagon pattern with guardians. This is why I open this thread as tridagon?? with one puzzle of mith's file.
In my view, it's difficult to put in the same bag both groups, but it's not a key point.
Finding puzzles of interest for the players is another story. As far as I know, all puzzles of the current T&E(3) file are derived from "Loki", the nugget, by more or less a "vicinity" search.
I have a code to search directly puzzles with the "loki" PM at start hoping to get one day another nugget.
Unhappily, till now, this gives more gold powder and thousand tons of sand to sort.
I have in mind some ideas to get a feasible process, but the result will continue to produce seeds. If one has the potential, then it will be possible to get other sets of puzzles.
This is already complex, and I see no easy way to get out of the loki vicinity that limiting the search to a typical loki pattern.