Børge wrote:
Glyn wrote:
EDIT.The inter-grid XY-chain in my PM accounts for Samurai #5D and #5E with no further changes required. So that looks a promising, but tricky addition to the code. I don't think you have a bug in there Richard.
I have analyzed both #5D and #5E and agree 100% with Glyn. The step that cracks both puzzles is the inter-grid XY-chain (2=3)r1c5:LLG-(3=7)r1c9:LLG[or r7c3:CG]-(7=9)r7c7-(9=5)r9c7-(5=2)r9c1:CG[or r3c7:LLG] => r1c7:LLG<>2
Until now I had not realized that JSduoku was capable of inter-grid solving.
Very impressive. Hopefully JC will provide some insight into how this is accomplished. My personal guess is that it is a "side effect" of his DLX based Brute Force solver.
No, it isn't related to the DLX based Brute Force solver.
It's simply because JSudoku uses the
whole puzzle for most solving techniques, whereas SudokuSolver probably limit itself to a single sub-grid for some techniques.