The New Sudoku Players' Forum

[Hajime]

Grid 3, locked singles 1r6c12 -> -1r5c12,-1r7c1

First some confusion: There are no candidates 1 in r5c12 of grid 3 and why should this lead to -1r7c1 ???

Grid 3, locked singles 1r6c12 -> ...

That is sure and because the cells g3r6c12 are the same as g4r1c67 they are still locked single cells for candidate "1".
And therefore all other candidates 1 in r1 of grid 4 can be eliminated: grid 4 -1r1c15. Agreed.
And also -1r9c6 because both locked cells can be seen by r9c6 (same column or same jigsaw-house).
And r2c7<>1 (same column, same jigsaw-house).

What I do is Join/Intersect all houses PER GRID. Grid 3, locked singles 1r6c12 does not lead to eliminations in grid 3.
In the view of grid 4 alone, r1c67 (same cells) are not locked anymore for candidate 1. Here is the omission. I see your point.

'Generalized Intersection' but not yet in code.

The code will be published as soon as SiSeSuSo is stable and all methods are programmed using "Generalized" code, not only for intersections or kites, but for all implemented methods.

[creint]

-1r5c12 was already excluded in your grid, so a bit confusing.

Both digits from 1g3r6c12 are seeing r7c1
1g3r6c1=1g4r1c6 is in diagonal \ from grid 4
1g3r6c2 is in column 2 of grid 2.

But you already got that.

Your solver is grid-based and because of that you sometimes need more strategies to complete all of them.
Sometimes it is necessary to view them as one grid to find strategies.

[Hajime]

But you already got that.

Sure... grid transcending intersections or locked singles.
Is that also possible for kites, skyscrapers, fish, x- or xy-chains or any other AIC or method. Long way to go...