
Re: Killer Sudoku 21 May 2021
That puzzle (link will expire) does seem quite difficult. While I can't offer a full analytical solution, here are some of the observations that helped me solve it. (I
did use "pencil" numbers for this one
![Smile [smile]](./images/smilies/msp_smile.gif)
)
(A) The sum of the numbers in cells f2 and f3 is 5, so those two cells must contain either [1,4] or [2,3]. (In the former case, 4 must be in f3 because the 10 cage in that quadrant can't contain [1,4,5] while 1 is also in f3.)
(B) In column d, the sum of the numbers in the four cells outside the 15 and 19 cages is 45 - 15 - 19 = 11, which means that those cells must contain [1,2,3,5].
(C) If the 10 cage in cells c3, c4 and d4 contains the number 1, it must be in d4 (because the other cells are "covered" by the 1 in c5). If not, the numbers in that cage must be [2,3,5].
(D) In the upper left quadrant, the sum of all the cages that are wholly or partially in the quadrant is 10 + 11 + 12 + 13 + 14 = 60, of which 45 is inside the quadrant and 15 in the combined four "outie" cells of the cages 10, 13 and 14 (cells a4, c4, d4 and d1, of which the latter two are among the four cells mentioned in (B).
(E) I tried to determine where the number 9 could go within the same quadrant. Its upper row is "covered" by the 9 in i1, and c3 obviously can't contain 9 either. When I placed 9 in a2 or c2, there was no correct way to complete rows 1 and 2. That left the 11 cage (cells b2 and b3) and cell a3 as the only possible positions of 9 within that quadrant.
The latter two observations made me think that the rather small sum of the four "outies" combined with the knowledge that
a4 had to contain a comparatively high number (
5,
6,
8 or
9) might be enough of a constraint to permit some inferences. I tried putting
9 in that cell, which soon turned out not to permit any correct solution to columns
c and
d (the
21 cage at the bottom).
Keeping my focus on the number
9, I then tried putting it in
a3, with
5 in
a4. Apropos (C), I don't remember if I managed to prove (using columns c and d) that the
10 cage couldn't contain
1 under this hypothesis, or if [2,3,5] in that cage was just a sub-hypothesis I was checking, but at any rate it led to a correct solution to the puzzle.
Even assuming that [2,3,5] in the
10 cage was the result of a valid inference, I would of course still have to check and exclude both variants of [6,8] in the
14 cage in cells
a3 and
a4 (with [2,9] in the
11 cage within the upper left quadrant) to make sure that the solution I'd found was unique.
Working through the branches of trial and error this way after you've found a correct solution may give you a better understanding of the set of conditions/constraints that force the unique solution to the puzzle. You may find (or at least approach) a full analytical solution
![Cool [cool]](./images/smilies/msp_cool.gif)
instead of a
gamebook-style catalog of all the branches (and twigs) of trial and error where all paths execpt one lead to a dead end
![Sad [sad]](./images/smilies/msp_sad.gif)
But I'm trying to cut back on the puzzling so I won't attempt any further analysis of this one
![Smile [smile]](./images/smilies/msp_smile.gif)