#717, September 26 - October 2, 2026: The Weekly 'Unsolvable' Sudoku Puzzle
by Richard Kröger, guest compiler
|
WARNING: This is the Weekly 'Unsolvable' Sudoku, rated above 'Extreme'.
This is currently unsolvable by my solver, except perhaps with trial and error strategies.
Using the solver will not help you. (much).
|
Changes March 2026
Due to the introduction of
Forcing Nets many Weekly Unsolvable puzzles can now
be solved. I have flagged all the Sudoku ones which are solvable. I am struggling to make any new ones so we'll
have to see if the Weekly can continue. The point of the Weekly was to provoke discussion and new ideas, so
that is progress. But my solution is unlikely to be optimal so alternatives are welcome.
Archive
Each week a new 'unsolvable' will be published and the previous will be accessible here from this archive section.
If you like very tough puzzles, these are for you.
Share and Enjoy
Supporter Members have access to all these puzzles going back to number 1.
Discussion...
Post an idea here...
... by: Dieter
#717 - initially populated cells: 23 = 28,4 %
No solver used, solver = brain.
basics: H7=4, H1=8
populated cells: 25 = 30,9 %
Strategy: numbers 4,2,5,6,7,8 occur most frequently, combine two cells with almost same candidates.
combinations of D9 = (2,4,6,7) and F9 = (2,4,6,8);
after basic checks 7 matrices left: 2/4, 2/6, 2/8, 4/2, 4/6, 6/4 and 7/2.
D9=4 & F9=6 solve with E6=4.
Optimization:
E6=4 & F9=6 solve.
... by: Norman Powell
Take 2: With the correct puzzle.
First you may be glad to know that AS's solver does solve the incorrect puzzle:
000902000000500003000001040008300900600020000050007010260400070070060005004008000 [NB I missed the first cell in my inital copy]
Finding the expected eliminations and resulting in solution found by Bowman's Bingo [Equivalent to Forcing Networks - Solution Found in my solver]
The solving route is longer, mainly because AS deliberately holds back from providing a solution found by Forcing Networks until all other possible methods are exhausted, whereas my solver looks for it in the Forcing Networks and declares a solution as soon as it is found.
Back to the real #717: This is solved resorting to a full implementation of Bowman's Bingo, which tests value after value through levels of trial values, until either a contradiction is found in the base level and an elimination is made or a solution is discovered in the process. Summary of the solution path:
Hidden Singles 8H1 & 4H7
Forcing Networks 1st Pass: Elimnate 1B3 as forces more than one value ON in D6.
Bowman's Bingo: Eliminate 4F5 after 20 trial values over 6 levels of trials.
Solved Cell: 8F5
Forcing Networks: 2nd Pass - 16 eliminations from values being ON causing contradictions: 1A3, 7B1, 2B3, 7C1, 7C4, 4D1, 3E7, 3E8, 3F3, 1G3, 3G7, 9H3, 1J1, 1J4, 9J4, 3J7
Hidden Single: 3F7
Rectangular Elimination: 1E2 and Strong Link E4, H4 would eliminate all the 1's from Box 7 so eliminate.
Forcing Networks: 6 Eliminations suggested before (not used in solution path so not listed) Solution Found [Equivalent to AS's Bowman's Bingo] while testing 7a9 for ON.
Bowman's Bingo is my method of last resort and is designed to return an elimination or the solution.
... by: Frans Goosens
#717
With trial and error
Combination A4=678 and G9=189
************************************************************
A4=6 G9=1 ( A8=8 ) Solved,
Total solving time is : 8 sec.
Number of logical steps is : 548
... by: Serban
# 717 H1=8 H7=4 Basics to 25
1. A9 = C5 =7 solve
2. G6=5 & G9=1 solve
... by: Norman Powell
Apologies, I seem to have an error in copying the puzzle: I solved:
000902000000500003000001040008300900600020000050007010260400070070060005004008000
in the above.
... by: Norman Powell
Apologies, I seem to have an error in copying the puzzle: I solved:
000902000000500003000001040008300900600020000050007010260400070070060005004008000
in the above.
... by: Norman Powell
AS's Solver only finds Hidden Singles 8H1 & 4H7.
My Solver:
Confined to Box: Value 3 confined to Column 5 and Box 2, eliminate 3G5 & 3J5.
[AS's Solver should have picked this up as a Pointing Pair?]
Alternative Inference Chain [Placed before Forcing Networks in my method order]:
3J1 has weak links to both ends of chain, which would cause a contradiction so eliminate.
Chain: -1H3+1H4-2H4+2J4-7J4+7J5-5J5+5J1-
Grouped X-Cycles [Placed after AIC in my methods order]:
1A3 & 1B3 has weak links to both ends of the following chain: which would cause a contradiction so eliminate:
-1H3+1H4-1E4+1[E2 E3]-1D1+1[A1 B1]-
Forcing Network Pass 1: Finds nothing until testing 9C3 for being ON, when it finds the solution:
547932168186574293329681547718346952693125784452897316265413879871269435934758621
[This is equivalent to AS's Bowman's Bingo]
I am not sure why AS's Solver did not proceed further?