I've received a lot of interesting comments and questions from Sudoku fans over the last few years and this page
is where I try to answer them. I'm also directing Str8ts feedback here. Please feel free to drop me a note on the side of the page. Or you can email me directly at andrew@str8ts.com.
the solver fails to eliminate, in cell A9, the 4 base digits 2.5.8.9, who can see the base cells, but it well eliminates the 2 others cell who can see the base cells (C9 and E8)
regards and thanks for your helpfull solver
Andrew Stuart writes:
Very useful to have a good example. Thank you Really, the original documentation on Exocet is so difficult to read I have included only what I can be absolutely sure of and that does not cause false positives. And that is only about a third of what is proposed by Bird. When I go back for another try check this puzzle first
Monday 2-Oct-2023
... by: Adriaan, UK
Thanks Andrew, great website! I understand you create Sudoku's as well. I assume you take a completed grid and then remove numbers/reduce but still making sure it's solvable. How do you do that? Is it trial-and-error, or is there a theorem that tells you whether a sudoku is valid? And for a Killer Sudoku: how do you determine the subdivision that still leads to a valid solution? I've been wondering about that! Maybe you can enlighten me.... or is that a trade secret ;-) Regards, Adriaan
Andrew Stuart writes:
Happy to share. Only trade secret is the code itself but I do like to discuss ideas. Have you seen this? Goes into some excruciating details And thanks for asking as I've just noticed the new version wasn't loaded on the website
Wednesday 27-Sep-2023
... by: Anton Lysogor, Russia
Hi Andrew! Thakns a lot for the best Sudoku I've ever seen. There is a bug after the last update. Single notes (with leading zero) are converted to usual numbers after each page refresh. So I have to make printscreens to complete the puzzles. Please fix it.
Andrew Stuart writes:
Hi Anton. Fixed this today. Should see version 2.19. Lets hope there are no knock on effects.
Sunday 24-Sep-2023
... by: John P Sidenfaden, California USA
Completed the 9/24 Zigsaw puzzle # 5339 correctly. All numbers are filled in but program did not confirm the win (green confirmation) nor display as a loss.
Very strange...
Andrew Stuart writes:
I've been fiddling with the puzzle data in the player and you should refresh until you see 2.18 Let me know if it still doesn't count right tomorrow?
Tuesday 29-Aug-2023
... by: Tuner, USA
I ran into a puzzle that I solved using what I call "double Medusa". I don't use colors, it confuses me, I use letters. A/B instead of colors. When A/B does not solve the puzzle, I go to C/D, E/F if necessary. Today I found a puzzle that I solved when A/B and C/D occupied the same cell in two different places indicating that D could not be correct because it occupied cells containing and A and another with a B. All D's are eliminated. Manual testing confirmed.
Is this new? Did I do good?
Andrew Stuart writes:
I think it is possible for two medusa nets to be isolated from one another. And there might be inferences one could deduce from that, especially as not much board space will be left. Reminds me of this
Thursday 24-Aug-2023
... by: aw, Germany
A strange constant: The Daily Sudokus on Wednesday and Thursday are rated as tough - and I can't solve the Wednesday puzzle and have no problem with the Thursday issue.
Any explanation ... ?
Andrew Stuart writes:
No sure. they are from the same pool, should be random within a grade and no particular selection for a day is made if they are the same grade. Perception when no pattern?
Wednesday 2-Aug-2023
... by: Sudhir, India
Wow!! I input an Expert grade puzzle and tool pointed to the critical clue, the way a guide does.
If you please could check why the grader is giving this soduku a 41 ("Gentle/Easy'). I was cracking my head trying to figure why I was taking so much time on a easy sudoku, but I found the solution uses a Y-wing, so it makes sense now.
Love the solver, it helps a lot.
Thank you
Andrew Stuart writes:
Yes this is a very easy puzzle with one bottleneck. Collapses quickly but the y-wing doesn't boost the score enough. I try and filter such puzzles from my players if I can. But useful for the strategy pages if people want to find those strategies
Monday 10-Jul-2023
... by: bruce amos, Canada brucemamos@icloud.com
Hi Andrew: Many thanks regarding you site. The sudoku methods are really outstanding. This comment is regarding the killer sudoku solver which I applied to a hard problem that I got off the net. The original problem:
I stepped throught the solver until it ran out of known strategies. When the solver got stuck we had candidates: C4:1234568, E3:3456789, E4:3456789 G1:23456789, G2:1234, G3:1234
It is possible to make progress in this position: Firstly, the points C4, E3, E4 total 19 as the total of all cages that intersect the rightmost five columns exceeds 45*9 by 19. This means that E3~3 since otherwise C4 and E4 would total 16 and have to be 7 and 9. This is impossible since C4 cannot be 7 or 9.
Also, the total of G1,G2,G3 must be 10 since the total of all cages that intersect the bottom two rows is 100. Since G2,G3 are both 1,2,3,4 G1 must be 5,6 or 7.
I'm sure that once the solver handles this type of situation it will noticably improved.
Andrew Stuart writes:
Hi Bruce. Yes I like these explanations. I think the solver can be removed if I can identify those overlaps. Thanks for the example
Sunday 2-Jul-2023
... by: John, Ireland
Dear Andrew, I'm a huge fan of yours. Thanks for this fascinating website. I would appreciate it if you could clarify two things for me:
1- How the grading system works? Take a look at these three sudokus please:
First one is dubbed Extreme (score: 546), second one is considered Very Hard (score: 592), and third one is again Extreme (score: 687). How is that possible? Generally speaking, which one should I consider the most reliable? The score, or the Extreme/Very Hard/etc. descriptions?
2- This puzzle (010700400200050070000900050009008002003090600700400300070000000060010900008006030) has been dubbed Extreme Grade, Overall Score: 1199, with 57 steps to solve. However, if you remove 3 from F7, it'll still be standard, Extreme puzzle, but the overall score will reduce to 645, with 42 steps to solve. How is that possible? How come removing a clue makes the puzzle easier?
Basically after I decide where the cut-off points are for grades (based on a statistical review of the scores) I do occasionally promote or demote a puzzle if I think it has too many very hard strategies or too few.
I try and avoid publishing ones right on the border as well but might happen
Thursday 5-Oct-2023
... by: dom Paris, France, France
Load Sudoku: CLICK TO LOADIt's about exocet J4.
the solver fails to eliminate, in cell A9, the 4 base digits 2.5.8.9, who can see the base cells, but it well eliminates the 2 others cell who can see the base cells (C9 and E8)
regards and thanks for your helpfull solver
Really, the original documentation on Exocet is so difficult to read I have included only what I can be absolutely sure of and that does not cause false positives. And that is only about a third of what is proposed by Bird. When I go back for another try check this puzzle first
Monday 2-Oct-2023
... by: Adriaan, UK
I understand you create Sudoku's as well. I assume you take a completed grid and then remove numbers/reduce but still making sure it's solvable. How do you do that? Is it trial-and-error, or is there a theorem that tells you whether a sudoku is valid? And for a Killer Sudoku: how do you determine the subdivision that still leads to a valid solution?
I've been wondering about that!
Maybe you can enlighten me.... or is that a trade secret ;-)
Regards,
Adriaan
Have you seen this?
Goes into some excruciating details
And thanks for asking as I've just noticed the new version wasn't loaded on the website
Wednesday 27-Sep-2023
... by: Anton Lysogor, Russia
Sunday 24-Sep-2023
... by: John P Sidenfaden, California USA
Very strange...
Let me know if it still doesn't count right tomorrow?
Tuesday 29-Aug-2023
... by: Tuner, USA
Is this new? Did I do good?
Thursday 24-Aug-2023
... by: aw, Germany
Any explanation ... ?
Wednesday 2-Aug-2023
... by: Sudhir, India
Wednesday 12-Jul-2023
... by: Victor, Brazil
Load Sudoku: CLICK TO LOADIf you please could check why the grader is giving this soduku a 41 ("Gentle/Easy'). I was cracking my head trying to figure why I was taking so much time on a easy sudoku, but I found the solution uses a Y-wing, so it makes sense now.
Love the solver, it helps a lot.
Thank you
Monday 10-Jul-2023
... by: bruce amos, Canada brucemamos@icloud.com
Many thanks regarding you site. The sudoku methods are really outstanding.
This comment is regarding the killer sudoku solver which I applied to a hard problem that I got off the net.
The original problem:
https://www.sudokuwiki.org/killersudoku.htm?bd=122212221113313341221221441331121112334444422311121122122122133123313311144412221,121100000814000019000008000013001400160018150024000000180000000000000016000033000000000000001900001718000000180800000000001200000017000703001800001600000013000000
I stepped throught the solver until it ran out of known strategies.
When the solver got stuck we had candidates:
C4:1234568, E3:3456789, E4:3456789
G1:23456789, G2:1234, G3:1234
It is possible to make progress in this position:
Firstly, the points C4, E3, E4 total 19 as the total of all cages that intersect the rightmost five columns exceeds 45*9 by 19. This means that E3~3 since otherwise C4 and E4 would total 16 and have to be 7 and 9. This is impossible since C4 cannot be 7 or 9.
Also, the total of G1,G2,G3 must be 10 since the total of all cages that intersect the bottom two rows is 100. Since G2,G3 are both 1,2,3,4 G1 must be 5,6 or 7.
I'm sure that once the solver handles this type of situation it will noticably improved.
Sunday 2-Jul-2023
... by: John, Ireland
I'm a huge fan of yours. Thanks for this fascinating website. I would appreciate it if you could clarify two things for me:
1- How the grading system works? Take a look at these three sudokus please:
920000030706000000000280605000038006000921000500740000803072000000000903060000012
701040906000000050609005000000100309080090020907006000000400703070000000403080105
009700050007050000100306970000500030200070009090004000081407006000020400030001700
First one is dubbed Extreme (score: 546), second one is considered Very Hard (score: 592), and third one is again Extreme (score: 687). How is that possible? Generally speaking, which one should I consider the most reliable? The score, or the Extreme/Very Hard/etc. descriptions?
2- This puzzle (010700400200050070000900050009008002003090600700400300070000000060010900008006030) has been dubbed Extreme Grade, Overall Score: 1199, with 57 steps to solve. However, if you remove 3 from F7, it'll still be standard, Extreme puzzle, but the overall score will reduce to 645, with 42 steps to solve. How is that possible? How come removing a clue makes the puzzle easier?
Thanks for your time! Really appreciate it.
Sorry for the late reply
Have you seen this document?
Basically after I decide where the cut-off points are for grades (based on a statistical review of the scores) I do occasionally promote or demote a puzzle if I think it has too many very hard strategies or too few.
I try and avoid publishing ones right on the border as well but might happen