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.
Post a Comment or Question here...
Monday 27-Dec-2021
... by: Ken Gregory, Alberta, Canada
The checkmark at Exocet doesn't work. The solver used Exocet strategy to solve a puzzle. I removed the checkmark next to Exocet, reloaded the puzzle to see how the solver would solve it without using Exocet, but it did used Exocet anyway.
Andrew Stuart writes:
I think I fixed it. Was swapped internally with APE above it Do try again
Friday 17-Dec-2021
... by: DougCube, California
The button for "Email This Board" gives "This board is empty!" for puzzles where many candidates have been removed but no single cell is solved. (There are such configurations that are solvable and have unique solution.)
The expectation is to still get the state of the board in the text format.
Andrew Stuart writes:
Something wrong with some cells. They have to contain at least one candidate or clue.
Try having " Auto Clear" ticked before entering clues
Otherwise, if you have cleared the board and entered some clues, press take step to get the candidates. I guess it's possible that some clue entries will leave a row or column with nothing in a cell before "take step"
Wednesday 3-Nov-2021
... by: Allan Patrick, United States
In "Unique Rectangles" it confused me when under subheading "Type 2 Unique Rectangles", second to last sentence you say "8's on brown cells are eliminated." which suggests to me that the 8's in H4 and H6 can be eliminated when it's actually the 8's highlighted in yellow that can be eliminated. Can this be made clearer? Thanks for this website.
Your solver correctly describes this as a BUG Sudoku and gives a solution with D1=6; D3=1; F3=5; D6=5 ;F6=1. This cannot be correct because it allows for an alternate solution with D1=6; D3=5; F3=1; D6=1; F6=5.
The correct solution is based on D3=6 chosen to eliminate the possibility of an alternate solution.
Andrew Stuart writes:
There seem to be three solutions to this puzzle. Are you sure you entered all the clues correctly or maybe the puzzle is faulty? There is no "correct" solution since there isn't a single solution. BUG is among a group of strategies that specifically rely on there being a single solution. But all logical strategies will be deceived ultimately.
Monday 25-Oct-2021
... by: Larry Zwahlen, Canton, GA, USA
The current jigsaw sudoku for 10/25/2021 has an error in it. A 3 has to be in row A of the yellow piece and a 3 also has to be in row B of the yellow piece. How can this be.
(I notice the player and solver have different colours so I've synchronized them)
Thursday 21-Oct-2021
... by: jamsaw, USA
On the home page, I noticed you are using different terms for cells. Sometimes they are called "cells" and other times "squares". The problem is, other sites have different names for units...some call a 3x3 unit "box", some call it "square", some call it "region". So, when reading the instructions or analysis, it become confusing if you have ever been exposed to other Sudoku sites. Is there any chance you could describe your terminology at the beginning of the explanations? I wish there were a common terminology where everyone had the same unit names in mind.
Also, where are you, where J follows H in the alphabet?
Andrew Stuart writes:
I agree there should be consistency. I think where the "square" is used it will be very early documentation and I quickly changed to "cell" as and when I spot square I'll change it. Generally though the Sudoku community came up with all manner of terms for this and that and it's been a problem from the start. Emergent divergence. Not just cells/squares
Also very early on people decided that "I" was too close to "1" for a row name and therefore J follows H for row names. This was generally agreed to be a good idea as coordinates need to be very clear. A large fraction continue to use the cumbersome rXcY format
First of all, I just want to say thanks for providing such an amazing piece of software and resources, it's been invaluable to me.
I believe I've found a bit of an inconsistency with the way the solver applies (at least) hidden singles in the solve path. In the sudoku provided, step 10 lists the following hidden singles:
4 found once at r1c2 in row, 4 candidates removed 6 found once at r1c1 in row, 1 candidate removed 8 found once at r9c5 in column/box, 1 candidate removed
The problem is with r1c1: the 6 is only found once in the row because it has been eliminated from r1c2 in the same round. This elimination is therefore not deducible from the state of the board at step 10.
Stepping through the steps manually on the solver separates these steps.
Maybe there is a reason the solve path wants to group these steps, but it seems bizarre to me.
Thanks again
Andrew Stuart writes:
The solver does it correctly - storing eliminations during the scan and then applying them at the end. knock on effects need to be separated for clarity. It is done in javascript. The Solve Path uses the compiled c/c++ engine to derive a solution which applies the eliminations immediately which is why 4 and effect 6 but 6 cannot effect 4 since it is done in number order. This didn't seem important when I wrote it since all the other strategies were the interesting ones and this was a process for clearing off candidates from singles to get to the meat. But I should address it in the next update.
If you are comparing a puzzle with a score 68 to another with a score of 72 I wouldn't say one was more difficult than the other since there are a host of subjective decisions (exact points for any strategy eg) but 68 compared to 300 or 3000 is significant. Its for the bulk division of all puzzles into grades but the dividing line is arbitrary. It's like a bell chart of human heights. If you want to divide people into short and tall the division will separate people with millimetre height differences around the average
Saturday 21-Aug-2021
... by: Frank Merritt, Illinois, USA
I believe you have a minor bug involving the labeling of shapes in the explanation of "Double Line / Box Reduction". You say "Down the center are two columns where 2 is present in shapes 4 and 5 -- and just those 2 shapes." But this is for "Andrew Stuart 7", and according to the labeling in the lower right corner of the Solver page for this configuration, those shapes are 1 and 2, not 4 and 5. I have never understood the labeling of shapes in Jigsaw configurations. Judging from the labels for Andrew Stuart 7 and for Cabbage, I believe they are labeled by whim rather than by any systematic procedure. Is that correct? If not, is the rule explained anywhere in your pages? P.S. Thanks very much for such an incredibly useful reference!
Andrew Stuart writes:
Correct, I've fixed the shape numbers in that sentence - many thanks!
Now I believe I have tried to assign the numbers in a top-left to bottom-right order but not every shape has that determination I can see I might go back and renumber, next update
Monday 16-Aug-2021
... by: Theo, Switzerland
I just love your Sudoku Solver. Thanks a lot !
One small question: I have edited off some possibles in any Sudoku. With the "auto clear" button on, when I enter a solution, the action taken by the solver is not to logically clear out just the values that have to be cleared in that box, row and column because of this new number, but it resets ALL possibilities back. Meaning, my edits are discarded, the possibilities I edited out earlier are set back again. Could there be an option to make a real clearing, not a reset to all possibilities?
This sounds so easy, am I missing something?
Best regards, Theo
Andrew Stuart writes:
Yeah I get you but you're asking for an "intelligent" auto-clear that would have to keep track of which candidates you wanted to keep since you entered them and which were derived from the clues. That would be too exhausting and also lead to contradictions in case the solver was forced to adopt a candidate as existing when it had previously removed it. I know people use the solver as a player. Best to have auto clear on for the clue entry and off other times.
Tuesday 10-Aug-2021
... by: brent hudson, United Kingdom
Hi Andrew,
Firstly, what a brilliant programme! I have used it a lot over the last 2 years as I've tried to improve my own solving skills. It has helped me improve a lot.
However, on this board I have been able to solve a puzzle that the solver cannot, and I would expect it to be able to. I am wondering why it cannot.
Here's the board:
There is a Killer Sudoku board I would like you to look at
Click on this link: https://www.sudokuwiki.org/killersudoku.htm?bd=121112211121133441112213122132112111332312322111311321221213311133221121114411121,300622000012002100000000001600040000000025003200191400000600000013000000000000130000151200210000000000000022130000210000000000180500000027001000000008000000000000
When it gets stuck, cells G6, H6, J5, and J6 sum to 13 (by summing columns 7, 8, and 9 and looking at the spill over). 13 in 4 cells in the same block has to contain 1. This means that J4 has to be 2 (the solver gets stuck with J4 being only 1 or 2).
If I set J4 to 2, the solver then solves the puzzle.
Why doesn't the solver see that J4 has to be 2?
I ask for two reasons:
1. The solver seems to solve much harder situations than this one. 2. If you do decide to update the solver again, maybe there's a piece of logic you could insert?
Again - thanks for doing the solver; it's a great tool!
Brent
Andrew Stuart writes:
Yes the Killer solver does lag behind the Sudoku in terms of solving power. Mainly because there are many more candidates and the chaining strategies can't bite. However I like your example below, gives me a clue to improve. Just have to figure out where exactly.
Monday 27-Dec-2021
... by: Ken Gregory, Alberta, Canada
Do try again
Friday 17-Dec-2021
... by: DougCube, California
The expectation is to still get the state of the board in the text format.
Try having " Auto Clear" ticked before entering clues
Otherwise, if you have cleared the board and entered some clues, press take step to get the candidates. I guess it's possible that some clue entries will leave a row or column with nothing in a cell before "take step"
Wednesday 3-Nov-2021
... by: Allan Patrick, United States
Wednesday 27-Oct-2021
... by: Carl, Florida
Load Sudoku: CLICK TO LOADThe correct solution is based on D3=6 chosen to eliminate the possibility of an alternate solution.
There is no "correct" solution since there isn't a single solution. BUG is among a group of strategies that specifically rely on there being a single solution. But all logical strategies will be deceived ultimately.
Monday 25-Oct-2021
... by: Larry Zwahlen, Canton, GA, USA
Sorry for the late reply
Try the solver to see how the solutions are derived
(I notice the player and solver have different colours so I've synchronized them)
Thursday 21-Oct-2021
... by: jamsaw, USA
The problem is, other sites have different names for units...some call a 3x3 unit "box", some call it "square", some call it "region". So, when reading the instructions or analysis, it become confusing if you have ever been exposed to other Sudoku sites. Is there any chance you could describe your terminology at the beginning of the explanations? I wish there were a common terminology where everyone had the same unit names in mind.
Also, where are you, where J follows H in the alphabet?
Also very early on people decided that "I" was too close to "1" for a row name and therefore J follows H for row names. This was generally agreed to be a good idea as coordinates need to be very clear. A large fraction continue to use the cumbersome rXcY format
I've also added a Glossary
Tuesday 19-Oct-2021
... by: Obi, UK
Load Sudoku: CLICK TO LOADI believe I've found a bit of an inconsistency with the way the solver applies (at least) hidden singles in the solve path. In the sudoku provided, step 10 lists the following hidden singles:
4 found once at r1c2 in row, 4 candidates removed
6 found once at r1c1 in row, 1 candidate removed
8 found once at r9c5 in column/box, 1 candidate removed
The problem is with r1c1: the 6 is only found once in the row because it has been eliminated from r1c2 in the same round. This elimination is therefore not deducible from the state of the board at step 10.
Stepping through the steps manually on the solver separates these steps.
Maybe there is a reason the solve path wants to group these steps, but it seems bizarre to me.
Thanks again
This didn't seem important when I wrote it since all the other strategies were the interesting ones and this was a process for clearing off candidates from singles to get to the meat. But I should address it in the next update.
If you are comparing a puzzle with a score 68 to another with a score of 72 I wouldn't say one was more difficult than the other since there are a host of subjective decisions (exact points for any strategy eg) but 68 compared to 300 or 3000 is significant. Its for the bulk division of all puzzles into grades but the dividing line is arbitrary. It's like a bell chart of human heights. If you want to divide people into short and tall the division will separate people with millimetre height differences around the average
Saturday 21-Aug-2021
... by: Frank Merritt, Illinois, USA
I have never understood the labeling of shapes in Jigsaw configurations. Judging from the labels for Andrew Stuart 7 and for Cabbage, I believe they are labeled by whim rather than by any systematic procedure. Is that correct? If not, is the rule explained anywhere in your pages?
P.S. Thanks very much for such an incredibly useful reference!
There is a complete list of the jigsaw shapes here
Now I believe I have tried to assign the numbers in a top-left to bottom-right order but not every shape has that determination I can see
I might go back and renumber, next update
Monday 16-Aug-2021
... by: Theo, Switzerland
One small question: I have edited off some possibles in any Sudoku. With the "auto clear" button on, when I enter a solution, the action taken by the solver is not to logically clear out just the values that have to be cleared in that box, row and column because of this new number, but it resets ALL possibilities back. Meaning, my edits are discarded, the possibilities I edited out earlier are set back again. Could there be an option to make a real clearing, not a reset to all possibilities?
This sounds so easy, am I missing something?
Best regards, Theo
Tuesday 10-Aug-2021
... by: brent hudson, United Kingdom
Firstly, what a brilliant programme! I have used it a lot over the last 2 years as I've tried to improve my own solving skills. It has helped me improve a lot.
However, on this board I have been able to solve a puzzle that the solver cannot, and I would expect it to be able to. I am wondering why it cannot.
Here's the board:
There is a Killer Sudoku board I would like you to look at
Click on this link:
https://www.sudokuwiki.org/killersudoku.htm?bd=121112211121133441112213122132112111332312322111311321221213311133221121114411121,300622000012002100000000001600040000000025003200191400000600000013000000000000130000151200210000000000000022130000210000000000180500000027001000000008000000000000
When it gets stuck, cells G6, H6, J5, and J6 sum to 13 (by summing columns 7, 8, and 9 and looking at the spill over). 13 in 4 cells in the same block has to contain 1. This means that J4 has to be 2 (the solver gets stuck with J4 being only 1 or 2).
If I set J4 to 2, the solver then solves the puzzle.
Why doesn't the solver see that J4 has to be 2?
I ask for two reasons:
1. The solver seems to solve much harder situations than this one.
2. If you do decide to update the solver again, maybe there's a piece of logic you could insert?
Again - thanks for doing the solver; it's a great tool!
Brent