YZF_Sudoku

Programs which generate, solve, and analyze Sudoku puzzles

Re: YZF_Sudoku

Postby yzfwsf » Wed Jan 27, 2021 2:46 am

ghfick wrote:Hi yzfwsf,

Excellent to have Grouped AICs added.

Have a look at :

.1.5...9.2...3...5..9.......4.6....8.3.....1...1.8.76....2....16....7.2..8..6.4..

If you move to just before the first Grouped AIC Type I and then 'Show Next Step' , you get a step labelled ALS but the visual is not correct [showing part of the Grouped AIC?] . 'Show Next Step' should show the first Grouped AIC as listed in the solution path.

I should have more for you later today or tomorrow.

Gordon


The bug mentioned has been fixed, and a bug in Grouped AIC has been fixed. Please download 620V2.
Release Version 620V2

BTW: More Grouped AIC puzzle
Code: Select all
9...3..8......2..4.4.8..2...5...3..7..1.2..4.4..9..8..1..2...3...2.6...5.....19..
yzfwsf
 
Posts: 410
Joined: 16 April 2019

Re: YZF_Sudoku

Postby jco » Thu Jan 28, 2021 3:09 pm

Hello yzfwsf,

I think I found a small bug in identification of a cell in a almost locked pair.
For the puzzle
.....6..3.4.....8..9...17..5...87.4...........6.94...7..27...9..5.....1.3..6.....

the first non-basic move given by YZF_SUDOKU is

Almost Locked Pair: 49 in r456c3 r5c1 r9c3 => r8c3<>49 r5c1<>1278

I think it should be 49 in r458c3, instead of r456c3.
That would make it agree with the image shown for that move.
Hidden Text: Show
this loop
Code: Select all
(4)r5c1=(4)r5c3-(4=9)r8c3-(9)r8c1=(9)r5c1
makes the same eliminations.

Another equally small issue: in the listing of moves we have moves named locked pair and naked pair.
I was expecting only naked pair/hidden pair (locked pair could be replaced by naked pair).

Regards,
jco.
EDIT: small correction in the hidden part.
Last edited by jco on Thu Jan 28, 2021 7:39 pm, edited 1 time in total.
JCO
jco
 
Posts: 280
Joined: 09 June 2020

Re: YZF_Sudoku

Postby ghfick » Thu Jan 28, 2021 5:35 pm

If the two cells that form the Naked Pair are confined to two houses (a row and a block or a column and a block), they are called a Locked Pair. Candidates can be eliminated from both houses.
ghfick
 
Posts: 147
Joined: 06 April 2016

Re: YZF_Sudoku

Postby yzfwsf » Thu Jan 28, 2021 5:38 pm

Hi jco:Thanks.
Almost Locked Pair: 49 in r456c3 r5c1 r9c3 => r8c3<>49 r5c1<>1278

Should be:Almost Locked Pair: 49 in r45c3 r5c1 r9c3 => r8c3<>49 r5c1<>1278
this loop
CODE: SELECT ALL
(4)r5c1=(4)r5c3-(4=9)r8c3-(9)r8c1=(9)r5c1
makes the same eliminations.

If you use find all steps function, will take following hint:
Grouped Continuous Nice Loop: (9=4)r9c3 - r5c3 = (4-9)r5c1 = r45c3 - 9r9c3 => r5c1<>1,r5c1<>2,r5c1<>7,r5c1<>8,r8c3<>4,r8c3<>9

I was expecting only naked pair/hidden pair (locked pair could be replaced by naked pair).

Locked pair is a naked pair in mini-row/mini-column.
yzfwsf
 
Posts: 410
Joined: 16 April 2019

Re: YZF_Sudoku

Postby jco » Thu Jan 28, 2021 8:12 pm

Hello yzfwsf,

yzfwsf wrote: If you use find all steps function, will take following hint:
Grouped Continuous Nice Loop: (9=4)r9c3 - r5c3 = (4-9)r5c1 = r45c3 - 9r9c3 => r5c1<>1,r5c1<>2,r5c1<>7,r5c1<>8,r8c3<>4,r8c3<>9


Thanks! (this one using the ALP cells).

Regards,
jco
JCO
jco
 
Posts: 280
Joined: 09 June 2020

Re: YZF_Sudoku

Postby rjamil » Thu Jan 28, 2021 9:39 pm

Hi all,

yzfwsf wrote:Hi jco:Thanks.
Almost Locked Pair: 49 in r456c3 r5c1 r9c3 => r8c3<>49 r5c1<>1278

Should be:Almost Locked Pair: 49 in r45c3 r5c1 r9c3 => r8c3<>49 r5c1<>1278

I think, showing r456c3 as mini-column is better option.

However, there is another ALP that eliminates the same as follows:
49 @ r789c1 r5c1 r9c3 => --49 @ r8c3 -1278 @ r5c1

R. Jamil
rjamil
 
Posts: 658
Joined: 15 October 2014
Location: Karachi, Pakistan

Re: YZF_Sudoku

Postby ghfick » Thu Jan 28, 2021 11:04 pm

JE mystery.png
JE mystery.png (131.17 KiB) Viewed 346 times


98.7..6....7.5..9.........74....32...3.....4...98....6.2..1......65....8......16.

yzfwsf wrote:Enumerate base pairs and apply JE rules to find out all reasonable number pairs. If these reasonable number pairs can delete a certain number, then it will become the final elimination.


I am returning to the JE compatibility check. How does your solver identify these 2 exclusions? The 2 exclusions do not arise from DPB's version of IBP?
ghfick
 
Posts: 147
Joined: 06 April 2016

Re: YZF_Sudoku

Postby yzfwsf » Fri Jan 29, 2021 5:42 am

Hi ghfick:
The double UR threat does not detect this elimination, but performs permutation detection on the base pair.
yzfwsf
 
Posts: 410
Joined: 16 April 2019

Re: YZF_Sudoku

Postby ghfick » Mon Feb 01, 2021 1:06 am

HI yzfwsf,

1) By permutation detection I think you mean that each possible pair of base cell digits is considered looking for contradictions? This search technique does not then seem to be available to human solvers. If not for humans, maybe this search should be an option; something separate from the JE exclusions.

2) I wanted to ask you more about adding a fourth cross line. So far, I have seen just two examples U383 and U385 where the fourth cross line is parallel to the other cross lines. In both cases, other ways are available to get the exclusions. YZF_Sudoku sometimes gives a set of extra "S" cells in a line that is not parallel to the other cross lines. I am unsure how this works.

...and a few more minor edits :
Junior Exocet: _ Base ... [add a space]
Senior Exocet: _ Base ... [add a space]
MSLS: _ XXCells .... [add a space]

All the best
Gordon
ghfick
 
Posts: 147
Joined: 06 April 2016

Re: YZF_Sudoku

Postby yzfwsf » Mon Feb 01, 2021 5:07 pm

ghfick wrote: By permutation detection I think you mean that each possible pair of base cell digits is considered looking for contradictions? This search technique does not then seem to be available to human solvers. If not for humans, maybe this search should be an option; something separate from the JE exclusions.

These eliminations have been listed separately.
I wanted to ask you more about adding a fourth cross line. So far, I have seen just two examples U383 and U385 where the fourth cross line is parallel to the other cross lines. In both cases, other ways are available to get the exclusions. YZF_Sudoku sometimes gives a set of extra "S" cells in a line that is not parallel to the other cross lines. I am unsure how this works.

The additional cross-lines works the same way.Search for jellyfish with Rank=-1 for a single base digit, that is, find 3 houses of cover set for covering 4 houses of defining set. The defining set is not include target cells and those cells that see two base cells.
If the base cells have this digit, the above rules ensure that the target cells must have that base digit.
yzfwsf
 
Posts: 410
Joined: 16 April 2019

Re: YZF_Sudoku

Postby yzfwsf » Wed Mar 10, 2021 6:10 am

Fixed some minor bugs, added Cell/Region Forcing Chain support for group nodes .
620V3 Release on my google drive.
yzfwsf
 
Posts: 410
Joined: 16 April 2019

Re: YZF_Sudoku

Postby yzfwsf » Fri Mar 12, 2021 4:43 am

Hi ghfick:
I am here to answer your question for the convenience of other forum friends.
In fact, my solver can use the "Modify Givens" menu to act as a simulator of algebraic methods
http://forum.enjoysudoku.com/endor-fins-1-2-ser-8-3-8-9-t38805.htmlpuzzle No.2
Code: Select all
...........98..76..5..4.98..9..3......76.....8....75..9.87..65.6..92.........5...

First delete all values 1/2/3/4, and then fill r1369c4 to 1/2/3/4, you will find the puzzle rate of SE6.6, which can be easily solved. The final number mapping is as follows: 1 <=> 3,2 <=> 2,4 <=> 4.
yzfwsf
 
Posts: 410
Joined: 16 April 2019

Re: YZF_Sudoku

Postby yzfwsf » Fri Mar 19, 2021 9:06 pm

At the request of ghfick, I support 4 kinds of views on the board, but I don't know how this function is helpful to the manual solver? There is no guarantee that there will be no bugs, please test and feedback.
The algorithm of the puzzle generator is optimized, and the speed of generating puzzles is much faster than before.
release version 620V4 on my google drive
yzfwsf
 
Posts: 410
Joined: 16 April 2019

Re: YZF_Sudoku

Postby ghfick » Sat Mar 20, 2021 6:05 pm

This latest addition to YZF_Sudoku looks VERY impressive and the real impact of the 4 view display can now be determined.
Gordon
ghfick
 
Posts: 147
Joined: 06 April 2016

Re: YZF_Sudoku

Postby yzfwsf » Sat Mar 27, 2021 3:39 am

Thanks to jco, I fixed the error of searching for the x rule (double threat) in SE. BTW: All Hardest sudoku of my solver is provided by an internal database, which is a part of the puzzle extracted from the database of champagne. No transformation has been made. It is only convenient for users to learn MSLS, JE, SE, SK-Loop and other related technologies. Other puzzles are randomly generated by the program and filtered by skfr score. Of course, to speed up, I used a pre-filter. The internal database can be replaced by an external database by the user, from Tool->Load Load the Extenal Puzzle menu. The external database can be created by the program using the following steps. Activate the grid pannel and use the Ctrl+Shift+M key combination to call up the creation wizard. Please provide a txt file,Each line is a puzzle with 81 characters. Don’t be free Line (including the end of the file).

Release 620V5
yzfwsf
 
Posts: 410
Joined: 16 April 2019

PreviousNext

Return to Software