The New Sudoku Players' Forum

by **Papy** » Fri Aug 18, 2006 7:25 pm

Hi,

Like every body I look after news technics to solve my gridsBut the job is hard and the new technics are very commplex
So when I find a new method for elimination, speed and easy I was happy and want to give it to the Planet Sudoku
Here us my metode I call it French Pairs. I think that it can be 'ameliored'
on the grid if you have TWO cells with the TWO same candidat
the two cells , corners of the rectangle, cannot have the two candidats
Example:

r1 c1 12 and r4 c4 12

So r1C4 and R4 C1 cannot have the 1 and 2 in the candidats
to be continue....

by **underquark** » Fri Aug 18, 2006 10:58 pm

Isn't this a unique rectangle?

by **ronk** » Sat Aug 19, 2006 12:26 am

underquark wrote:Isn't this a unique rectangle?

No. With a unique rectangle, the four cells must be in two boxes.

by **tso** » Sat Aug 19, 2006 3:39 am

Papy wrote:... I call it French Pairs. I think that it can be 'ameliored' on the grid if you have TWO cells with the TWO same candidat the two cells , corners of the rectangle, cannot have the two candidats

Example:

r1 c1 12 and r4 c4 12

So r1C4 and R4 C1 cannot have the 1 and 2 in the candidats
to be continue....

This tactic does not work. If r1c1 and r4c4 are both 1, then one or both r1c4 and r4c1 can be 2 (and vice versa). Only if r1c1<>r4c4 could you eliminate 1 and 2 from r1c4 and r4c1.

Given this...

Code: Select all: *--------------------------------------------------* | 12 . . | 1234 . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | 1245 . . | 12 . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | . . . | . . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | *--------------------------------------------------*

... no candidates can be eliminated from any of these four cells.

If however, the rectangle were in two rows, two columns and two boxes ...

Code: Select all: *--------------------------------------------------* | 12 . . | 1234 . . | . . . | | . . . | . . . | . . . | | 1245 . . | 12 . . | . . . | |----------------+----------------+----------------| | . . . | . . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | . . . | . . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | *--------------------------------------------------*

... then you could deduce that at least ONE of these four statements is true:

r1c4=3
r1c4=4
r3c1=4
r3c1=5

... otherwise the puzzle would have two solutions ... and though this bit of information may or may not help you with the solution, you still could not immediately eliminate both the candidates 1 and 2 from r1c4 and r3c1.

by **Papy** » Sat Aug 19, 2006 7:56 am

Exact I dont see that problem

Perhaps with three cells?

At the beginning I use tree corner of a rectangle
To be continue....

by **tso** » Sun Aug 20, 2006 5:45 pm

Papy wrote:Exact I dont see that problem

Perhaps with three cells?

At the beginning I use tree corner of a rectangle
To be continue....

If by this, you mean this ...

Code: Select all: *--------------------------------------------------* | 12 . . | 12 . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | 12 . . | 1234 . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | . . . | . . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | *--------------------------------------------------*

... you still cannot make any valid deduction regarding r4c4, as there are two valid possibilites for the other three cells ...

Code: Select all: *--------------------------------------------------* | 1 . . | 2 . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | 2 . . | 1.34 . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | . . . | . . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | *--------------------------------------------------*

... or ...

Code: Select all: *--------------------------------------------------* | 2 . . | 1 . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | 1 . . | .234 . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | . . . | . . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | *--------------------------------------------------*

This is not enough to reduce the possiblities in r4c4.

by ab » Sun Aug 20, 2006 6:14 pm

ok papy look here:
http://forum.enjoysudoku.com/viewtopic.php?t=3957&highlight=unique+rectangle
there's a puzzle with a unique solution and the same two candidates are in four cells at the corners of a rectangle

by **Papy** » Sun Aug 20, 2006 7:50 pm

No I was thinking not true pairs

12.............. 23
.
.
.
.
.
.
.
13........... 123

But I don't study them . I have ended my 17 clues generator today.
Papy

by **coloin** » Sun Aug 20, 2006 8:21 pm

Ah I see....

If the ONLY pencil marks in r1c1 and r4c4 are 1 and 2

Code: Select all: *--------------------------------------------------* | 12only. . | x . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | x . . | 12only . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | . . . | . . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | *--------------------------------------------------*

Then this precludes 1 and 2 in both r1c4 and r4c1

Does this happen often ? Or maybe it has been described before !

C

by **Ruud** » Mon Aug 21, 2006 1:39 am

coloin wrote:Does this happen often ? Or maybe it has been described before !

It happens all the time.

This pattern is better known as XY-Wing.

Ruud

by **keith** » Mon Aug 21, 2006 2:07 am

coloin wrote:Ah I see....

If the ONLY pencil marks in r1c1 and r4c4 are 1 and 2

Code: Select all
*--------------------------------------------------* | 12only. . | x . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | x . . | 12only . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | . . . | . . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | *--------------------------------------------------*

Then this precludes 1 and 2 in both r1c4 and r4c1

Does this happen often ? Or maybe it has been described before !

C

With no other information, you can make no conclusions on the contents of any other cells This is not even a possible UR.

Ruud, can you please complete the XY-wing? I do not see it.

I wish it were so simple! But then, if it were so simple, I would not be so interested!

Keith

by **udosuk** » Mon Aug 21, 2006 6:17 am

coloin wrote:Ah I see....

If the ONLY pencil marks in r1c1 and r4c4 are 1 and 2

Code: Select all
*--------------------------------------------------* | 12only. . | x . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | x . . | 12only. . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | . . . | . . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | *--------------------------------------------------*

Then this precludes 1 and 2 in both r1c4 and r4c1

coloin, I don't think this could preclude 1 & 2 in both r1c4 & r4c1, unless you can somehow prove that r1c1 & r4c4 must have different values (as tso has demonstrated above)...

What Ruud described as XY-wing is this situation:

Code: Select all: *--------------------------------------------------* | 12 . . | 23 . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | 13 . . | 123 . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | |----------------+----------------+----------------| | . . . | . . . | . . . | | . . . | . . . | . . . | | . . . | . . . | . . . | *--------------------------------------------------*

Obviously, r4c4 could not be 3, otherwise we'll have no possible candidate to fill in r1c1. This is xy-wing (a particular application of the Almost Locked Subsets technique)...

Anyway I couldn't see any reason to rename this technique to "French Pairs" unless the original discoverer of the Almost Locked Subsets concept is also French...

And despite being a not-too-incompetent reader of the so called "French-English", I couldn't at all figure out the above configuration from Papy's original description:

Papy wrote:Here us my metode I call it French Pairs. I think that it can be 'ameliored' on the grid if you have TWO cells with the TWO same candidat the two cells , corners of the rectangle, cannot have the two candidats
Example:

r1 c1 12 and r4 c4 12

So r1C4 and R4 C1 cannot have the 1 and 2 in the candidats

Anyone who could understand the above description as xy-wing deserves to be appointed as the next British ambassador in Paris... :!:

by **Papy** » Mon Aug 21, 2006 7:26 am

yex not a new name for X wing

The name was for th first solurtion... but it's bad.
So bye bye French pair (today but tomorow be an other day..)

I have wriiten a Valid Gid and 17 clues Sudohu generator
Are sone people interrested to TEST it?
http://rapidshare.de/files/30182806/Solveur.exe.html

Papy

by **fermat** » Tue Aug 22, 2006 3:13 am

Papy wrote:I have wriiten a Valid Gid and 17 clues Sudohu generator
Are sone people interrested to TEST it?
http://rapidshare.de/files/30182806/Solveur.exe.html
Papy

The program didn't run for me. It errored (in french) that it couldn't open "c:/grilles/" ....some more that I forget.

Edited to add: that is a hint you are using linux/unix.
Slashes "/\" are often different under windows.
Never mind, no one will have C:/Grilles.

by **Papy** » Tue Aug 22, 2006 7:13 am

Hi,
The soft Win32 look at a file
'c:/Grilles/Australie.txt'
You can create a empty file!
It'a a test version I forget that but other things also
I will change the exe today
But If you create the folder and the file
Go to the 17 clues generator, look at the speed and after on the HD uyou will have a file 17_Clues.txt
Look at the number of valids and diffefent sudoku!
Yopu wil nderstand why my other pseudo is John Fullspeed

Soory
Papy

The New Sudoku Players' Forum

The French Pairs...

The French Pairs...

Re: The French Pairs...

French paire

Re: French paire

French Pair

Incorrect

No name

Re: No name

miistake