17-clue and 18-clue Sudoku update

Everything about Sudoku that doesn't fit in one of the other sections

Postby gsf » Mon May 14, 2007 11:12 pm

gfroyle wrote:
Havard wrote:hi there gfroyle. Good to have you back.

This is getting a bit confusing, since I have found quite a few more already.

Havard


Sorry, my fault... I did not add *ALL* of the new ones that you found to the list...

You are correct that the list should have contained 39382.

However putting your new ones into *my* program has now increased the number a bit more... I am up to 39393 now.

BTW, gsf's file of 2007-05-14 contains 39305 lines of which 1 line is duplicated....

More updates later..

Gordon

now I'm even more confused
the gfroyle catalog I snarfed this morning contains 39304 puzzles
all of the posted 17s up to 2007-01-22 (36637) are accounted for in the 39304
I show havard's additions (116) also covered in the 39304
can you post one grid from havard that is not in the 39304 so I can see what went awry
thanks
gsf
2014 Supporter
 
Posts: 7306
Joined: 21 September 2005
Location: NJ USA

Postby ronk » Tue May 15, 2007 12:00 am

Havard wrote:which brings the catalog up to 39386 puzzles.

Here are four more:
Code: Select all
...2....91...8.....7..6..4.3.....5.....4.7......9.........5.3...4........92......
....31.........5.2.........3.7.........8..4..1..5......9....637...21...........9.
..7...49.....5.....2.......51.....8....7.2...3..6........4..6..15.............7..
..7...43.....5.....2.......51.....8....7.2...3..6........4..6..15.............7..

[edit: and another four]
Code: Select all
.1....5.......2......8.....3.2..6.......7.1.........8.4...5..73........62..1.....
.6..8..9.3..51..............7.4.35........1.......9...4.13............6.5........
.6..8..9.3..21..............7.4.35........1.......9...4.13............6.5........
.8....3.......15......7..6....8.....5.4...2..1.........6.....71...35.......4.....
ronk
2012 Supporter
 
Posts: 4764
Joined: 02 November 2005
Location: Southeastern USA

Postby gfroyle » Tue May 15, 2007 1:48 am

gsf wrote:now I'm even more confused


Havard is adding the new ones to the bottom of his original post...

So even if you *think* you have read it, and incorporated them into your listing, then you need to check again.

Anyway, I have now added

- all of Havards
- the ones given one post above by Ronk
- a few more new ones

The current total is 39421 and is in the usual place and contains all that I know from all sources...

Glenn, my comment on the web page (about attribution) was inspired by the format that I saw you were using.... unfortunately I do not think that I can recover the originator for the ones that people have sent me privately (mostly from Kohei Noshita) although I could try and trawl through my email records. The main problem is that once I get them, I use my canonizer and the original one "disapppears". Because I periodically sort them, I don't have an original NUMBER either..

Clearly these were bad decisions but at the time I didn't realize that it might turn into a global effort!

Gordon
gfroyle
 
Posts: 214
Joined: 21 June 2005

Postby gsf » Tue May 15, 2007 2:19 am

gfroyle wrote:
gsf wrote:now I'm even more confused


Havard is adding the new ones to the bottom of his original post...
So even if you *think* you have read it, and incorporated them into your listing, then you need to check again.

I thought that post was in this thread, 6 updates, last dated sun eve may 13, with 116 puzzles
gfroyle wrote:Anyway, I have now added

- all of Havards
- the ones given one post above by Ronk
- a few more new ones

The current total is 39421 and is in the usual place and contains all that I know from all sources...

Glenn, my comment on the web page (about attribution) was inspired by the format that I saw you were using.... unfortunately I do not think that I can recover the originator for the ones that people have sent me privately (mostly from Kohei Noshita) although I could try and trawl through my email records. The main problem is that once I get them, I use my canonizer and the original one "disapppears". Because I periodically sort them, I don't have an original NUMBER either..

Clearly these were bad decisions but at the time I didn't realize that it might turn into a global effort!

thanks
I'll hold off then until you and havard are done
and then I'll merge
it turns out the only non-gfroyle attribution I have pre 2007 is one from wolfgang
gsf
2014 Supporter
 
Posts: 7306
Joined: 21 September 2005
Location: NJ USA

Postby gfroyle » Tue May 15, 2007 4:22 am

gsf wrote:I'll hold off then until you and havard are done
and then I'll merge


Will we ever be done?:)

My search is heuristic in nature and so could run indefinitely...

Gordon
gfroyle
 
Posts: 214
Joined: 21 June 2005

Postby Havard » Tue May 15, 2007 8:58 am

I'll keep updating this post: http://forum.enjoysudoku.com/viewtopic.php?p=43947#p43947

Congratulations ronk on the new finds! What are your method? Would you, gfroyle, like to share yours too?

Havard
Havard
 
Posts: 377
Joined: 25 December 2005

Postby ronk » Tue May 15, 2007 11:05 am

Havard wrote:Congratulations ronk on the new finds! What are your method?

Same 2-off/2-on search as you, but just on the 2599 puzzles added between gfroyle39227.dat and gfroyle36628.dat. Since you started with the later, I thought some 17s might get lost there.
ronk
2012 Supporter
 
Posts: 4764
Joined: 02 November 2005
Location: Southeastern USA

Postby Ocean » Tue May 15, 2007 2:38 pm

Welcome back to the forum, Gordon!

Since I was too late with the previous attempt, I had to try a little bit more. These two 17s were not in the latest published sets, as far as I can judge:

Code: Select all
.......1......2..3...4...........5..4.16.......71......5....2......8..4..3.91....
.......1......2..3...4...........5..6.17.......41......5....2......8..6..3.91....


(Experimenting with a different algorithm from the one used by Havard and ronk, so my search should not compete directly. Results come slowly, but if this one is new, it does at least work.)
Ocean
 
Posts: 442
Joined: 29 August 2005

Postby coloin » Tue May 15, 2007 2:51 pm

gfroyle wrote:Will we ever be done?

Well, we just might one day........
It depends on what we are dealing with.....

If, as I suspect [knowing sudoku puzzles there is always a puzzle which proves you wrong] there is a puzzle out there which is remote - ie there is no puzzle associated with it within a 2off/2on distance, then what we are dealing with is a pocket of 17s [majority] which are all related within 2off/2on of each other.

Havard will find ALL these in the next five days.:D

There will be a presently an unknown number of puzzles who dont fit in this pocket. These will either be singular or small groups.

As I have questioned before these will have been found by other methods not 2off/2on. [ whatever dukuso or wolfgang did .[Or Ocean, presently.]]

Perhaps a 2off/1on technique on an 18 clue puzzle.:?: [Now theres another search for Haverd [to do easily:!: ]

Suffice to say I dont know if any puzzle exists which proves this.

Depending on the frequency of these puzzles occurring it remains to be seen if there exists a 17 which doesnt have a related 3off/3on cousin.
At the rate we are progressing who knows what is round the corner ?

C
coloin
 
Posts: 1638
Joined: 05 May 2005

Postby Havard » Tue May 15, 2007 3:19 pm

coloin wrote:Perhaps a 2off/1on technique on an 18 clue puzzle.:?: [Now theres another search for Haverd [to do easily:!: ]

Suffice to say I dont know if any puzzle exists which proves this.

Heh, a lot of thinking about this is being done at the moment:
http://www.sudoku.frihost.net/forum/viewtopic.php?p=4632#p4632
:)

Havard
Havard
 
Posts: 377
Joined: 25 December 2005

22 million 18s

Postby gfroyle » Tue May 15, 2007 10:56 pm

By popular request, there are 22 million 18-clue Sudoku puzzles available from my website.. they are split into 44 files, each containing 500000 puzzles that use about 70Mb each.

The first file is called

people.csse.uwa.edu.au/gordon/sudoku18_aa.zip { broken link }

and then to get the rest just replace aa with ab, ac, ad, ... , br

(old-style Unix people will recognise the legacy of using the command "split" in default mode!)

They are compressed with zip because it is easily accessible on Win/Mac/Linux rather than because it does the best compression.

Enjoy!

Gordon
gfroyle
 
Posts: 214
Joined: 21 June 2005

Re: 22 million 18s

Postby gsf » Wed May 16, 2007 7:09 am

here's one more
Code: Select all
100000080000009200700030000260500000000000000000000073030608000090000000000000107
gsf
2014 Supporter
 
Posts: 7306
Joined: 21 September 2005
Location: NJ USA

Postby Pat » Wed May 16, 2007 7:09 am

Havard wrote:
coloin wrote:Perhaps a 2off/1on technique on an 18-clue puzzle

Heh, a lot of thinking about this is being done at the moment:
http://www.sudoku.frihost.net/forum/viewtopic.php?p=4632#p4632


so now Havard's
http://forum.enjoysudoku.com/viewtopic.php?p=43947#p43947
is not being updated?

i see the newer finds are being posted at
http://www.sudoku.frihost.net/forum/viewtopic.php?t=2923
User avatar
Pat
 
Posts: 3449
Joined: 18 July 2005

Re: 22 million 18s

Postby gfroyle » Wed May 16, 2007 7:53 am

gsf wrote:here's one more
Code: Select all
100000080000009200700030000260500000000000000000000073030608000090000000000000107


??? My programs say that this is isomorphic to #3734 on the existing list (the new ones have been added to the end of the list, so this should be 3734 on earlier lists too).
gfroyle
 
Posts: 214
Joined: 21 June 2005

Re: 22 million 18s

Postby gsf » Wed May 16, 2007 8:06 am

gfroyle wrote:
gsf wrote:here's one more
Code: Select all
100000080000009200700030000260500000000000000000000073030608000090000000000000107


??? My programs say that this is isomorphic to #3734 on the existing list (the new ones have been added to the end of the list, so this should be 3734 on earlier lists too).

here's the original grid I have for 3734
Code: Select all
....31....4....2..............27..5.3.1....9.6........5.....6.3.9.8........4.....

does that check?
[edit] it doesn't check with the latest sudoku17
would you post the grid this is isomorphic to
thanks
gsf
2014 Supporter
 
Posts: 7306
Joined: 21 September 2005
Location: NJ USA

PreviousNext

Return to General