The New Sudoku Players' Forum

by **David P Bird** » Fri Jul 20, 2012 12:02 pm

champagne wrote: so do you really state that the almost shark logic says we can clear 24 candidates

You mis-understand the interpretation of the results. We have 24 Potential Eliminations, but exactly one of them is wrong. If we can identify which one it is, or narrow it down to being contained in a sub-set, then we could make all the others. That's what I've been attempting to do.

by **champagne** » Fri Jul 20, 2012 12:12 pm

David P Bird wrote:
champagne wrote: so do you really state that the almost shark logic says we can clear 24 candidates

You mis-understand the interpretation of the results. We have 24 Potential Eliminations, but exactly one of them is wrong. If we can identify which one it is, or narrow it down to being contained in a sub-set, then we could make all the others. That's what I've been attempting to do.

Accepted, but in such conditions, the solving power does not exist. The tool using it has still to be found.

Am I right

by **David P Bird** » Fri Jul 20, 2012 1:00 pm

champagne wrote:Accepted, but in such conditions, the solving power does not exist. The tool using it has still to be found.

Am I right

Absolutely, and I'm beginning to give up hope. What I've been trying to find is two sub-patterns in the PEs one or other must be true. Although we won't know which one it is it would allow all the other potential eliminations to be made.

I'm aware that in the most difficult puzzles the chances of this search being successful might be small, but for the difficult/tough puzzles it might be better.

by **ronk** » Fri Jul 20, 2012 2:48 pm

David P Bird wrote:How come you can coach daj <here> in this thread, and object to my post – which is not a discussion!

Because you post a POV which you have not yet successfully explained to anyone else in your sharks thread.

David P Bird wrote:As champagne is the owner of the thread, I will comply with his wishes if he asks me to delete that post.

Forum organization is also in the purview of moderators.

by **David P Bird** » Fri Jul 20, 2012 5:14 pm

Ronk, I'm sorry but as before you have completely missed the purpose of that post which was to say that a near miss Multi-fish might still show some potential. I don't care how it is found. The last sentence reads:

I admit that I'm not sure how you would get the same PE list from a XSudo approach when there is one less truth than there should be, but it must be possible.

I consider that you are rather more guilty of drifting threads off-topic than me, and you are practicing a "do what I say, not what I do" policy.

I've said my piece – now do what the heck you want, but the option I would prefer is for you to junk this complete exchange of ours.

by **daj95376** » Fri Jul 20, 2012 6:12 pm

Moving forward. Hopefully, champagne doesn't mind individual solutions being posted for puzzles in the "03 NN ..." file.

Puzzle 4993 seems to be a cousin of puzzle 3218.

Code: Select all: .2.4....9..7......8....2.1...1..5.7..6..9...5...2..........8.5....36...2.3..2.6.. # 4993;elev;2354;11.10;1.20;1.20;730

Hidden Text: Show

I did not resolve the column scenario, but it exists and should be very similar.

by **ronk** » Fri Jul 20, 2012 7:34 pm

daj95376 wrote:Puzzle 4993 seems to be a cousin of puzzle 3218.

Code: Select all
.2.4....9..7......8....2.1...1..5.7..6..9...5...2..........8.5....36...2.3..2.6.. # 4993;elev;2354;11.10;1.20;1.20;730

Hidden Text: Show
Code: Select all
1 R15689 \ c16b6 + r1c57,r5c4,r6c25,r8c27,r9c9 (2) 5 R1 68 \ c13 + r1c57,r5c4,r6c25,r8c27,r9c9 (1) 7 R15689 \ c16 + r1c57,r5c4,r6c25,r8c27,r9c9 (3) 8 R15689 \ c38b6 + r1c57,r5c4,r6c25,r8c27,r9c9 (2)

Another 0-rank multi-fish in the "03 NN ..." file, very very nice :!:

I missed that one.

Here is an almost-sk-loop with 8 exclusions for #14596. I don't know whether it contains an exocet, even though the solution seems to indicate it does. This pattern appears identical to something I've posted before, but it's not likely to be the same puzzle.

___

___ clickable thumbnail

puzzle+logic set: Show

[edit: add missing AURs to logic set, and add missing r1c3<>1 and r7c1<>1 to listing]

by **David P Bird** » Fri Jul 20, 2012 11:29 pm

daj Well done for finding your truth & link set for 4993.

Your column set is the same as I had for my almost solution. Looking at it again I find I can also get a " box adjusted" balance for this puzzle which produces this eliminations grid:

Code: Select all: *-------------------------*-------------------------*-------------------------* | - 2 - | 4 1578-3 - | 578-3 - 9 | | 3469-15 - 7 | - - 369-1 | - 2346-8 - | | 8 - 3469-5 | - - 2 | - 1 - | *-------------------------*-------------------------*-------------------------* | 2349 - 1 | - - 5 | - 7 - | | - 6 - | 178 9 - | (12348) - 5 | | - 578-49 - | 2 178-34 - | (13489) - (13468) | *-------------------------*-------------------------*-------------------------* | 26 - 26 | - - 8 | - 5 1347 | | - 1578-49 - | 3 6 - | 178-49 - 2 | | - 3 - | 5 2 - | 6 - 178-4 | *-------------------------*-------------------------*-------------------------*

I was only using simple balances for this exercise and, as Xsudo has the exclusive sponsorship contract for this thread, I can't explain adjusted balances further, but look at box 6, which is the box you included for two of you links.

How do theses exclusions tally with yours? (I'm lost with the Xsudo diagrams that don't reflect what the summary shows.)

BTW Exclusive sponsorship contracts for the Olympics have caused a lot of hassle here. A small shop keeper in my town on the route of the torch procession had a window display with 5 coloured hula hoops but was told to take them down. In the Olympic Park fish and chips will be available, but not our traditional pie and chips as only McDonalds can sell chips otherwise, and so it goes on with the other brands.

by **champagne** » Sat Jul 21, 2012 5:36 am

David P Bird wrote:I was only using simple balances for this exercise and, as Xsudo has the exclusive sponsorship contract for this thread, I can't explain adjusted balances further, but look at box 6, which is the box you included for two of you links. .

David, a rank 0 logic is a rank o logic, no need for XSUDO.

Allan Barker can be thanked to have stressed on that specific pattern, but the logic is now freely available.

I never check the results of my solver.

For players, and new patterns found by hand, XSUDO remains a good way to check the validity before posting the results.

The rest seems to me part of an endless fight I don't understand.

champagne

by **daj95376** » Sat Jul 21, 2012 7:11 am

Code: Select all: *-------------------------*-------------------------*-------------------------* | - 2 - | 4 1578-3 - | 578-3 - 9 | | 3469-15 - 7 | - - 369-1 | - 2346-8 - | | 8 - 3469-5 | - - 2 | - 1 - | *-------------------------*-------------------------*-------------------------* | 2349 - 1 | - - 5 | - 7 - | | - 6 - | 178 9 - | (12348) - 5 | | - 578-49 - | 2 178-34 - | (13489) - (13468) | *-------------------------*-------------------------*-------------------------* | 26 - 26 | - - 8 | - 5 1347 | | - 1578-49 - | 3 6 - | 178-49 - 2 | | - 3 - | 5 2 - | 6 - 178-4 | *-------------------------*-------------------------*-------------------------*

David, your results are very close to those of my template analyzer.

Code: Select all: <1578> accepted = 52 template combinations <1578> <>1 r2c16 <1578> <>3 r1c57,r6c5 <1578> <>4 r6c25,r8c27,r9c9 <1578> <>5 r2c1,r3c3 <1578> <>7 r6c6,r7c7 <1578> <>8 r4c7,r256c8,r4c9 <1578> <>9 r6c2,r8c27 <1578> r1c57,r5c4,r6c25,r8c27,r9c9 locked for candidates

by **David P Bird** » Sat Jul 21, 2012 10:18 am

daj95376 wrote:David, your results are very close to those of my template analyzer.

Thanks for that, but it poses the question does your template analyser give the same eliminations as the truth and link sets?

I don't really understand how Xsudo ranks an elimination set and thought that my balance-adjusting method would change a rank 0 to a rank 1. That's why I didn't use it. Now I think may have shown that that's not necessarily true, so I will have to re-examine the list of almost patterns I found.

champagne My un-adjusted 4993 PE set would leave box 6 with (1578) occupying 5 cells and (23469) occupying 4. The invalid PE is therefore confined to the three cells in that box with candidates shown in brackets and all the others are good.

I think that we only need to explore this further if my elimination set is rank 1.

by **champagne** » Wed Jul 25, 2012 3:32 pm

I did not check the list, but these puzzles should have something "exotic" not seen by my current code

puzzle,
seq number
er
ep
ed
identtification
potential eliminations/10

Hidden Text: Show

EDIT The list has been revised and puzzles solved wrongly marked in the update of the data base cleared

EDIT2 : a problem seen in the update dated 12_05 corresponding puzzles have been cleared waiting for a clarification

by **David P Bird** » Wed Jul 25, 2012 11:51 pm

The first and last of these puzzle have multi-fish eliminations with identical cover sets. I thought they were listed in decreasing order of potential eliminations, so I wonder how they evaluate so differently.

9..8..7...8..6......5..4...6...8.....3.9..6....2....1..7..3.9....1.....4.....2.5.;34189;11.30;11.30;9.40;GP;12_05;10
(1245)Shark:r3689,c36789
Home Set Min Truths = 8 + 2 more in c7, Away Set Max Truths = 10 (Adjusted Balance)
=> Home r3c89, r6c69, r8c68, r9c39 <> 1245
=> Away r1c2, r2c14, r4c24, r5c15, r7c14 <> 36789
22 Candidates, 17 Cells (No eliminations possible in c7 but two of the home cells must be true and two must be false)

98.7..6..7...9......5..4...6...7.....3..8.9.......2.5..7..3.8....1....2.........4;41106;10.90;10.90;9.80;GP;12_07;4
(1245)Shark:r3689,c36789
Home Set Min Truths = 7 + 2 more in c7, Away Set Max Truths = 9 (Adjusted Balance)
=> Home r3c89, r6c39, r8c69, r9c368 <> 1245
=> Away r2c24, r4c24, r5c4, r7c4 <> 36789
21 Candidates 15 Cells (No eliminations possible in c7 but two of the home cells must be true and two must be false)
( (1245) becomes a naked quad c4, and (39) becomes a hidden pair in b5 producing (6789) naked quad in b8)

ronk, if you can spare the time, please would you check if Xsudo can arrive at the same eliminations?
(daj I'm pretty sure that templates will get most or all of them but that's only of secondary interest to me.)

by **ronk** » Thu Jul 26, 2012 12:54 am

David P Bird wrote:ronk, if you can spare the time, please would you check if Xsudo can arrive at the same eliminations?

For each puzzle, two additional exclusions r24c7<>3 are valid.

These are relatively simple 0-rank multi-fish, at least for a software program. After recent revisions, I'm surprised champagne's program missed them.

by **champagne** » Thu Jul 26, 2012 6:26 am

Hi david and ronk,

Sorry, I likely made a wrong manipulation updating my data base,
I have to understand where and why, then I'll revise the list.

Edit; The list has been revised.
David, I noticed your remark about the potential.
The only thing I can say at that point is that the potential does not apply systematically to the floors leading to a rank 0 logic,
but a bug can not be excluded.

Last remark, the current code does not cover the pattern using boxes with a row or column basis.

The New Sudoku Players' Forum

Exotic patterns a resume

Re: Exotic patterns a resume

Re: Exotic patterns a resume

Re: Exotic patterns a resume

Re: Exotic patterns a resume

Re: Exotic patterns a resume

Re: Exotic patterns a resume

Re: Exotic patterns a resume

Re: Exotic patterns a resume

Re: Exotic patterns a resume

Re: Exotic patterns a resume

Re: Exotic patterns a resume

Re: Exotic patterns a resume

Re: Exotic patterns a resume

Re: Exotic patterns a resume

Re: Exotic patterns a resume