The New Sudoku Players' Forum

[champagne]

the file of potential hardest puzzles has shown it had a cut-off to high for the search done on exotic patterns used to solve such puzzles.

I prepared in the skfr google project the entire file of puzzles rating (skfr) 10.0 and more I had available

warning: that file contains all the puzzles I used as seed in my search including already known (see the readme file)


5million puzzles file

[RW]

Another front page puzzle by AI in my morning paper claiming to be the new world's hardest
8..........36......7..9.2...5...7.......457.....1...3...1....68..85...1..9....4..

ER only 10.7, q1 95991 and q2 97638. Can't get suexrat on my 64-bit system (is there any way to use this program under win7 64?). Doesn't seem hard enough to make the hardest database anyway.

Is this even a new puzzle, or has it been posted here before?

[champagne]

does not seem to be part of the 5 millions puzzles file

champagne

[ronk]

Another front page puzzle by AI in my morning paper claiming to be the new world's hardest
8..........36......7..9.2...5...7.......457.....1...3...1....68..85...1..9....4..

ER only 10.7, q1 95991 and q2 97638.

And an overlay of <1368>-jellyfish for 23 exclusions at the start reduces the ER from 10.7 to 9.0.

Code: Select all

 base \ ---------------- cover ----------------------
1r3459\c19  + r3c6 +(r4c5)+ r4c7 + r5c2 + r9c5 + r9c6
3r3459\c149 + r3c6 + r4c5 +(r4c7)+ r5c2 + r9c5 + r9c6
6r3459\c139 +(r3c6)+ r4c5 + r4c7 + r5c2 + r9c5 + r9c6
8r3459\c48  + r3c6 + r4c5 + r4c7 + r5c2 + r9c5 + r9c6

"(rMcN)" indicates cell void of candidate

logic set pastable into Xsudo: Show

Code: Select all

     16 Truths = {1R3459 3R3459 6R3459 8R3459}
     16 Links = {1c19 3c149 6c139 8c48 5n2 49n5 39n6 4n7}
     23 Eliminations --> r6c139<>6, r1c9<>136, r2c19<>1, r7c14<>3, r8c19<>3, r9c56<>2, r1c4<>3,
     r1c3<>6, r2c8<>8, r3c6<>4, r4c5<>2, r4c7<>9, r5c2<>2, r8c1<>6, r9c5<>7

[champagne]

Another front page puzzle by AI in my morning paper claiming to be the new world's hardest
8..........36......7..9.2...5...7.......457.....1...3...1....68..85...1..9....4..

ER only 10.7, q1 95991 and q2 97638.

And an overlay of <1368>-jellyfish for 23 exclusions at the start reduces the ER from 10.7 to 9.0.

Interesting point that I did not check

So we have now a puzzle with an "exotic pattern" identical to what has been found in our file of "potential hardest"

[David P Bird]

So we have now a puzzle with an "exotic pattern" identical to what has been found in our file of "potential hardest"

[Edit] What follows is mistaken - I should have double checked it.
Please check out that puzzle with this one:
98.7..6......96..........5.7....94...3.4.......287....6...8.7...2.....3...1.....5;9470;GP;cy4;1235;
which you published <here>

If found this completely by chance as I recognised the multi-fish Shark pattern it needed as being familiar.

I think the solution grids are morphs of one another, but I haven't the tools to analyse the similarities between the givens further.
DPB

[champagne]

So we have now a puzzle with an "exotic pattern" identical to what has been found in our file of "potential hardest"

Please check out that puzzle with this one:
98.7..6......96..........5.7....94...3.4.......287....6...8.7...2.....3...1.....5;9470;GP;cy4;1235;
which you published <here>

If found this completely by chance as I recognised the multi-fish Shark pattern it needed as being familiar.

I think the solution grids are morphs of one another, but I haven't the tools to analyse the similarities between the givens further.

DPB

I checked for a possible morph, but it is not unless my canonicalisation process is bugged.

After that, I don't see clearly what I can do.

champagne

[David P Bird]

champagne, sorry for wasting your time, it was my mistake. I had two instances of my helper spreadsheet running but through my bad typing both were using the same input source.

DPB

[champagne]

champagne, sorry for wasting your time, it was my mistake. I had two instances of my helper spreadsheet running but through my bad typing both were using the same input source.

DPB

no problem David.




I had not yet introduced in my solver the use of multi_fishes.
It's done now.
here is the start for the last "AI claimed to be the new world's hardest"

as noticed ronk, after that, the most difficult step is in serate a dynamic contradiction chain

Hidden Text: Show

8..........36......7..9.2...5...7.......457.....1...3...1....68..85...1..9....4..
multi_fish floors 1368

Code: Select all

8g    16+   6+    |3+    13+   13+   |136+  X     136+ 
1+    1+    3g    |6g    18+   18+   |18+   8+    1+   
16+   X     6+    |38+   X     138+  |X     8+    136+ 

136+  X     6+    |38+   368+  X     |168+  8+    16+   
136+  1368+ 6+    |38+   X     X     |X     8+    16+   
6+    68+   6+    |1g    68+   68+   |68+   3g    6+   

3+    3+    1g    |3+    3+    3+    |3+    6g    8g   
36+   36+   8g    |X     36+   36+   |3+    1g    3+   
36+   X     6+    |38+   1368+ 1368+ |X     X     3+   

16 Truths = {1R3459 3R3459 6R3459 8R3459 }
16 Links = {1c19 3c149 6c139 8c48 3n6 4n57 5n2 9n56 }

23 eliminations
1r1c9 1r2c1 1r2c9 3r1c4 3r1c9 3r7c1 3r7c4 3r8c1
3r8c9 6r1c3 6r1c9 6r6c1 6r6c3 6r6c9 6r8c1 8r2c8
4r3c6 2r4c5 9r4c7 2r5c2 27r9c5 2r9c6

starting in that position after eliminations

Code: Select all

A      B    C    |D    E     F     |G     H    I     
8      1246 2459 |247  12357 1234  |13569 4579 4579 
2459   124  3    |6    12578 1248  |1589  4579 4579 
1456   7    456  |348  9     138   |2     458  13456
----------------------------------------------------
123469 5    2469 |2389 368   7     |168   2489 12469
12369  1368 269  |2389 4     5     |7     289  1269 
2479   2468 2479 |1    268   2689  |5689  3    2459 
----------------------------------------------------
2457   234  1    |2479 237   2349  |359   6    8     
247    2346 8    |5    2367  23469 |39    1    279   
23567  9    2567 |2378 1368  1368  |4     257  2357 


HP 36 box 3
1 r2c7=1
2 r3c8=8
-> row 3 digit 5 in box 1
NP 249 box 1
3 r3c4=4

and the position where chains are needed in my new solver

Code: Select all

A      B    C    |D    E     F     |G    H    I     
8      16   249  |27   12357 123   |36   4579 4579 
249    24   3    |6    2578  28    |1    4579 4579 
156    7    56   |4    9     13    |2    8    36   
---------------------------------------------------
123469 5    2469 |2389 368   7     |68   249  12469
12369  1368 269  |2389 4     5     |7    29   1269 
2479   2468 2479 |1    268   2689  |5689 3    2459 
---------------------------------------------------
2457   234  1    |279  237   2349  |359  6    8     
247    2346 8    |5    2367  23469 |39   1    279   
23567  9    2567 |2378 1368  1368  |4    257  2357

EDIT changed initial table (PM limited to floors) and print of Truths Links
changed 23 eliminations and not 22

[ronk]

16 Truths = {3R1368 4R1368 5R1368 9R1368 }
16 Links = {1c136 3c6 4c38 8c8 9c136 3n6 4n57 5n2 9n56 }

Dyslexia?

[champagne]

16 Truths = {3R1368 4R1368 5R1368 9R1368 }
16 Links = {1c136 3c6 4c38 8c8 9c136 3n6 4n57 5n2 9n56 }

Dyslexia?

I must agree that Allan put digits first and rows or columns after the letter.
This is against my own logic, but I have to stick to Allan's rules
I looked at old posts to see what is done for a mix of sets rows and columns.
It seems that here the form 1368R3459 would be correct for the sets, but I am no sure.

I'll adjust the code and revise the post above.
Do you have the expected form for a mix of rows and columns in Truths or links

[ronk]

Do you have the expected form for a mix of rows and columns in Truths or links

Separate terms are required.

It seems that here the form 1368R3459 would be correct for the sets, but I am no sure.

I've witten that form and Xsudo reads it just fine, even though it doesn't write it.

I must agree that Allan put digits first and rows or columns after the letter.
This is against my own logic ...

Row number(s) before the letter 'r'? AFAIK no one does that! What are you smoking?

[daj95376]

While champagne is updating his post:

22 eliminationss
1r1c9 1r2c1 1r2c9 3r1c4 3r1c9 3r7c1 3r7c4 3r8c1
3r8c9 6r1c3 6r1c9 6r6c1 6r6c3 6r6c9 6r8c1 8r2c8
4r3c6 2r4c5 9r4c7 2r5c2 27r9c5 2r9c6

23 eliminationss
1r1c9 1r2c1 1r2c9 3r1c4 3r1c9 3r7c1 3r7c4 3r8c1
3r8c9 6r1c3 6r1c9 6r6c1 6r6c3 6r6c9 6r8c1 8r2c8
4r3c6 2r4c5 9r4c7 2r5c2 27r9c5 2r9c6

[champagne]

While champagne is updating his post:

22 eliminationss
1r1c9 1r2c1 1r2c9 3r1c4 3r1c9 3r7c1 3r7c4 3r8c1
3r8c9 6r1c3 6r1c9 6r6c1 6r6c3 6r6c9 6r8c1 8r2c8
4r3c6 2r4c5 9r4c7 2r5c2 27r9c5 2r9c6

23 eliminationss
1r1c9 1r2c1 1r2c9 3r1c4 3r1c9 3r7c1 3r7c4 3r8c1
3r8c9 6r1c3 6r1c9 6r6c1 6r6c3 6r6c9 6r8c1 8r2c8
4r3c6 2r4c5 9r4c7 2r5c2 27r9c5 2r9c6

the post is updated, the fix has been easier than expected,
this is a counting mistake, the number of eliminations was not in my print
I missed the double digit in r9c5

I'll add the count in my print to avoid the same mistake later

thanks

[ronk]

Andrew Stuart has proposed the number of quads backdoors of maximum size as a new metric for difficult puzzles.

"Quads backdoors" means, in addition to the backdoors guesses, the techniques used are singles, locked candidates, naked and hidden subsets (pairs, triples, and quads).