The hardest sudokus (new thread)

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

Re: The hardest sudokus (new thread)

Postby eleven » Tue Sep 12, 2023 1:42 pm

Paquita wrote:These are new yet unpublished T&E(3) puzzles
Hidden Text: Show
98.76.5..7.45..6....5......6.7.9.....5.87..96..8..4.7..4..........98...3...4...21 ED=11.7/1.2/1.2
98.76.5..7.54..8...46......5.8.94...46.8......795......9....4.....6...3......7.21 ED=11.7/1.2/1.2
987......6...........85....4.352.1..1.2..43.......3....4.3..21....2...58....184.3 ED=11.7/1.2/1.2
98.76.5..7.58.49...64......67.58....5.9.47.....89...5.4.6............63........21 ED=11.7/1.2/1.2
98.76.5..7.58.49...64......67.58....5.9..7....489...5.4.6.........4..63........21 ED=11.7/1.2/1.2
9876.........958.......4...76.......3.92..7...283..6..2.6....97.73...2.8......36. ED=11.7/1.2/1.2
9876.........958.......4...76.......3.92..7...283..6..2.6....97.9....36...3...2.8 ED=11.7/1.2/1.2
98.76.5..7.58.49...64......67.98....5.9.47.....85...9.4.6............63........21 ED=11.7/1.2/1.2
98.76.5..7.45......6..48...6...57...54.89..6..7........9....83....98.4.6.......21 ED=11.7/1.2/1.2
98.76.5..7.45..6...6..48...6...57...54.89.....7........9....83....98.4.6.......21 ED=11.7/1.2/1.2
...

The first 8 ones are simple after placing the only extra candidate of the TH.
The 9th and 10th (same after basics) go down to ER 8.5, because 4r4c4 directly implies 8r6c1 => 8r6c1.
eleven
 
Posts: 3151
Joined: 10 February 2008

Re: The hardest sudokus (new thread)

Postby marek stefanik » Tue Sep 12, 2023 3:17 pm

And we can finish like this:
Code: Select all
,----------------,-----------------,---------------,
| 9    8   x23-1 | 7      6    123 | 5    14   234 |
| 7    123  4    | 5     x23-1 9   | 6    18   238 |
|y123  6    5    |z123    4    8   |x23   7    9   |
:----------------+-----------------+---------------:
| 6   x23-1 1239 | 1234   5    7   | 123  489  48  |
| 5    4   y123  | 8      9   z123 | 7    6   x23  |
| 8    7    1239 |x23-146 123  46  | 123  49   5   |
:----------------+-----------------+---------------:
| 124  9    6    | 124    12   5   | 8    3    7   |
| 123 y123  7    | 9      8   x23-1| 4    5    6   |
| 34   5    8    | 346    7    46  | 9    2    1   |
'----------------'-----------------'---------------'
x=r3c7=r5c9
If the remaining digits in r35 cross (r3c1=r5c6 and r3c4=r5c3), they are both eliminated from r1c36, i.e. contra.
Therefore y=r3c1=r5c3 and z=r3c4=r5c6.
yb7p5, NS xr8c6, xb2p5, xb1p3, xb4p2, xb5p7, stte

Marek
marek stefanik
 
Posts: 359
Joined: 05 May 2021

Re: The hardest sudokus (new thread)

Postby hendrik_monard » Tue Sep 12, 2023 4:41 pm

mith wrote:@hendrik, JPF, Paquita - I was planning to scrape the thread for puzzles since ph2010 to make sure no depth 3 puzzles haven't been overlooked (and to add the depth 2 puzzles to the ph/T&E(2) database eventually), but if any of you are able to provide just the list of puzzles (regardless of depth) you have posted it would be helfpul as a check. Don't worry about duplicates or morphs, I will catch those on my end. Will also check against Paquita's combined T&E(2) file.

Hi Mith, when you write:"the list of puzzles (regardless of depth) you have posted", do you mean recently (and since when), or ever?
hendrik_monard
 
Posts: 96
Joined: 19 April 2021
Location: Leuven (Louvain) Belgium

Re: The hardest sudokus (new thread)

Postby mith » Tue Sep 12, 2023 6:27 pm

hendrik_monard wrote:
mith wrote:@hendrik, JPF, Paquita - I was planning to scrape the thread for puzzles since ph2010 to make sure no depth 3 puzzles haven't been overlooked (and to add the depth 2 puzzles to the ph/T&E(2) database eventually), but if any of you are able to provide just the list of puzzles (regardless of depth) you have posted it would be helfpul as a check. Don't worry about duplicates or morphs, I will catch those on my end. Will also check against Paquita's combined T&E(2) file.

Hi Mith, when you write:"the list of puzzles (regardless of depth) you have posted", do you mean recently (and since when), or ever?


For the te3 database, I just need to check since my last update (2022-11-06). I should have all of yours before that, since I did the same check before that update too.

For te2/ph, apparently my last update locally was 2022-03-19 but I don't know how thorough that update was so I will probably just check back to ph2010 (here).
mith
 
Posts: 996
Joined: 14 July 2020

Re: The hardest sudokus (new thread)

Postby hendrik_monard » Tue Sep 12, 2023 7:37 pm

Hi Mith,
Tomorrow I'll put all the puzzles I posted (since april 2021) in a file on google drive.
hendrik_monard
 
Posts: 96
Joined: 19 April 2021
Location: Leuven (Louvain) Belgium

Re: The hardest sudokus (new thread)

Postby Paquita » Tue Sep 12, 2023 8:55 pm

This is funny

98765.43.64.3......53......57..36...3.497...5.9.4.5..3.3..6.28....793.........3..
9.76..4..64.3......53......57...6...3.497...5.9.4.5..3....6.28....79..........3..

Although I use T&E(2) seeds now, I still get T&E(3) puzzles, like these : the first is an expanded, T&E(2), 11.7, the second a minimal (T&E(3), 11.8) of the first. I think it is because I now expand puzzles, there are more like these in my recent results. Not only is the minimal higher rated, it has become another T&E. (the puzzle is already in the T&E(3) file).
Paquita
 
Posts: 132
Joined: 11 November 2018

Re: The hardest sudokus (new thread)

Postby denis_berthier » Wed Sep 13, 2023 2:35 am

Paquita wrote:This is funny
98765.43.64.3......53......57..36...3.497...5.9.4.5..3.3..6.28....793.........3..
9.76..4..64.3......53......57...6...3.497...5.9.4.5..3....6.28....79..........3..
Although I use T&E(2) seeds now, I still get T&E(3) puzzles, like these : the first is an expanded, T&E(2), 11.7, the second a minimal (T&E(3), 11.8) of the first. I think it is because I now expand puzzles, there are more like these in my recent results. Not only is the minimal higher rated, it has become another T&E. (the puzzle is already in the T&E(3) file).


Hi Paquita
Maybe you didn't post the right 1st puzzle, but 98765.43.64.3......53......57..36...3.497...5.9.4.5..3.3..6.28....793.........3.. is in T&E(1), with SER 9.1.

Remember that for proving that a puzzle is in T&E(2) with SudoRules, you have to prove two things:
- it can be solved with T&E(2)
- it can't be solved with T&E(1)

Is the result surprising? Not that much. When starting from some non-minimal puzzle, there's not much one can say about puzzles obtained by deleting one or more givens - in particular about its minimals - except that they can't be strictly easier in the T&E hierarchy and the B, BpB or BpBB sub-classifications (which is already better than what you can say about their SER).
What would be surprising (and impossible) is the case where the first puzzle would be the expansion of the second.
.
denis_berthier
2010 Supporter
 
Posts: 4213
Joined: 19 June 2007
Location: Paris

Re: The hardest sudokus (new thread)

Postby Paquita » Wed Sep 13, 2023 8:29 am

Ok yes, the puzzle comes from this one
98.76.5..7...48.....65.9...5.9...48.47........68..4.9..57...........63.5......2..;11.7;1.2;1.2;monh
It was expanded
98.76.5..7.5.48.....65.9...5.9..748.47.....5..68.54.9..57...........63.5.....52..
and then modified.
98.7625..7.5.48.....65.9...5.9..748.47.....5..68.54.9..57...........63.5.....52..
then maxlexed
98765.43.64.3......53......57..36...3.497...5.9.4.5..3.3..6.28....793.........3..
then to min
9.76..4..64.3......53......57...6...3.497...5.9.4.5..3....6.28....79..........3..

The starting puzzle is T&E(2), 11.7.

Currently I do not check for T&E(1), is it possible that an 11.0+ puzzle is T&E(1)? In other words, should I check it?
Paquita
 
Posts: 132
Joined: 11 November 2018

Re: The hardest sudokus (new thread)

Postby hendrik_monard » Wed Sep 13, 2023 8:32 am

Hi Mith,

I put on my google drive the collection of all 55088 puzzles submitted by me from April 19th 2021 to August 15th 2023.
All lines have the same format: puzzle tab SER tab date of submission(YYMMDD)

Code: Select all
https://drive.google.com/file/d/1LxrL5R5vmYqj6f7FBEmPL5ot7OZY1pXH/view?usp=drive_link


Hendrik
Last edited by hendrik_monard on Wed Sep 13, 2023 9:02 am, edited 2 times in total.
hendrik_monard
 
Posts: 96
Joined: 19 April 2021
Location: Leuven (Louvain) Belgium

Re: The hardest sudokus (new thread)

Postby denis_berthier » Wed Sep 13, 2023 8:44 am

Paquita wrote:Currently I do not check for T&E(1), is it possible that an 11.0+ puzzle is T&E(1)? In other words, should I check it?

I would say it's extremely unlikely. The highest SER I can remember seeing in T&E(1) was 9.5 (I can be wrong about the exact value.)
However, I don't know all the arbitrary thresholds, bugs and inconsistencies that might lead to such an extraordinary situation as a 11.0+ being in T&E(1); as a result, I can't have a 100% certainty.


Indeed, I have an example in the ph2010 database:
Code: Select all
.6.7.........8......5..9.1.8...6......9.25.3..7......4.....35.1....9..2...2...3..;11.00;1.20;1.20;FLK;final2 2018_07;2112870;21; #756628

Problem is, it's not an example. The real SER is 9.2. By a pure coincidence, I was about to post about it elsewhere.
denis_berthier
2010 Supporter
 
Posts: 4213
Joined: 19 June 2007
Location: Paris

Re: The hardest sudokus (new thread)

Postby denis_berthier » Wed Sep 13, 2023 8:52 am

hendrik_monard wrote:I put on my google drive the collection of all 55088 puzzles submitted by me from April 19th 2021 to August 15th 2023.
All lines have the same format: puzzle tab SER tab date of submission(YYMMDD)
https://drive.google.com/file/d/1LxrL5R ... drive_link


Hi Hendrik
The link is broken.
denis_berthier
2010 Supporter
 
Posts: 4213
Joined: 19 June 2007
Location: Paris

Re: The hardest sudokus (new thread)

Postby hendrik_monard » Wed Sep 13, 2023 8:55 am

Paquita wrote:
Currently I do not check for T&E(1), is it possible that an 11.0+ puzzle is T&E(1)? In other words, should I check it?

Yes, possibly. I can give some examples of puzzles that were solved by my F1_Solver. I guess they are in T&E(1).
..1..23...4..1..5.5.......6...7.......7.3.8.....2.8...9.......4.6.....9...81..2.. 13739 11.20;11.20;9.90;col;H126;2049;21 98.7..6..7.........51.3....8....9.2..1.........5.....4...8.2.6.....1...5...9..27. 75249 11.10;1.20;1.20;col;2015_08;1683116;22
hendrik_monard
 
Posts: 96
Joined: 19 April 2021
Location: Leuven (Louvain) Belgium

Re: The hardest sudokus (new thread)

Postby denis_berthier » Wed Sep 13, 2023 9:05 am

hendrik_monard wrote:
Paquita wrote:
Currently I do not check for T&E(1), is it possible that an 11.0+ puzzle is T&E(1)? In other words, should I check it?

Yes, possibly. I can give some examples of puzzles that were solved by my F1_Solver. I guess they are in T&E(1).
..1..23...4..1..5.5.......6...7.......7.3.8.....2.8...9.......4.6.....9...81..2.. 13739 11.20;11.20;9.90;col;H126;2049;21 98.7..6..7.........51.3....8....9.2..1.........5.....4...8.2.6.....1...5...9..27. 75249 11.10;1.20;1.20;col;2015_08;1683116;22

No, both are in T&E(2).
denis_berthier
2010 Supporter
 
Posts: 4213
Joined: 19 June 2007
Location: Paris

Re: The hardest sudokus (new thread)

Postby hendrik_monard » Wed Sep 13, 2023 9:06 am

denis_berthier wrote:Hi Hendrik
The link is broken.

Thanks for seeing this. Apparently, the forum software did this when I submitted the post. I edited the original post and now it should be all right.
hendrik_monard
 
Posts: 96
Joined: 19 April 2021
Location: Leuven (Louvain) Belgium

Re: The hardest sudokus (new thread)

Postby eleven » Wed Sep 13, 2023 3:55 pm

Paquita wrote:...
then to min
9.76..4..64.3......53......57...6...3.497...5.9.4.5..3....6.28....79..........3..

This puzzle (T&E(3), 11.8) can be solved relatively easily with TH, oddagons, a UR and coloring (i am sure there are shorter paths).
Hidden Text: Show
After TH => 6r5c2 and basics, RT 128 r1c26,r5c6 -> -1r79c6
Code: Select all
+-------------------------+---------------------------+-------------------------+
|   9      *28      7     |   6        5      *128    | 4       3       128     |
|   6       4       128   |   3      @*128     79     | 15789   12579   12789   |
|  #128     5       3     | @#28-1     4       79     | 16789   12679   126789  |
+-------------------------+---------------------------+-------------------------+
|   5       7       128   |  *128      3       6      | 189     1249    12489   |
|   3       6       4     |   9        7       128    | 18      12      5       |
|  @128     9       128   |   4      @*128     5      | 67      67      3       |
+-------------------------+---------------------------+-------------------------+
|   47      13      59    |   15       6       34     | 2       8       79      |
| @#28      238     56    |   7        9       238    | 156     1456    146     |
|   47   @#*128     569   |@#*1258     128     248    | 3       5679    679     |
+-------------------------+---------------------------+-------------------------+

Now we have a lot of lurking 28 oddagons (size 5,7).
1r3c4 => 5r7c4, 1r9c5: kills all 1's in the *-ed 7-cell oddagon => -1r3c4 [Edit: had a wrong cell marked *]
1r9c5 => 5r7c4 => (odd #) 1r3c1 => oddagon @ => -1r9c5 [Edit: corrected @ in the grid]
Then a skyscraper in c15 eliminates 1r3c2
Code: Select all
+----------------------+----------------------+----------------------+
| 9    #*28     7      | 6      5     #128    | 4      3      128    |
| 6      4     *28     | 3     *28+1   79     | 15789  12579  12789  |
| 1      5      3      |#28     4      79     | 6789   2679   26789  |
+----------------------+----------------------+----------------------+
| 5      7      128    |#28     3      6      | 189    1249   12489  |
| 3      6      4      | 9      7      128    | 18     12     5      |
| 28     9      128    | 4     #128    5      | 67     67     3      |
+----------------------+----------------------+----------------------+
| 47     13     59     | 15     6      34     | 2      8      79     |
| 28     238    56     | 7      9      238    | 156    1456   146    |
| 47   #*28+1   569    | 15   #*28     248    | 3      5679   679    |
+----------------------+----------------------+----------------------+

The *-ed 5-cell oddagon has extra candidates 1r9c2 and 1r2c5
1r2c5 => 7-cell oddagon # => 1r9c2
=> 1r9c2
Code: Select all
+--------------------+----------------------+------------------------+
| 9      28     7    |   6      5      128  |   4      3      128    |
| 6      4      28   |   3      128    79   |   15789  1259   1278   |
| 1      5      3    | *#28     4      79   |   679-8  9-2    2678   |
+--------------------+----------------------+------------------------+
| 5      7      128  | *#28     3      6    |   189    1249   1248   |
| 3      6      4    |   9      7    *#128  |  #18    *12     5      |
| 28     9      128  |   4      128    5    |   67     67     3      |
+--------------------+----------------------+------------------------+
| 7      3      5    |   1      6      4    |   2      8      9      |
| 28     28     6    |   7      9      3    |   15     145    14     |
| 4      1      9    |   5      28     28   |   3      67     67     |
+--------------------+----------------------+------------------------+

kites for (*) and 8 (#) => -2r3c8, -8r3c7
Code: Select all
+-------------------+-------------------+--------------------+
| 9     28    7     |  6     5     128  |  4     3     128   |
| 6     4     28    |  3     128   9    | #158  #25-1  7     |
| 1     5     3     |  28    4     7    |  6     9     28    |
+-------------------+-------------------+--------------------+
| 5     7     128   |  28    3     6    |  9     124   248-1 |
| 3     6     4     |  9     7     128  |  18    12    5     |
| 28    9     128   |  4     128   5    |  7     6     3     |
+-------------------+-------------------+--------------------+
| 7     3     5     |  1     6     4    |  2     8     9     |
| 28    28    6     |  7     9     3    | #15   #145   14    |
| 4     1     9     |  5     28    28   |  3     7     6     |
+-------------------+-------------------+--------------------+

Note the UR 15 r28c67 => -1r2c8,r4c9
The rest can be solved with skyscraper and kite.
Last edited by eleven on Wed Sep 13, 2023 6:52 pm, edited 1 time in total.
eleven
 
Posts: 3151
Joined: 10 February 2008

PreviousNext

Return to General