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 tarek » Thu Nov 18, 2010 4:54 pm

I will start collating puzzles submitted to this thread after the last database publication later tonight ....

They will be automatically scrutinized over their q1,q2,Sx9 & Sxt ratings. For SE rating, I will consider the submitted SE rating (if there is one) ... I will not SE rate puzzles submitted to this thread (other than the ones I already have).

So if you've rated a puzzle that scores ER>11.6 then post it with the SE rating please.

tarek
User avatar
tarek
 
Posts: 2624
Joined: 05 January 2006

Re: The hardest sudokus (new thread)

Postby ronk » Thu Nov 18, 2010 5:24 pm

tarek wrote:So if you've rated a puzzle that scores ER>11.6 then post it with the SE rating please.

That bar seems mighty high. Did you perhaps mean 10.6?
ronk
2012 Supporter
 
Posts: 4764
Joined: 02 November 2005
Location: Southeastern USA

Re:

Postby tarek » Thu Nov 18, 2010 7:14 pm

ronk wrote:
tarek wrote:So if you've rated a puzzle that scores ER>11.6 then post it with the SE rating please.

That bar seems mighty high. Did you perhaps mean 10.6?
It is very high, the idea is to limit the number of records in the datbase.

Previously, I wrote:Suggested database filter:
Code: Select all
Entries should have:
Q1  >=99000 OR
Q2  >=99100 OR
SX9 >=4200 OR
SXT >=2000 OR
ER  >=11.6
These figures were based on my personal collection.
The ER threshold is as high as other thresholds. If a puzzle fails the ER threshold then it can still make the list by passing ANY of the other thresholds.

The current database records before the update stand at 86 ... I didn't receive any suggestions regarding these filters/thresholds since I originally posted them. The idea was to make sure that the hardest puzzle is in the database. If a puzzle fails all criteria then IMO it can't be the hardest puzzle.

tarek
User avatar
tarek
 
Posts: 2624
Joined: 05 January 2006

Hardest Database Update

Postby tarek » Thu Nov 18, 2010 9:07 pm

The following puzzles have been added to the database as they passed the SE threshold, All other puzzles posted on this thread since the last database was posted didn't pass any of the thresholds.

]PuzzleR Poster ER EP ED
..9...4...7.3...2.8...6...71..8....6....1..7.....56...3....5..1.4.....9...2...7.. eleven 11.8 11.8 11.6
5.......9.2.1...7...8...3...4.6.........5.......2.7.1...3...8...6...4.2.9.......5 m_b_metcalf 11.8 11.8 3.4
....567.....1.....6..37......4.....13...6.9...8.....2.5....73....2.....8.1.....4. eleven 11.7 11.7 11.2
1...5...9.......3..79........7.1...65....42..6..8.......5.6...7.....2.4....3..8.. eleven 11.7 1.2 1.2
5.......9.2.1...7...8...3...4...2.......5.......7.6.1...3...8...6...4.2.9.......5 StrmCkr 11.6 11.6 3.4

Here is the latest database with a gsf friendly 1st line:

List deleted to avoid order confusion with recent updated form of list


tarek
{EDIT: List deleted to avoid order confusion with recent updated form of list]
Last edited by tarek on Thu Dec 30, 2010 11:44 am, edited 1 time in total.
User avatar
tarek
 
Posts: 2624
Joined: 05 January 2006

Re: The hardest sudokus (new thread)

Postby champagne » Sat Nov 27, 2010 9:31 am

Code: Select all
9.....8..
...7...6.
....8.5.4
.3.8.....
4...2.9..
..1..6.3.
..41.....
.6...7...
5...4...2   swampy soil


despite a relatively low SE rating

ED=10.8/10.8/10.0

That puzzle is seen by my solver as tough as "Golden Nugget" and "Silver Plate"

But it has no SK loop and no EXOCET pattern.

As a matter of fact, it would be for the time being my toughest puzzle.

champagne

EDIT : I omitted to check the rank 0 logic, it has one, what could lower significantly the final complexity
champagne
2017 Supporter
 
Posts: 5720
Joined: 02 August 2007
Location: France Brittany

Re: swampy soil checks

Postby tarek » Sat Nov 27, 2010 1:10 pm

Hi Champagne,

I've checked that puzzle:
Code: Select all
22 clues minimal
q1 95243
q2 98057
sx9 1784
sxt 1161
Singles Backdooor size 2
FNBHTWXYKO Backdoor size 1
It scores high on the q2 rating but overall relatively lower than the difficilt puzzles that you've mentioned.

tarek
User avatar
tarek
 
Posts: 2624
Joined: 05 January 2006

Re: swampy soil checks

Postby champagne » Sat Nov 27, 2010 3:23 pm

tarek wrote:Hi Champagne,

I've checked that puzzle:
Code: Select all
22 clues minimal
q1 95243
q2 98057
sx9 1784
sxt 1161
Singles Backdooor size 2
FNBHTWXYKO Backdoor size 1
It scores high on the q2 rating but overall relatively lower than the difficilt puzzles that you've mentioned.

tarek


Hi tarek,

It's not new for me that correlation between different rating index is poor for "hardest puzzles".

The rating done by my solver is in line with the set of rule in use in it.

It would be interesting to compare paths, but it would be a boring task. The lowest print size I got (using the rank 0 logic) is still over 400K.

Generally speaking, in that current game, the highest rating given by SE are by far not the hardest seen by my solver.

May-be some players will give their own feeling.

champagne
champagne
2017 Supporter
 
Posts: 5720
Joined: 02 August 2007
Location: France Brittany

Re: swampy soil checks

Postby tarek » Sat Nov 27, 2010 7:52 pm

champagne wrote:It's not new for me that correlation between different rating index is poor for "hardest puzzles".

The rating done by my solver is in line with the set of rule in use in it.

I can understand it when the q1,q2,sx9,sxt,ER collective over-rates a puzzle (Similar to a sensitive but not specific test) ....

But you are saying that the hardest puzzle can be missed by these tests !! I am intersted to see if that turns out to be true ... I still stand by the opinion that the hardest puzzle should be detected by at least one of these tests.

tarek
User avatar
tarek
 
Posts: 2624
Joined: 05 January 2006

HS

Postby champagne » Sun Nov 28, 2010 4:33 am

tarek wrote:
champagne wrote:It's not new for me that correlation between different rating index is poor for "hardest puzzles".

The rating done by my solver is in line with the set of rule in use in it.

I can understand it when the q1,q2,sx9,sxt,ER collective over-rates a puzzle (Similar to a sensitive but not specific test) ....

But you are saying that the hardest puzzle can be missed by these tests !! I am intersted to see if that turns out to be true ... I still stand by the opinion that the hardest puzzle should be detected by at least one of these tests.

tarek


AFAIK None of these solver detect the SK loop, nor an Exocet, nor a rank o logic, some tools to crack easily very hard puzzles.

Even Golden Nugget and Silver Plate are solved thru "relatively easy" path using the Exocet property. (published)

I remain convinced that the final rating should apply the best known technique than can be applied to solve a puzzle.

My statement regarding "swampy soil" is just that

1) using my basic set of rules, (tagging with AHS AAHS), no easy path comes and the path is very complex and boring, the same order of magnitude as for Golden Nugget/ Silver Plate

2) Using all the set of special tools, I make very little progress

3) unless somebody comes with fresh stuff, it is at the end the toughest puzzle I have.

For sure, a new tool will come one day and change the relative rating of all these puzzles. The last siginficant change came form Allan Barker attack of puzzles with, among others, the rank o logic.

Let me take a reverse example with that statement from Mauricio


Code: Select all
To solve it, I only need to suppose uniqueness and to use not more than hidden pair.
+-------+-------+-------+
| . . . | . . 1 | . . 2 |
| . 1 . | . 3 . | . 4 . |
| . . 5 | 6 . . | . . . |
+-------+-------+-------+
| . . 2 | 7 . . | 8 . . |
| . 8 . | . . 4 | . 3 . |
| 1 . . | . 9 . | . . . |
+-------+-------+-------+
| . . . | 3 . . | 5 . . |
| . 9 . | . 8 . | . 1 . |
| 6 . . | . . . | . . 7 |
+-------+-------+-------+ ER=11.3/11.3/10.5


can you tell me which of your rating index shows that this is an easy puzzle :D
(my solver sees the symmetry of given)

champagne
champagne
2017 Supporter
 
Posts: 5720
Joined: 02 August 2007
Location: France Brittany

Re: HS

Postby RW » Sun Nov 28, 2010 9:16 am

champagne wrote:
Code: Select all
To solve it, I only need to suppose uniqueness and to use not more than hidden pair.
+-------+-------+-------+
| . . . | . . 1 | . . 2 |
| . 1 . | . 3 . | . 4 . |
| . . 5 | 6 . . | . . . |
+-------+-------+-------+
| . . 2 | 7 . . | 8 . . |
| . 8 . | . . 4 | . 3 . |
| 1 . . | . 9 . | . . . |
+-------+-------+-------+
| . . . | 3 . . | 5 . . |
| . 9 . | . 8 . | . 1 . |
| 6 . . | . . . | . . 7 |
+-------+-------+-------+ ER=11.3/11.3/10.5

My symmetry techniques must be a bit rusty... After basic symmetry and SSTS I get stuck here:

Code: Select all
 *-----------------------------------------------------------*
 | 7     36    89    | 49    45    1     | 36    58    2     |
 | 28    1     69    | 59    3     278   | 67    4     568   |
 | 34    24    5     | 6     27    28    | 1     79    389   |
 |-------------------+-------------------+-------------------|
 | 49    45    2     | 7     6     3     | 8     59    1     |
 | 59    8     67    | 2     1     4     | 679   3     569   |
 | 1     367   36    | 8     9     5     | 246   267   46    |
 |-------------------+-------------------+-------------------|
 | 28    27    1     | 3     247   269   | 5     269   4689  |
 | 35    9     47    | 45    8     267   | 246   1     346   |
 | 6     235   348   | 1     245   29    | 2349  289   7     |
 *-----------------------------------------------------------*

What is the next move?

RW
RW
2010 Supporter
 
Posts: 1000
Joined: 16 March 2006

Re: HS

Postby champagne » Sun Nov 28, 2010 10:06 am

RW wrote:My symmetry techniques must be a bit rusty... After basic symmetry and SSTS I get stuck here:

Code: Select all
 *-----------------------------------------------------------*
 | 7     36    89    | 49    45    1     | 36    58    2     |
 | 28    1     69    | 59    3     278   | 67    4     568   |
 | 34    24    5     | 6     27    28    | 1     79    389   |
 |-------------------+-------------------+-------------------|
 | 49    45    2     | 7     6     3     | 8     59    1     |
 | 59    8     67    | 2     1     4     | 679   3     569   |
 | 1     367   36    | 8     9     5     | 246   267   46    |
 |-------------------+-------------------+-------------------|
 | 28    27    1     | 3     247   269   | 5     269   4689  |
 | 35    9     47    | 45    8     267   | 246   1     346   |
 | 6     235   348   | 1     245   29    | 2349  289   7     |
 *-----------------------------------------------------------*

What is the next move?

RW




Mauricio produced at that time 2 puzzles.

I did not pay enough attention and I took the second one, not solved so easily.
(One should check that the statement refers to that one and not only to the first one)

At the point where you are locked, my solver produces an AIC using the symmetry property.

Here is the tagged PM with symmetry


Code: Select all
__ A___ B____ C____ |D___ E____ F_____ |G_____ H_____ I_____
1||7___ 3A6a_ 8a9A_ |4A9a 4a5A_ 1_____ |3a6A__ 5a8A__ 2_____
2||2a8A 1____ 6A9a_ |5a9A 3____ 2A7G8_ |6G7g__ 4_____ 5A68__
3||3a4A 2A4a_ 5____ |6___ 2G7g_ 28____ |1_____ 7G9g__ 3A89G_

4||4a9A 4A5a_ 2____ |7___ 6____ 3_____ |8_____ 5A9a__ 1_____
5||5A9a 8____ 6G7g_ |2___ 1____ 4_____ |67G9f_ 3_____ 5a69__
6||1___ 36A7G 36___ |8___ 9____ 5_____ |2b46__ 2B6c7g 46____

7||2A8a 2G7g_ 1____ |3___ 24f7G 26b9__ |5_____ 26C9__ 4F68A9
8||3A5a 9____ 4g7G_ |4a5A 8____ 2c6B7g |2C46__ 1_____ 3a46__
9||6___ 235A_ 34G8A |1___ 245a_ 29____ |23A49F 28a9__ 7_____


and the AIC killing all 'A' tagged candidates

[]2r2c6-A/7r2c6-G_7r6c8-g/2r6c8-B_2r6c7-b/2r8c7-C_2r8c6-c/2r2c6-A

champagne
champagne
2017 Supporter
 
Posts: 5720
Joined: 02 August 2007
Location: France Brittany

Re: HS

Postby RW » Sun Nov 28, 2010 12:28 pm

champagne wrote:Mauricio produced at that time 2 puzzles.

I did not pay enough attention and I took the second one, not solved so easily.

Okay that explains it, thanks. I also ended up using a simple chain (not using the symmetry property):

[r9c8]-2-[r8c7]=2=[r8c6]-2-[r2c6]=2=[r2c1]=8=[r7c1]-8-[r9c3]=8=[r9c8]

Not very complicated, but certainly more complex than a hidden pair.

RW
RW
2010 Supporter
 
Posts: 1000
Joined: 16 March 2006

Re: The hardest sudokus (new thread)

Postby eleven » Sun Nov 28, 2010 6:21 pm

Champagne,

because this puzzle has a pattern different to the known hardest, i would be interested, how you classify it (ER is 11.5):
Code: Select all
 +-------+-------+-------+
 | . . 3 | . . 6 | . 8 . |
 | . . . | 1 . . | 2 . 6 |
 | 6 . . | . 7 . | . . 5 |
 +-------+-------+-------+
 | . . 4 | . . . | . . . |
 | 3 . . | . . 4 | . 9 . |
 | . 9 8 | . 2 . | . . . |
 +-------+-------+-------+
 | . 3 . | . . 8 | . 6 . |
 | 8 . . | . . . | 1 . 7 |
 | . . . | 5 . . | . . . |
 +-------+-------+-------+
eleven
 
Posts: 1578
Joined: 10 February 2008

Re: The hardest sudokus (new thread)

Postby champagne » Sun Nov 28, 2010 7:08 pm

eleven wrote:Champagne,

because this puzzle has a pattern different to the known hardest, i would be interested, how you classify it (ER is 11.5):
Code: Select all
 +-------+-------+-------+
 | . . 3 | . . 6 | . 8 . |
 | . . . | 1 . . | 2 . 6 |
 | 6 . . | . 7 . | . . 5 |
 +-------+-------+-------+
 | . . 4 | . . . | . . . |
 | 3 . . | . . 4 | . 9 . |
 | . 9 8 | . 2 . | . . . |
 +-------+-------+-------+
 | . 3 . | . . 8 | . 6 . |
 | 8 . . | . . . | 1 . 7 |
 | . . . | 5 . . | . . . |
 +-------+-------+-------+



I checked your puzzle and I made a run including the former 11.8

009000400070300020800060007100800006000010070000056000300005001040000090002000700 11.8
003006080000100206600070005004000000300004090098020000030008060800000107000500000 11.5

My first run is done excluding all special tools (symmetry, SK loop, EXOXET, R0 logic)

The 2 puzzles are in the same area
. print around 250K
. Maximum stuff used AAHS size 2 cells
. 7/8 nested derived weak links at maximum.

The second one seems slightly harder than the first one.

If you want to compare with the last one i posted, that one gave the following results

. print over 600K
. necessity to use AAHS over 2 cells
. but a maximum of 4 nested derived links;


If we now look for special tools

I think I already noticed that your 11.8 puzzle has something close to an EXOCET (r56c4). The second one has an EXOCET (r46c1).
None have the SK loop nor a rank0 logic (i did not check for more than 4 floors).

Using the adequate special tool, the print is significantly reduced, but still around 150K.


champagne
champagne
2017 Supporter
 
Posts: 5720
Joined: 02 August 2007
Location: France Brittany

Re: The hardest sudokus (new thread)

Postby ronk » Sun Nov 28, 2010 8:30 pm

champagne wrote: checked your puzzle and I made a run including the former 11.8

009000400070300020800060007100800006000010070000056000300005001040000090002000700 11.8
003006080000100206600070005004000000300004090098020000030008060800000107000500000 11.5
...
If we now look for special tools

I think I already noticed that your 11.8 puzzle has something close to an EXOCET (r56c4). The second one has an EXOCET (r46c1).
None have the SK loop nor a rank0 logic (i did not check for more than 4 floors).

Using the adequate special tool, the print is significantly reduced, but still around 150K.

For the ER=11.5 puzzle, what "special tool" was used ... and what were the exclusions?
ronk
2012 Supporter
 
Posts: 4764
Joined: 02 November 2005
Location: Southeastern USA

PreviousNext

Return to General