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 » Thu Sep 23, 2010 4:20 pm

champagne wrote:sorry eleven, I am somehow lost.
I reacted to the eliminations 3r5c2 and 3r8c5.

Ah, i already said, that they cant be eliminated from the loop.

In a simple SK loop with exactly 2 digits for 2 cells (without any ORs ) you can directly eliminate a common pair (like 37 in column 2) from the rest of the unit. But i saw later, that here the reverse loop has ORs in other units and leaves more possibilities open for column 2 (or row 8).

However i can find all of Ron's eliminations manually as described (its just a bit boring). I doubt, that this would be so easy using your methods.
eleven
 
Posts: 1561
Joined: 10 February 2008

Re: The hardest sudokus (new thread)

Postby champagne » Thu Sep 23, 2010 4:28 pm

eleven wrote:However i can find all of Ron's eliminations manually as described (its just a bit boring). I doubt, that this would be so easy using your methods.


:?: :?: :?:

2 hidden pairs and a basic nice loop with not more than 2 AHS/AC

2r8c13 - 2r8c79 = 2r79c8 - 2r13c8 = AC:r1c7r3c9 (2r1c7r3c9 - 7r1c7r3c9) = 7r2c79 - 7r2c13 = 7r13c2 - 7r79c2 = AC:r7c1r9c3 (7r7c1r9c3 - 2r7c1r9c3)

as shown above, what simpler can we expect.

All the eliminations of the rank 0 logic shown by ronk are there.

I really don't catch your point

champagne
champagne
2017 Supporter
 
Posts: 5677
Joined: 02 August 2007
Location: France Brittany

Re: The hardest sudokus (new thread)

Postby eleven » Thu Sep 23, 2010 5:27 pm

Since i know SK-loops, i can find them in 10 minutes.
But why should i look for your loop over more than 20 cells including 2 AHS/AC ? Before i try to find something like that, i have spent 2 days looking for shorter chains.
eleven
 
Posts: 1561
Joined: 10 February 2008

Re: The hardest sudokus (new thread)

Postby champagne » Thu Sep 23, 2010 5:49 pm

eleven wrote:Since i know SK-loops, i can find them in 10 minutes.
But why should i look for your loop over more than 20 cells including 2 AHS/AC ? Before i try to find something like that, i have spent 2 days looking for shorter chains.



Basically no comment on such a statement. Each player has it's own preferences.

Only one remark: that nice loop covers exactly (in the form using ALS) the cells of the potential SK loop (I stick to a strict definition, so, for me, we have no SK loop here)

This gives an additionnal chance to find it.

Generally speaking, in puzzles having such symmetries, Nice loops when they do exist are often similar to that one as far as I could see.

champagne
champagne
2017 Supporter
 
Posts: 5677
Joined: 02 August 2007
Location: France Brittany

Re: The hardest sudokus (new thread)

Postby ronk » Thu Sep 23, 2010 6:57 pm

FWIW, this is the pencilmark grid at the time of the Explainer's ER=11.3 move, r3c9<>2.

Code: Select all
1.......9.4...2.8...6...3.....4.3.7.....6.....2.5.8.....9...1...8.7...4.5.......6 # ED=11.3/11.3/3.8 gsf-game0119

 1      357    28     | 368    34578  4567   | 247    256    9
 379    4      357    | 169    1359   2      | 567    8      157
 28     579    6      | 189    4578   14579  | 3      125    47-2
----------------------+----------------------+---------------------
 689    169    158    | 4      129    3      | 2589   7      1258
 3478   1359   34578  | 129    6      179    | 2458   1359   23458
 3479   2      1347   | 5      179    8      | 469    169    134
----------------------+----------------------+---------------------
 247    367    9      | 2368   23458  456    | 1      235    78
 236    8      123    | 7      1359   169    | 259    4      235
 5      137    247    | 12389  2348   149    | 78     239    6
ronk
2012 Supporter
 
Posts: 4764
Joined: 02 November 2005
Location: Southeastern USA

Re: The hardest sudokus (new thread)

Postby eleven » Thu Sep 23, 2010 7:56 pm

champagne wrote:Basically no comment on such a statement. Each player has it's own preferences.

A last comment on this:
Any manual player, who would not check this grid for an SK loop, is not advanced. As i said, such a check can be done in 10 minutes. Do you know, how long it takes, to check such a grid for finned fishes on paper ? I guess you dont, but i can tell you, longer, whatever preferences you have.
So for humans SK eliminations have to be rated easier. That means, that in all these hardest puzzles with SK loop (including non simple ones) you should make the eliminations, before you even start your rating.

Its a pity, that there are no players anymore on the players forum.
eleven
 
Posts: 1561
Joined: 10 February 2008

Re: The hardest sudokus (new thread)

Postby champagne » Thu Sep 23, 2010 8:10 pm

ronk wrote:FWIW, this is the pencilmark grid at the time of the Explainer's ER=11.3 move, r3c9<>2.


After the second rank 0 logic (found manually) , I have the same PM plus 6r4c7 and 6r6c1.

My solver starts then using the last level of tagging, which reflects usually a significant step in difficulty, but I had some much lower SE ratings at that level.

No chance anyway to compare solutions at that level, processes are to different.

champagne
champagne
2017 Supporter
 
Posts: 5677
Joined: 02 August 2007
Location: France Brittany

Re: The hardest sudokus (new thread)

Postby champagne » Fri Sep 24, 2010 6:55 am

after ronk's analysis of SE hardest step in that puzzle, I went deeper in the rest of the path.
next elimination for my solver is

6r4c7;6r6c1 (after several steps at the highest level of tagging),

giving the same PM.

At that moment, the solver makes a step with 10 cycles in the search of derived weak links.

The maximum number of cycles in the search of derived weak links seems to be one of the best parameters to correlate SE rating with the results of my solver.

In fact, in the last pattern games where I participated, I sorted out puzzles to rate using 3 parameters

. level of tagging
. maximum number of cycles in the search of derived weak links
. print size.

it worked pretty well.

next steps don't eliminate 2r3c9. I find in sequence

3r5c8;3r8c9
5r1c8
7r2c9;7r7c1
4r5c1
7r5c3
4r7c6

champagne
champagne
2017 Supporter
 
Posts: 5677
Joined: 02 August 2007
Location: France Brittany

Re: The hardest sudokus (new thread)

Postby eleven » Thu Oct 28, 2010 2:06 pm

A first contribution to the hardest list:
Code: Select all
 +-------+-------+-------+
 | . . 9 | . . . | 4 . . |
 | . 7 . | 3 . . | . 2 . |
 | 8 . . | . 6 . | . . 7 |
 +-------+-------+-------+
 | 1 . . | 8 . . | . . 6 |
 | . . . | . 1 . | . 7 . |
 | . . . | . 5 6 | . . . |
 +-------+-------+-------+
 | 3 . . | . . 5 | . . 1 |
 | . 4 . | . . . | . 9 . |
 | . . 2 | . . . | 7 . . |
 +-------+-------+-------+
 eleven's 11.8
eleven
 
Posts: 1561
Joined: 10 February 2008

Re: The hardest sudokus (new thread)

Postby champagne » Thu Oct 28, 2010 3:32 pm

eleven wrote: eleven's 11.8


fine, I'll check my data base (so far I have 10 ER 11.8 in the data base but have not finished the rating)

It would be better to get ER EP ED could you edit your post

thanks

champagne
Last edited by champagne on Thu Oct 28, 2010 4:43 pm, edited 1 time in total.
champagne
2017 Supporter
 
Posts: 5677
Joined: 02 August 2007
Location: France Brittany

Re: The hardest sudokus (new thread)

Postby champagne » Thu Oct 28, 2010 3:44 pm

I had a quick look at the path proposed by the solver for that puzzle.

No SK loop, no pure EXOCET, but something very close (I voluntarily don't say where).

At the end, the path is "relatively easy".

It would be a good puzzle for "ttt"

champagne
champagne
2017 Supporter
 
Posts: 5677
Joined: 02 August 2007
Location: France Brittany

Re: The hardest sudokus (new thread)

Postby eleven » Thu Oct 28, 2010 4:30 pm

champagne wrote:fine, I'll check my data base (so far I have 10 ER 10.8 in the data base but have not finished the rating)

Hope you mean 11.8, otherwise i copied the wrong puzzle.

It would be better to get ER EP ED could you edit your post

How can i ? I should change this soon, it takes long to rate collections with an average rating of 10.5.
eleven
 
Posts: 1561
Joined: 10 February 2008

Re: The hardest sudokus (new thread)

Postby champagne » Thu Oct 28, 2010 4:50 pm

eleven wrote:
champagne wrote:fine, I'll check my data base (so far I have 10 ER 10.8 in the data base but have not finished the rating)

Hope you mean 11.8, otherwise i copied the wrong puzzle.

It would be better to get ER EP ED could you edit your post

How can i ? I should change this soon, it takes long to rate collections with an average rating of 10.5.


1) I edited my post, it was 11.8.

2) I don't catch your point. If you have the ER, you normally got the EP;ED as well.

3) Regarding the time requested to rate puzzles 10.5 and above, all the players of the pattern game know that. :D
My computer started the rating of my "hardest" data base may be 2 months ago (with interruptions due to the pattern game). I hope to have finished that run in the near future and restart then the search for new hardest.

champagne
champagne
2017 Supporter
 
Posts: 5677
Joined: 02 August 2007
Location: France Brittany

Re: The hardest sudokus (new thread)

Postby eleven » Thu Oct 28, 2010 6:19 pm

Sorry, i dont even know, what EP and ED stands for. One of them is the rating of the first step, if i remember right.

My Explainer version just gives me the rating and the techniques used.

[Added:] Ah, i see, there is an explanation in the Patterns game intro. So whats the command line for serate you would like ?
eleven
 
Posts: 1561
Joined: 10 February 2008

Re: The hardest sudokus (new thread)

Postby champagne » Thu Oct 28, 2010 6:39 pm

eleven wrote:Sorry, i dont even know, what EP and ED stands for. One of them is the rating of the first step, if i remember right.

My Explainer version just gives me the rating and the techniques used.

[Added:] Ah, i see, there is an explanation in the Patterns game intro. So whats the command line for serate you would like ?


that one works

java -Xrs -Xmx500m -cp SudokuExplainer.jar diuf.sudoku.test.serate -f%g;%r;%p;%d;%e -i"puz.txt" -o"puz_rat.txt"


-f is the format command, here a puzzle ouput is of the form
puzzle;er;ep;ed;elapsedtime

-i is the input file command followed by the name (must be between quotes if there is a blank in the name)

-o is the output file command with the same remark.

note: I guess that if you want to insert blanks in the format, it must be within quotes as well eg:

-f"%g ED=%r/%p/%d %e" should give

puzzle ED=er/ep/ed elapsedtime


champagne
champagne
2017 Supporter
 
Posts: 5677
Joined: 02 August 2007
Location: France Brittany

PreviousNext

Return to General