The New Sudoku Players' Forum

[champagne]

I just got that puzzle with the highest known SE rating

Code: Select all

98.7.....7.....6....6.5.....4...5.3...79..5......2...1..85..9......1...4.....3.2. ;11.9;11.9;11.8


Well done. I couldn't understand if this was a very recent find or that it had been there for some time but posted elsewhere.

tarek

It came out of my search in the 22 clues field. It has been generated yesterday and rated this morning. (some hours to get the result!!!)
It should be a new one

champagne

[tarek]

It came out of my search in the 22 clues field. It has been generated yesterday and rated this morning. (some hours to get the result!!!)
It should be a new one

Great, I'll take your word for it. I'll be using your rating for the time being & it should be in the next update.

tarek

[eleven]

Congratulations, champagne, you seem to have better filters than me

I found this puzzle and 649 of your list (including the first 11) in my collection of 4.5 mio "step 1" 22 clues, but never rated them.

Note the poor q2 ratings of some puzzles in your list, e.g. your puzzle #625 (ER 11.1) has q2 412 - like a simple newspaper puzzle.

[champagne]

Congratulations, champagne, you seem to have better filters than me

I found this puzzle and 649 of your list (including the first 11) in my collection of 4.5 mio "step 1" 22 clues, but never rated them.

Note the poor q2 ratings of some puzzles in your list, e.g. your puzzle #625 (ER 11.1) has q2 412 - like a simple newspaper puzzle.

So it's not really a new one. As you are a native English speaker, I let you find a name for it.

Regarding the filters, it's very simple. I use now the SE clone skfr.

It's working properly up to the area 10.5 _ 11 with a speed ratio exceeding 1 : 100 in the upper field.

I worked exclusively with skfr in the game 146 and got some rating equalling SE rating +-0.2 for ratings over 10.

As you likely know, it's not yet an operational program, but if you want to test it, you can get a copy of the code (or you can join the team to help developing).

My target is to give up as soon as possible the SE rating and to replace it by skfr rating, but I have still 3 steps to implement and many tests to achieve.

The team will restart the work after the holidays period.

champagne

[champagne]

if I am right we were missing a 11.4 diamond.

Here is one

Code: Select all

98.7.....6...8.7....7..5....4.....3...86..5......2...1..98..6......3..4......1..2 ;11.4;11.4;11.4


champagne

[ttt]

Hi champagne,
Congrats!
And what it's name? suggestion: "puzzle for ttt", please .

[champagne]

Hi champagne,
Congrats!
And what it's name? suggestion: "puzzle for ttt", please .

Hi "ttt"

Nice to see you again.

I think it worth a name for historical reasons, but you should break it easily.

Let's eleven make a proposal for the name, he also produced it.

I am interested to know if you find something using XSUDO and the UR option.

I identified during the search some puzzles that seem very hard to break, but I have to rework the entire data base to sort them out, I am lacking time to do that for the time being.

Let us know what you find with that one.

cheers

champagne

[eleven]

So it's not really a new one. As you are a native English speaker, I let you find a name for it.

Of course it is a new one. Though i had it in my collection, i never saw it So i think, it's up to you to give it a name. If you want a suggestion, i would call it "Champagne dry" (btw my English is based on dict.leo.org), which beats a Golden Nugget, when you are thirsty.

Well done with the skfr. Hope you will manage to improve the Explainer rating with a new standard.

Personally i have stopped all sudoku activities after my hardest search (apart from solving newspaper puzzles). Let's see, if i will get hungry again.

[dobrichev]

I just got that puzzle with the highest known SE rating

Code: Select all

98.7.....7.....6....6.5.....4...5.3...79..5......2...1..85..9......1...4.....3.2. ;11.9;11.9;11.8


Congratulations!

The puzzle has 0 cells solvable with all but one of the givens. Maximal sizes of unavoidables hit by a single clue are between 46 and 55.

Hidden Text: Show

Code: Select all

98.7.....7.....6....6.5.....4...5.3...79..5......2...1..85..9......1...4.....3.2.       22
        1       0       36      46      #Clue,pos,numUA,maxUAsize
        2       1       60      52      #Clue,pos,numUA,maxUAsize
        3       3       134     50      #Clue,pos,numUA,maxUAsize
        4       9       42      52      #Clue,pos,numUA,maxUAsize
        5       15      81      53      #Clue,pos,numUA,maxUAsize
        6       20      97      54      #Clue,pos,numUA,maxUAsize
        7       22      54      48      #Clue,pos,numUA,maxUAsize
        8       28      89      54      #Clue,pos,numUA,maxUAsize
        9       32      8       55      #Clue,pos,numUA,maxUAsize
        10      34      119     50      #Clue,pos,numUA,maxUAsize
        11      38      135     52      #Clue,pos,numUA,maxUAsize
        12      39      211     53      #Clue,pos,numUA,maxUAsize
        13      42      18      47      #Clue,pos,numUA,maxUAsize
        14      49      64      50      #Clue,pos,numUA,maxUAsize
        15      53      93      55      #Clue,pos,numUA,maxUAsize
        16      56      212     53      #Clue,pos,numUA,maxUAsize
        17      57      93      49      #Clue,pos,numUA,maxUAsize
        18      60      91      52      #Clue,pos,numUA,maxUAsize
        19      67      116     49      #Clue,pos,numUA,maxUAsize
        20      71      68      52      #Clue,pos,numUA,maxUAsize
        21      77      51      50      #Clue,pos,numUA,maxUAsize
        22      79      156     54      #Clue,pos,numUA,maxUAsize
                        2028    55      0       #numUA,maxUAsize,nSolved

[champagne]

So it's not really a new one. As you are a native English speaker, I let you find a name for it.

Of course it is a new one. Though i had it in my collection, i never saw it So i think, it's up to you to give it a name. If you want a suggestion, i would call it "Champagne dry" (btw my English is based on dict.leo.org), which beats a Golden Nugget, when you are thirsty.

Well done with the skfr. Hope you will manage to improve the Explainer rating with a new standard.

Personally i have stopped all sudoku activities after my hardest search (apart from solving newspaper puzzles). Let's see, if i will get hungry again.

Your suggestion is accepted.


I understand your position, but in case you would come back, we could share the task to dig in the 22 23 area where i suspect that the best puzzles are.

You made such a good job that I will suffer to produce truly new puzzles. On the other side, i learned from the pattern game that you have to be patient. Most often, I got the best puzzles after 8 to 10 cycles.

champagne

[ronk]

The puzzle has 0 cells solvable with all but one of the givens.

Would you please rephrase that so as to clarity its meaning? TIA

[champagne]

I just got that puzzle with the highest known SE rating

Code: Select all

98.7.....7.....6....6.5.....4...5.3...79..5......2...1..85..9......1...4.....3.2. ;11.9;11.9;11.8


Congratulations!

The puzzle has 0 cells solvable with all but one of the givens. Maximal sizes of unavoidables hit by a single clue are between 46 and 55.

Hidden Text: Show

Code: Select all

98.7.....7.....6....6.5.....4...5.3...79..5......2...1..85..9......1...4.....3.2.       22
        1       0       36      46      #Clue,pos,numUA,maxUAsize
        2       1       60      52      #Clue,pos,numUA,maxUAsize
        3       3       134     50      #Clue,pos,numUA,maxUAsize
        4       9       42      52      #Clue,pos,numUA,maxUAsize
        5       15      81      53      #Clue,pos,numUA,maxUAsize
        6       20      97      54      #Clue,pos,numUA,maxUAsize
        7       22      54      48      #Clue,pos,numUA,maxUAsize
        8       28      89      54      #Clue,pos,numUA,maxUAsize
        9       32      8       55      #Clue,pos,numUA,maxUAsize
        10      34      119     50      #Clue,pos,numUA,maxUAsize
        11      38      135     52      #Clue,pos,numUA,maxUAsize
        12      39      211     53      #Clue,pos,numUA,maxUAsize
        13      42      18      47      #Clue,pos,numUA,maxUAsize
        14      49      64      50      #Clue,pos,numUA,maxUAsize
        15      53      93      55      #Clue,pos,numUA,maxUAsize
        16      56      212     53      #Clue,pos,numUA,maxUAsize
        17      57      93      49      #Clue,pos,numUA,maxUAsize
        18      60      91      52      #Clue,pos,numUA,maxUAsize
        19      67      116     49      #Clue,pos,numUA,maxUAsize
        20      71      68      52      #Clue,pos,numUA,maxUAsize
        21      77      51      50      #Clue,pos,numUA,maxUAsize
        22      79      156     54      #Clue,pos,numUA,maxUAsize
                        2028    55      0       #numUA,maxUAsize,nSolved

I did not work on that side, but I am not surprised that that puzzle has specific properties.
The rules applied in Sudoku Explainer are "relatively basic" and fit well with such analysis.

As I mentioned earlier, I suspect that the 22 23 clues field will produce the best puzzles toward such properties. The problem is to sort them out.

I would be interested later on to have an idea of the specific properties of the puzzles that appear really as the toughest to my solver.

BTW, I am convinced that the next breakthrough in the puzzle generation will be based on a clear view of the qualities of the puzzles to generate . Here, "you all" have a key role to play.

champagne

[eleven]

The puzzle has 0 cells solvable with all but one of the givens.

When i understand that right, you cannot place a number without regarding all of the givens. Is this a rare property under the hardest puzzles ?

[dobrichev]

The puzzle has 0 cells solvable with all but one of the givens.

When i understand that right, you cannot place a number without regarding all of the givens.

Yes.

Is this a rare property under the hardest puzzles ?

I am not expert in the hard puzzles area. Few pages before there was some discussion, preceded by discussion here. There is a fast tool available and you can do analysis. I published results from one of the patterns game - see the links.

Cheers,
MD

[eleven]

Thanks, so it will not be much work to find it out.

I will try to adopt your program (i want to add the numbers you can place with n-1 clues and see what puzzles remain unsolved, with which UR sizes), when i have more time again.