The New Sudoku Players' Forum

by **eleven** » Thu Aug 04, 2011 12:12 pm

champagne wrote:I made a preliminary processing of eleven's entire file of hardests.

I found in my database some of them with the following origin

If you could post the puzzles in my list, which have been published earlier, i will remove them. I guess, that they have an ER less 11 (?).

Please also note, that i found these 103 non minimal puzzles in my list.

Hidden Text: Show

Dont ask me, how they got in there

Their minimals give 25 new 11+ up to 11.4 ( 68 10.6+) puzzles, which i want to add to the list at the weekend, together with a few 100 new dob-0/11+ puzzles. Apart from "Second flush" i did not find one with ER higher than 11.7. About 1500 are still to rate, this should be finished then.

by **champagne** » Thu Aug 04, 2011 5:56 pm

eleven wrote:If you could post the puzzles in my list, which have been published earlier, i will remove them. I guess, that they have an ER less 11 (?).

Please also note, that i found these 103 non minimal puzzles in my list.

About 1500 are still to rate, this should be finished then.

1) adjusting the database of hardest to clean the morphs is a boring task. I suggest to do it through PM.

2) I just loaded the entire tarek pearly database. It adds new redundancy.

At that point, I am wondering whether it should not be time to build a shared data base with all hardest with the "first identified source".
The open question is "where and how".

3) I never checked for minimal puzzle when I entered them in the database so, surely, I have some.

4) I don't intend to publish a new list before I have finished the current cycle (except for exceptional ones if any). You should at that time have closed your last rating. I'll filter my new ones using your last list.

If you agree, I'll send you to morrow 2matching list (through PM),

One for tarek pearly 6000
One for the others with possible redundancy.

champagne

by **eleven** » Thu Aug 04, 2011 10:36 pm

Yes, fine.

[Added:]
These lists have no ER.
There is one puzzle in my list, published earlier by Coloin with ER 11.8 (the others have ER11.4-):

Code: Select all: .....678........327.....5....1.4.....9.2.....6....7.5.3....58....29......4..1....;col 08 05;1325 ER 11.8/11,8/3.4

by **eleven** » Sun Aug 07, 2011 12:14 pm

The (triple-filtered) dob-0 puzzles turned out to need longer to rate than the sets i rated before, so rating is not yet finished. However, if not a really high rated puzzle will pop up over night, i will not publish, what comes later.
Some sets had a very high average rating, e.g. a set of about 350 puzzles, which are from the non miminal dob-0 puzzles with avgmaxUA > 49.5, that i got from my 21's. They had an average ER of more than 10.8 (and one 11.8). Also 22's with avgmaxUA > 49.5 gave average ER's > 10.7, but no extremely high ratings.
In the new list there are 10 20 clues, 386 21's, 267 22's and 27 23's (including Second Flush).

I uploaded both a list with my 690 new ER 11+ puzzles elevens_new11.zip and my new complete list elevens_hardest_V2.zip, where the 103 non minimal puzzles and the 229 earlier published puzzles are removed, and the new 11+ added.

by **champagne** » Wed Aug 10, 2011 8:10 am

deleted

champagne

by **ronk** » Wed Aug 10, 2011 1:26 pm

champagne wrote:I think It contains, as it is, all the puzzles rating 11 and more I identified.
I have to finish the rating of a big lot to have the entire list of rating elected in the area 10.5 10.9 with the criteria I took.

I hope your "criteria" is such that final results can be duplicated with ratings programs currently in the public domain.

by **champagne** » Wed Aug 10, 2011 7:23 pm

ronk wrote:
champagne wrote:I think It contains, as it is, all the puzzles rating 11 and more I identified.
I have to finish the rating of a big lot to have the entire list of rating elected in the area 10.5 10.9 with the criteria I took.

I hope your "criteria" is such that final results can be duplicated with ratings programs currently in the public domain.

it would be very nice to have moderators reading the readme files before emitting "as usual" unfriendly comments.

I left the pattern game after having been attacked daily by a bad loser, I am wondering if I must not just leave that forum for ever.

So many hours work for such a bad comment, what a pity

champagne

by **ronk** » Wed Aug 10, 2011 9:19 pm

champagne wrote:
ronk wrote:
champagne wrote:I think It contains, as it is, all the puzzles rating 11 and more I identified.
I have to finish the rating of a big lot to have the entire list of rating elected in the area 10.5 10.9 with the criteria I took.

I hope your "criteria" is such that final results can be duplicated with ratings programs currently in the public domain.

it would be very nice to have moderators reading the readme files before emitting "as usual" unfriendly comments.

Firstly, where are these 'readme files' I failed to read?

Secondly, why do you object to using rating programs in the public domain? If 'hardest lists' are based on a proprietary program, then the lists become archaic should the proprietor be hit by a truck ... or simply leave the hobby. We once had a hardest-list established by ravel which searched for the minimum number of steps required to reach a solution. It was known as the RMS (ravel's minimum step) rating. When ravel left, the list effectively became frozen in time.

Thirdly, proprietary ratings programs means the search activity for new hard puzzles is limited to a small clique, often a clique of one. Anyone should be able to participate in such a search and be confident that, if established criteria is met, the puzzle won't be rejected due to another's proprietary filter or ratings program. A searcher could obviously still use a pre-filter of his/her choosing, but the criteria of this pre-filter would not necessarily be one of the criteria for inclusion in the list.

by **champagne** » Thu Aug 11, 2011 7:36 am

I voluntarily discarded the file and the link to my collection of "potential hardest" due to the unfriendly context.

That collection contains one occurrence of all puzzles rating 11 and more (SE rating) and some puzzles (diamonds and pearls) in the area 10.5 10.9.

If anybody is interested in getting that file I'll give him a link through pm.

The file contains a readme.txt explaining in more details the way the file has been built and how it could be maintained.

I also discarded my first 2 files of generation in the 22 area which are obsolete

champagne

by **ttt** » Thu Aug 11, 2011 4:10 pm

champagne wrote:That collection contains one occurrence of all puzzles rating 11 and more (SE rating) and some puzzles (diamonds and pearls) in the area 10.5 10.9.
If anybody is interested in getting that file I'll give him a link through pm.
The file contains a readme.txt explaining in more details the way the file has been built and how it could be maintained.

I have not much time for Sudoku now..., but I'm interesting on that

. Send to me, please.

ttt

by **champagne** » Fri Aug 12, 2011 9:59 am

ttt wrote:
champagne wrote:That collection contains one occurrence of all puzzles rating 11 and more (SE rating) and some puzzles (diamonds and pearls) in the area 10.5 10.9.
If anybody is interested in getting that file I'll give him a link through pm.
The file contains a readme.txt explaining in more details the way the file has been built and how it could be maintained.

I have not much time for Sudoku now..., but I'm interesting on that . Send to me, please.

ttt

I sent you the link and more comments through pm

champagne

by **eleven** » Tue Aug 16, 2011 8:20 am

Champagne,

i cant see any reason, why you withdrawed the public link to your list. I also did not find it on your site.

Though i agree with ronk, that public rating programs are preferable, i am happy about each serious approach, which gives ratings to hard(est) collections.
We know, that in the current situation there is definitely only one rating, which is commonly accepted to reflect the difficulty of extremely hard puzzles for human solvers.

We had a nice example for one of champagne's methods in the formerly extremely overrated puzzle "Coloin". So i am interested, where else we can find big differences to the ER list.

by **champagne** » Tue Aug 16, 2011 10:48 am

eleven wrote:Champagne,

i cant see any reason, why you withdrawed the public link to your list. I also did not find it on your site.

Just take it as a sign that one more unfriendly post always from the same source put me in a very bad mood.

I'll put it back one day but for the time being, I'll deliver the link exclusively trough PM.

eleven wrote:Champagne,

We know, that in the current situation there is definitely only one rating,
which is commonly accepted to reflect the difficulty of extremely hard puzzles for human solvers.

That's why you and me, for months are exclusively working with SE rating.
The collection of puzzles (about 9000 as of to day) I made has been rated using the original Sudoku Explainer serate rating,
except may be for some of yours, that could have been rated using a fix from lksudoku.

eleven wrote:Champagne,

i am happy about each serious approach, which gives ratings to hard(est) collections.
So i am interested, where else we can find big differences to the ER list.

I have in my database identification of several properties leading to much shorter solutions.

I'll later rework all the file to mark the following properties

symmetry of given
SK loop
EXOCET
multi-fish pattern (rank 0 or near rank 0 logic)
other multi-floors having a high potential for elimination.

All these properties downgrade dramatically the puzzle in a player's view

champagne

by **champagne** » Wed Aug 17, 2011 10:16 am

I updated to-day my database of "potential hardest".

It contains now 9788 puzzles including 7545 puzzles rated more than 10.9 by SE

All puzzles previously published I identified are now re rated.

The last entry comes from my own search in the 22 field, with nothing really exciting.

The 2 best ratings I got were

98.7.......7.6.........57..4...3..2..1...6..3...9..8..2......4..3......1..86..5.0;11.80;1.20;1.20
98.7..6....5.4.......9...8.7.....8.336......9..2....1..3.6....8....2.........1.40;11.60;11.60;10.00

champagne

by **ronk** » Wed Aug 17, 2011 12:45 pm

champagne wrote:The 2 best ratings I got were

98.7.......7.6.........57..4...3..2..1...6..3...9..8..2......4..3......1..86..5.0;11.80;1.20;1.20
98.7..6....5.4.......9...8.7.....8.336......9..2....1..3.6....8....2.........1.40;11.60;11.60;10.00

One 0-rank move involving 4-layers, followed by one naked pair and singles solves the second puzzle.

Hidden Text: Show

[edit: add hidden XSUDO image and pastable logic set]

The New Sudoku Players' Forum

The hardest sudokus (new thread)

Re: The hardest sudokus (new thread)

Re: The hardest sudokus (new thread)

Re: The hardest sudokus (new thread)

Re: The hardest sudokus (new thread)

Re: The hardest sudokus (new thread)

Re: The hardest sudokus (new thread)

Re: The hardest sudokus (new thread)

Re: The hardest sudokus (new thread)

Re: The hardest sudokus (new thread)

Re: The hardest sudokus (new thread)

Re: The hardest sudokus (new thread)

Re: The hardest sudokus (new thread)

Re: The hardest sudokus (new thread)

Re: The hardest sudokus (new thread)

Re: The hardest sudokus (new thread)