- Code: Select all
1..9....3....8.6....47...1.5.7.3...6.2.6..9.......8....4..9.1....8....5.9..4....2 6.0/1.2/1.2
- late addition for previous game
Patterns Game
312 — late addition — 6.0
by Pat » Mon May 21, 2018 2:22 pm
::: 312 :::
-
Pat - Posts: 4056
- Joined: 18 July 2005
Re: Patterns Game
by champagne » Mon May 21, 2018 5:31 pm
papy999 wrote:::: comment :::
champagne 337 puzzles sent by pm (ER>=10.8 OR ER>=10.5 and pearl OR ER>=10.3 and diamond)
Patrice said there are 198 in your list
so there are at least 139 new ones and perhaps more
as I am not expecting ::: mores ::: in high ratings, I closed the game and processed the files
after posted puzzles, I find 164 new puzzles in your lot.
Having added several lots (including old game submissions), I intend to release a new situation of the data base of potential hardest shortly
- champagne
- 2017 Supporter
- Posts: 7356
- Joined: 02 August 2007
- Location: France Brittany
Re: Patterns Game
by Patrice » Mon May 21, 2018 5:37 pm
champagne wrote:And I was not surprised to see that you have the skills for automatic submissions (this was the case for gsf and likely mike and patrice have such a tool). I have no skill at all in this field, one reason to work more in "search mode" for the games
I had a bot when I was under Windows Vista, that could get pages from the pattern game thread and show for each ER my best ones against played ones (JAVA program)
It could also submit a puzzles by just a click on my list of best generated puzzles
But now with the change to Windows 10 and to Chrome, I lost this ability. Too much works to adapt... So when I play, it's full manual !
I didn't participate to game 313, because I made an analysis of PH1512, found 198 puzzles to play (many of them with skfr rating) and thought it was not fair to use them for the game... Next time I will not do this investigation before the end of a game !
BTW : papy999 : which technology are you using ?
- Patrice
- Posts: 1944
- Joined: 07 November 2010
- Location: Paris France
Re: Patterns Game
by dobrichev » Mon May 21, 2018 7:20 pm
Patrice wrote:I had a bot when I was under Windows Vista, that could get pages from the pattern game thread and show for each ER my best ones against played ones (JAVA program)
Oh, this famous portable Java.
- dobrichev
- 2016 Supporter
- Posts: 1850
- Joined: 24 May 2010
Re: Patterns Game
by Patrice » Tue May 22, 2018 8:04 pm
::: comment :::
dobrichev : yes portable Java
but the bot used : (program in Java) <=> Watij <=> IE
and the bot was a part of a multithread application that included (all parts in Java)
- generator {+/-n}
- brute solver
- serate rating engine
- watchman (for the opening of the game and to follow the game)
- dashboard (comparison for each ER of my bests and played ones + number of generated puzzles)
- player (to post a puzzle automatically - by a single click in a list)
I intended to write a "strategy engine" to get an autonomous player that can do the choice of puzzle to post... but I did not finalized it
I used it until nov 2012
watij is usualy used to test web application and enables access to the differents HTML elements of a web page, you can get the content of a tag, an element ... from your Java application
and you can set an input element or simulate the click on a button or a link... watij can work with IE, Chrome or Firefox
So with the change to windows 10 the problem comes from IE and watij, not from Java
new version of watij (non strictly compatible)
Chrome, a preferable choice for the browser
-------------------
When I came back to the game in sept 2017, I develop a new set of tools : everything is written in C or C++ ; I have only :
- generator {+/-n}
- brute solver
- minlex
- equivalent serate rating engine (a C version that mimics exactly serate - limited to non nested rules, that is 9.5) about 20 to 100 time faster
- memory database for generated puzzles (30 M puzzles needs only 1GB of memory)
and I use skfr to filter puzzles that have a nested level and serate to rate a small selection of those.
I never verify the rating with serate for non nested rating (<=9.5) ! My C version includes the serate bugs !
dobrichev : yes portable Java
but the bot used : (program in Java) <=> Watij <=> IE
and the bot was a part of a multithread application that included (all parts in Java)
- generator {+/-n}
- brute solver
- serate rating engine
- watchman (for the opening of the game and to follow the game)
- dashboard (comparison for each ER of my bests and played ones + number of generated puzzles)
- player (to post a puzzle automatically - by a single click in a list)
I intended to write a "strategy engine" to get an autonomous player that can do the choice of puzzle to post... but I did not finalized it
I used it until nov 2012
watij is usualy used to test web application and enables access to the differents HTML elements of a web page, you can get the content of a tag, an element ... from your Java application
and you can set an input element or simulate the click on a button or a link... watij can work with IE, Chrome or Firefox
So with the change to windows 10 the problem comes from IE and watij, not from Java
new version of watij (non strictly compatible)
Chrome, a preferable choice for the browser
-------------------
When I came back to the game in sept 2017, I develop a new set of tools : everything is written in C or C++ ; I have only :
- generator {+/-n}
- brute solver
- minlex
- equivalent serate rating engine (a C version that mimics exactly serate - limited to non nested rules, that is 9.5) about 20 to 100 time faster
- memory database for generated puzzles (30 M puzzles needs only 1GB of memory)
and I use skfr to filter puzzles that have a nested level and serate to rate a small selection of those.
I never verify the rating with serate for non nested rating (<=9.5) ! My C version includes the serate bugs !
- Patrice
- Posts: 1944
- Joined: 07 November 2010
- Location: Paris France
by g.r.emlin » Wed May 23, 2018 4:37 am
::: game 0313 idle 2018-05-23+04:37:34-0000 -- with no further entries the game will close at 2018-05-23+07:00:00-0000 :::
- Code: Select all
game 0313 2018-05-18+23:00:00-0000 duration 1d13h dealer coloin pages 2463-2472 (open) 6 players
_____________________________________________________________________________________________________________
1.2/1.2/1.2 m_b_metcalf 5.0/1.2/1.2 Patrice 7.6/7.6/7.3 m_b_metcalf 9.5/9.5/9.5 papy999
1.5/1.2/1.2 coloin 5.2/5.2/3.4 dobrichev 7.7/7.6/3.4 papy999 9.6/9.6/9.6 champagne
1.7/1.2/1.2 m_b_metcalf 5.6/3.4/3.4 m_b_metcalf 7.8/7.3/7.3 papy999 9.7/9.7/9.6 dobrichev
2.0/2.0/2.0 papy999 5.7/1.2/1.2 papy999 7.9/7.6/3.4 dobrichev 9.8/9.8/9.8 dobrichev
2.3/2.0/2.0 papy999 5.8/1.2/1.2 papy999 8.0/7.3/3.4 dobrichev 9.9/9.9/9.9 dobrichev
2.5/1.2/1.2 papy999 5.9/1.2/1.2 dobrichev 8.1/1.2/1.2 papy999 10.0/10.0/10.0 dobrichev
2.6/2.0/2.0 m_b_metcalf 6.0/1.2/1.2 papy999 8.2/7.6/7.3 papy999 10.1/10.1/10.1 dobrichev
2.8/2.0/2.0 m_b_metcalf 6.2/1.2/1.2 dobrichev 8.3/8.3/8.3 dobrichev 10.2/10.2/10.2 dobrichev
3.0/2.0/2.0 m_b_metcalf 6.5/2.0/2.0 dobrichev 8.4/8.4/8.4 papy999 10.3/10.3/10.1 dobrichev
3.2/2.0/2.0 m_b_metcalf 6.6/3.4/3.4 dobrichev 8.5/8.5/8.5 m_b_metcalf 10.4/10.4/10.4 dobrichev
3.4/3.4/3.4 dobrichev 6.7/6.7/6.6 dobrichev 8.6/8.6/8.5 papy999 10.5/10.5/10.5 champagne
3.6/2.0/2.0 papy999 6.8/3.4/3.4 papy999 8.7/8.7/3.4 papy999 10.6/10.6/10.6 dobrichev
3.8/2.0/2.0 papy999 6.9/2.0/2.0 dobrichev 8.8/8.8/3.4 dobrichev 10.7/10.7/10.6 papy999
4.0/1.2/1.2 papy999 7.0/2.0/2.0 dobrichev 8.9/8.9/8.5 m_b_metcalf 10.8/10.8/10.5 dobrichev
4.2/3.4/3.4 m_b_metcalf 7.1/7.1/7.1 papy999 9.0/9.0/9.0 dobrichev 10.9/10.9/10.9 dobrichev
4.4/1.2/1.2 m_b_metcalf 7.2/7.2/7.2 papy999 9.1/9.1/9.1 m_b_metcalf 11.1/11.1/10.9 papy999
4.5/3.4/3.4 m_b_metcalf 7.3/7.3/7.3 dobrichev 9.2/9.2/9.2 m_b_metcalf 11.2/11.2/10.4 papy999
4.6/2.0/2.0 papy999 7.4/7.4/7.4 papy999 9.3/9.3/9.3 champagne 11.3/11.3/11.1 champagne
4.7/2.0/2.0 dobrichev 7.5/7.5/7.3 dobrichev 9.4/9.4/9.4 papy999 11.4/11.4/3.4 papy999
_____________________________________________________________________________________________________________
open ratings : 4.8 4.9 5.1 5.4 6.1 11.0 11.5-11.8
submitter entries score response notes
----------- ------- ----- -------- -----
coloin 1 -19 0 dealer
Patrice 1 75 5h05m
champagne 4 577 10m29s rare-2,diamond*3,pearl
m_b_metcalf 15 1018 12m11s diamond*3,pearl*2,premature-entry
dobrichev 28 2923 42m26s rare-0,rare-1,diamond*11,pearl*7,premature-entry
papy999 27 3313 22m01s rare-0*3,rare-2,diamond*6,pearl*6,highest-rating=132,last-entry=369,dup-puzzle
----------- ------- ----- -----
6 76 7887 rare-0*4,rare-1,rare-2*2,diamond*23,pearl*16
-
g.r.emlin - Moderator
- Posts: 11580
- Joined: 07 December 2007
- Location: www
Re: Patterns Game
by dobrichev » Wed May 23, 2018 6:11 am
Yes Patrice, that it impressive.
In the above list only the determination of the morph of the puzzle that gives the target rating due to a known bug/feature in serate is missing.
Obviously this powerful toolset gives you advantage, and that was clearly seen by puzzles you played when you participate in the game.
Surely most of your efforts are worth, but ... reproducing the bugs ... hmmm ... gives strange satisfaction ... hmmmmm ... close to perversion.
OK, you can adapt to the game reality and be able to always win. And then what? Wait for some else to reproduce the rating bugs this becoming a real competitor, or do your own efforts in removal of the bugs and propaganda for replacing the official rating?
In the above list only the determination of the morph of the puzzle that gives the target rating due to a known bug/feature in serate is missing.
Obviously this powerful toolset gives you advantage, and that was clearly seen by puzzles you played when you participate in the game.
Surely most of your efforts are worth, but ... reproducing the bugs ... hmmm ... gives strange satisfaction ... hmmmmm ... close to perversion.
OK, you can adapt to the game reality and be able to always win. And then what? Wait for some else to reproduce the rating bugs this becoming a real competitor, or do your own efforts in removal of the bugs and propaganda for replacing the official rating?
- dobrichev
- 2016 Supporter
- Posts: 1850
- Joined: 24 May 2010
Re: Patterns Game
by champagne » Wed May 23, 2018 6:37 am
dobrichev wrote:In the above list only the determination of the morph of the puzzle that gives the target rating due to a known bug/feature in serate is missing.
Having partially studied the problem, I would say that we have no clear definition of a bug.
In the case of serate, I called "bug" some oddities that should not be as
- the shortest "length" for a chain does not always come in Nishio
- A hidden triplet in a UR has not the same rating depending on the number of cells not solved in the region
Also, some constraints obviously not compulsory with again some examples in URs
The closest of the serate code you are, the less deviations you have and, as far as I know, Patrice mostly translated the java code in a C++ code with the minimum of changes. Doing so, he reproduced options made in serate.
- champagne
- 2017 Supporter
- Posts: 7356
- Joined: 02 August 2007
- Location: France Brittany
by g.r.emlin » Wed May 23, 2018 7:02 am
::: game 0313 closed (idle for 2d18h) -- preliminary results :::
- Code: Select all
game 0313 2018-05-18+23:00:00-0000 duration 1d13h dealer coloin pages 2463-2472 (final) 6 players
_____________________________________________________________________________________________________________
1.2/1.2/1.2 m_b_metcalf 5.0/1.2/1.2 Patrice 7.6/7.6/7.3 m_b_metcalf 9.5/9.5/9.5 papy999
1.5/1.2/1.2 coloin 5.2/5.2/3.4 dobrichev 7.7/7.6/3.4 papy999 9.6/9.6/9.6 champagne
1.7/1.2/1.2 m_b_metcalf 5.6/3.4/3.4 m_b_metcalf 7.8/7.3/7.3 papy999 9.7/9.7/9.6 dobrichev
2.0/2.0/2.0 papy999 5.7/1.2/1.2 papy999 7.9/7.6/3.4 dobrichev 9.8/9.8/9.8 dobrichev
2.3/2.0/2.0 papy999 5.8/1.2/1.2 papy999 8.0/7.3/3.4 dobrichev 9.9/9.9/9.9 dobrichev
2.5/1.2/1.2 papy999 5.9/1.2/1.2 dobrichev 8.1/1.2/1.2 papy999 10.0/10.0/10.0 dobrichev
2.6/2.0/2.0 m_b_metcalf 6.0/1.2/1.2 papy999 8.2/7.6/7.3 papy999 10.1/10.1/10.1 dobrichev
2.8/2.0/2.0 m_b_metcalf 6.2/1.2/1.2 dobrichev 8.3/8.3/8.3 dobrichev 10.2/10.2/10.2 dobrichev
3.0/2.0/2.0 m_b_metcalf 6.5/2.0/2.0 dobrichev 8.4/8.4/8.4 papy999 10.3/10.3/10.1 dobrichev
3.2/2.0/2.0 m_b_metcalf 6.6/3.4/3.4 dobrichev 8.5/8.5/8.5 m_b_metcalf 10.4/10.4/10.4 dobrichev
3.4/3.4/3.4 dobrichev 6.7/6.7/6.6 dobrichev 8.6/8.6/8.5 papy999 10.5/10.5/10.5 champagne
3.6/2.0/2.0 papy999 6.8/3.4/3.4 papy999 8.7/8.7/3.4 papy999 10.6/10.6/10.6 dobrichev
3.8/2.0/2.0 papy999 6.9/2.0/2.0 dobrichev 8.8/8.8/3.4 dobrichev 10.7/10.7/10.6 papy999
4.0/1.2/1.2 papy999 7.0/2.0/2.0 dobrichev 8.9/8.9/8.5 m_b_metcalf 10.8/10.8/10.5 dobrichev
4.2/3.4/3.4 m_b_metcalf 7.1/7.1/7.1 papy999 9.0/9.0/9.0 dobrichev 10.9/10.9/10.9 dobrichev
4.4/1.2/1.2 m_b_metcalf 7.2/7.2/7.2 papy999 9.1/9.1/9.1 m_b_metcalf 11.1/11.1/10.9 papy999
4.5/3.4/3.4 m_b_metcalf 7.3/7.3/7.3 dobrichev 9.2/9.2/9.2 m_b_metcalf 11.2/11.2/10.4 papy999
4.6/2.0/2.0 papy999 7.4/7.4/7.4 papy999 9.3/9.3/9.3 champagne 11.3/11.3/11.1 champagne
4.7/2.0/2.0 dobrichev 7.5/7.5/7.3 dobrichev 9.4/9.4/9.4 papy999 11.4/11.4/3.4 papy999
_____________________________________________________________________________________________________________
open ratings : 4.8 4.9 5.1 5.4 6.1 11.0 11.5-11.8
submitter entries score response notes
----------- ------- ----- -------- -----
coloin 1 -19 0 dealer
Patrice 1 75 5h05m
champagne 4 577 10m29s rare-2,diamond*3,pearl
m_b_metcalf 15 1018 12m11s diamond*3,pearl*2,premature-entry
dobrichev 28 2923 42m26s rare-0,rare-1,diamond*11,pearl*7,premature-entry
papy999 27 3313 22m01s rare-0*3,rare-2,diamond*6,pearl*6,highest-rating=132,last-entry=369,dup-puzzle
----------- ------- ----- -----
6 76 7887 rare-0*4,rare-1,rare-2*2,diamond*23,pearl*16
-
g.r.emlin - Moderator
- Posts: 11580
- Joined: 07 December 2007
- Location: www
Re: Patterns Game
by dobrichev » Wed May 23, 2018 7:32 am
in continuation of the off-topic
A bug is a discrepancy between the specification and the result.
We have no specifications.
To me this is a bug unless from manual solver's perspective shortest length chain isn't easier identifiable.
A normal evolution of the rating should be primarily based on experts' consensus, and when this doesn't work for any reason, on authority of the people involved.
I don't know the history of creation of Sudoku Explainer, but seems that it is primarily based on the second option, and its popularity is due to Nicolas Juillerat's expertise in this field.
In the last decade I see no progress.
champagne wrote:Having partially studied the problem, I would say that we have no clear definition of a bug.
A bug is a discrepancy between the specification and the result.
We have no specifications.
champagne wrote:the shortest "length" for a chain does not always come in Nishio
To me this is a bug unless from manual solver's perspective shortest length chain isn't easier identifiable.
A normal evolution of the rating should be primarily based on experts' consensus, and when this doesn't work for any reason, on authority of the people involved.
I don't know the history of creation of Sudoku Explainer, but seems that it is primarily based on the second option, and its popularity is due to Nicolas Juillerat's expertise in this field.
In the last decade I see no progress.
- dobrichev
- 2016 Supporter
- Posts: 1850
- Joined: 24 May 2010
Re: Patterns Game
by dobrichev » Wed May 23, 2018 8:07 am
::: comment :::
Back to the game.
One with no trumping and no self-trumps.
The distribution by skfr ratings based on my cache of 1,991,451 puzzles
The pattern is short in low-ratings. Below are the counts grouped by skfr ER.
The peak is in ER=8.9 and it isn't an artefact of the searching bias.
All puzzles above 10.5 are pearls, and most in the range 10.0 - 10.5 are pearls too.
I stopped generation after less than one new puzzle appeared in 10 minutes on a 4-core machine.
It was fun.
Back to the game.
One with no trumping and no self-trumps.
The distribution by skfr ratings based on my cache of 1,991,451 puzzles
Hidden Text: Show
The pattern is short in low-ratings. Below are the counts grouped by skfr ER.
Hidden Text: Show
The peak is in ER=8.9 and it isn't an artefact of the searching bias.
All puzzles above 10.5 are pearls, and most in the range 10.0 - 10.5 are pearls too.
I stopped generation after less than one new puzzle appeared in 10 minutes on a 4-core machine.
It was fun.
- dobrichev
- 2016 Supporter
- Posts: 1850
- Joined: 24 May 2010
Re: Patterns Game
by champagne » Wed May 23, 2018 2:46 pm
Hi mladen,
thanks for the stats.
Just one comment, not only for the game, I am currently focusing on the rating range 1.0_6.2.
Even in this pattern with few low ER, this is about 18% of the puzzles generated.
On top of it, each higher rating ends with low ratings, so any improvement in this area will pay everywhere.
To follow (in case of success)
thanks for the stats.
Just one comment, not only for the game, I am currently focusing on the rating range 1.0_6.2.
Even in this pattern with few low ER, this is about 18% of the puzzles generated.
On top of it, each higher rating ends with low ratings, so any improvement in this area will pay everywhere.
To follow (in case of success)
- champagne
- 2017 Supporter
- Posts: 7356
- Joined: 02 August 2007
- Location: France Brittany
Re: Patterns Game
by Patrice » Fri May 25, 2018 6:35 pm
dobrichev wrote:do your own efforts in removal of the bugs and propaganda for replacing the official rating?
I think this is not worthwhile !
The new version of rating would not be accepted by the Sudoku Community, because the improvement is very low and serate has been used more than 10 years !
I think a new rating software should take in account other rules like :
- SK loop or D Loop
- Finned XWing.. and all other finned rule
- exocet
- and other rules...
...
and the rating of each rule should be choosen by people who uses their brain (and not a computer ) to solve sudoku
This would be a real challenge
- Patrice
- Posts: 1944
- Joined: 07 November 2010
- Location: Paris France
Re: Patterns Game
by dobrichev » Fri May 25, 2018 7:28 pm
Patrice wrote:The new version of rating would not be accepted by the Sudoku Community, because the improvement is very low and serate has been used more than 10 years !
If you meant your already done modification, then I can't comment because you know it.
With the rest I agree.
- dobrichev
- 2016 Supporter
- Posts: 1850
- Joined: 24 May 2010
by g.r.emlin » Fri May 25, 2018 11:02 pm
::: game 0314 start LIGHTNING at 2018-05-26+16:00:00-0000 :::
- Code: Select all
dealer status
candidate patterns (via private message to g.r.emlin)
2 coloin 3 Patrice 5 200e200w 7 Pat 87 g.r.emlin
past patterns
19 TTHsieh 1 pseudocoup 30 Mauricio 41 m_b_metcalf 9 gsf
1 claudiarabia 1 ravel 33 JPF 43 champagne
1 dobrichev 1 wintder 3 g.r.emlin 58 tarek
1 draco 2 200e200w 3 papy999 63 Pat
1 Eioru 24 coloin 3 Patrice 6 ronk
-
g.r.emlin - Moderator
- Posts: 11580
- Joined: 07 December 2007
- Location: www
