The New Sudoku Players' Forum

by **gsf** » Mon Apr 07, 2008 1:49 pm

Code: Select all: . . 1 . . 2 3 . . . 3 . . . . . 4 . 2 . . 6 . . . . 7 . . 8 4 . . . . 5 . . . . . . . . . 5 . . . . 1 8 . . 7 . . . . 8 . . 6 . 2 . . . . . 3 . . . 4 2 . . 1 . . ED=6.5/2.5/2.5

by **g.r.emlin** » Tue Apr 08, 2008 12:12 pm

::: game 0025 idle -- with no further entries the game will close at 2008-04-08+14:00:00-0000 :::

Code: Select all: game 0025 2008-04-05+21:20:00-0000 duration 1d16h dealer coloin pages 185-192 (open) 6 players ________________________________________________________________________________________________________ 1.2/1.2/1.2 Mauricio 4.5/4.2/3.4 Mauricio 7.3/7.3/7.3 Mauricio 9.0/9.0/9.0 gsf 1.5/1.2/1.2 coloin 4.6/3.4/3.4 Mauricio 7.4/7.4/7.4 gsf 9.1/9.1/9.1 gsf 1.7/1.2/1.2 Mauricio 4.7/2.0/2.0 Mauricio 7.5/7.5/7.5 Mauricio 9.2/9.2/9.2 gsf 2.0/2.0/2.0 gsf 5.0/2.0/2.0 Mauricio 7.6/7.6/7.1 gsf 9.3/9.3/9.3 gsf 2.3/2.0/2.0 Mauricio 5.2/2.0/2.0 Mauricio 7.7/7.7/3.4 gsf 9.4/9.4/9.4 gsf 2.5/2.5/2.5 Mauricio 5.6/4.2/3.4 Mauricio 7.8/7.8/7.8 gsf 9.5/9.5/9.4 gsf 2.6/2.5/2.5 Mauricio 5.7/3.4/3.0 Mauricio 7.9/7.9/3.0 gsf 9.6/9.6/9.3 JPF 2.8/2.5/2.5 gsf 6.2/1.2/1.2 JPF 8.0/8.0/3.0 m_b_metcalf 9.7/9.7/9.4 JPF 3.0/3.0/3.0 gsf 6.5/2.5/2.5 gsf 8.2/8.2/8.2 gsf 9.8/9.8/9.8 gsf 3.2/3.2/3.0 Mauricio 6.6/6.6/6.6 Mauricio 8.3/8.3/8.3 Mauricio 9.9/9.9/9.4 gsf 3.4/3.4/3.4 Mauricio 6.7/3.4/3.4 Mauricio 8.4/8.4/8.4 gsf 10.0/10.0/9.4 gsf 3.6/3.6/3.6 gsf 6.8/1.2/1.2 gsf 8.5/8.5/8.5 gsf 10.1/10.1/9.3 gsf 3.8/2.0/2.0 Mauricio 6.9/1.2/1.2 gsf 8.6/8.6/8.6 Mauricio 10.2/1.2/1.2 JPF 4.0/1.2/1.2 gsf 7.0/7.0/3.4 gsf 8.7/8.7/8.4 gsf 10.3/10.3/9.9 JPF 4.2/4.2/3.4 Mauricio 7.1/7.1/7.1 m_b_metcalf 8.8/8.8/8.8 Mauricio 10.4/10.4/9.0 gsf 4.4/4.4/4.4 gsf 7.2/7.2/7.2 m_b_metcalf 8.9/8.9/8.9 Mauricio 10.5/10.5/10.4 gsf ________________________________________________________________________________________________________ open ratings : 4.8 5.1 5.4 5.8-6.1 8.1 10.6 10.7 submitter entries score response notes ----------- ------- ----- -------- ----- coloin 1 -19 0 dealer tarek 0 0 59.00s m_b_metcalf 3 131 11m50s diamond*2,pearl,lower-pearl JPF 5 780 6h15m pearl*3 Mauricio 24 1847 7m01s diamond*9,pearl*8 gsf 31 3907 3m38s diamond*14,pearl*13,highest-rating,last-entry ----------- ------- ----- ----- 6 64 6646 diamond*25,pearl*25

by **Pat** » Tue Apr 08, 2008 12:18 pm

the above shows as 12:12 GMT
as far as i can tell, the time now is not yet 12 GMT

by **g.r.emlin** » Tue Apr 08, 2008 2:22 pm

::: game 0025 closed (idle for 1d00h) -- preliminary results :::

Code: Select all: game 0025 2008-04-05+21:20:00-0000 duration 1d16h dealer coloin pages 185-192 (idle) 6 players ________________________________________________________________________________________________________ 1.2/1.2/1.2 Mauricio 4.5/4.2/3.4 Mauricio 7.3/7.3/7.3 Mauricio 9.0/9.0/9.0 gsf 1.5/1.2/1.2 coloin 4.6/3.4/3.4 Mauricio 7.4/7.4/7.4 gsf 9.1/9.1/9.1 gsf 1.7/1.2/1.2 Mauricio 4.7/2.0/2.0 Mauricio 7.5/7.5/7.5 Mauricio 9.2/9.2/9.2 gsf 2.0/2.0/2.0 gsf 5.0/2.0/2.0 Mauricio 7.6/7.6/7.1 gsf 9.3/9.3/9.3 gsf 2.3/2.0/2.0 Mauricio 5.2/2.0/2.0 Mauricio 7.7/7.7/3.4 gsf 9.4/9.4/9.4 gsf 2.5/2.5/2.5 Mauricio 5.6/4.2/3.4 Mauricio 7.8/7.8/7.8 gsf 9.5/9.5/9.4 gsf 2.6/2.5/2.5 Mauricio 5.7/3.4/3.0 Mauricio 7.9/7.9/3.0 gsf 9.6/9.6/9.3 JPF 2.8/2.5/2.5 gsf 6.2/1.2/1.2 JPF 8.0/8.0/3.0 m_b_metcalf 9.7/9.7/9.4 JPF 3.0/3.0/3.0 gsf 6.5/2.5/2.5 gsf 8.2/8.2/8.2 gsf 9.8/9.8/9.8 gsf 3.2/3.2/3.0 Mauricio 6.6/6.6/6.6 Mauricio 8.3/8.3/8.3 Mauricio 9.9/9.9/9.4 gsf 3.4/3.4/3.4 Mauricio 6.7/3.4/3.4 Mauricio 8.4/8.4/8.4 gsf 10.0/10.0/9.4 gsf 3.6/3.6/3.6 gsf 6.8/1.2/1.2 gsf 8.5/8.5/8.5 gsf 10.1/10.1/9.3 gsf 3.8/2.0/2.0 Mauricio 6.9/1.2/1.2 gsf 8.6/8.6/8.6 Mauricio 10.2/1.2/1.2 JPF 4.0/1.2/1.2 gsf 7.0/7.0/3.4 gsf 8.7/8.7/8.4 gsf 10.3/10.3/9.9 JPF 4.2/4.2/3.4 Mauricio 7.1/7.1/7.1 m_b_metcalf 8.8/8.8/8.8 Mauricio 10.4/10.4/9.0 gsf 4.4/4.4/4.4 gsf 7.2/7.2/7.2 m_b_metcalf 8.9/8.9/8.9 Mauricio 10.5/10.5/10.4 gsf ________________________________________________________________________________________________________ open ratings : 4.8 5.1 5.4 5.8-6.1 8.1 10.6 10.7 submitter entries score response notes ----------- ------- ----- -------- ----- coloin 1 -19 0 dealer tarek 0 0 59.00s m_b_metcalf 3 131 11m50s diamond*2,pearl,lower-pearl JPF 5 780 6h15m pearl*3 Mauricio 24 1847 7m01s diamond*9,pearl*8 gsf 31 3907 3m38s diamond*14,pearl*13,highest-rating,last-entry ----------- ------- ----- ----- 6 64 6646 diamond*25,pearl*25

by **gsf** » Tue Apr 08, 2008 2:23 pm

::: more :::

Code: Select all: 001002300080000040500600007003800009000000000400009200700006005060000080004300100 6.7/4.2/3.4 001002300090000040200600007003400005000000000500007800700008006080000020004900100 6.8/6.6/4.2 001002300020000040500800007003700005000000000800001900700008009040000030002500400 6.8/2.0/2.0 001002300020000010400500006007600005000000000500004800600008004090000030002700100 5.9/2.0/2.0 008005600020000010900400008007300004000000000300006500400009002010000070002600800 9.5/9.5/9.5 001002300020000040500600007008700009000000000100008600900007005040000010003200400 6.8/2.5/2.5 008005600020000080100400009005300004000000000900007500400003001060000020002700800 4.0/4.0/3.0 001002300090000040500600009006100008000000000700004200400006005080000090003800100 6.2/4.2/3.0 001002300090000010800600007003100008000000000600007500700004006020000090005300100 6.9/6.9/3.4 008005600020000010100400003007300004000000000900007500400003001050000020002700800 6.8/6.8/3.4 001002300090000040200600005008400007000000000500001800700008003030000020004900100 6.8/6.8/3.0 001002300090000040200600007003100008000000000600007500700004006020000090004300100 6.7/6.7/3.4 001002300060000040500800007004700009000000000800006200700004005010000030002900100 7.9/7.9/3.4 001002300090000010500600002007300006000000000800004700600005008020000070003400100 5.7/4.2/3.4 001002300020000040500600009003100008000000000800007500900005006040000010002300400 5.2/2.5/2.5 001002300060000050900700006007400008000000000100003200400009003050000070006800400 9.6/9.6/9.4 001002300020000040800600007003100005000000000600008900700009008040000030002500400 6.9/2.0/2.0 001002300030000040500100006006300002000000000200009800600008005040000060002700900 5.8/1.2/1.2 001002300040000050500600007007400006000000000800009200600001004050000030002700800 7.7/7.7/7.2 008005900020000010400300008006700002000000000900006400300009006010000050007200300 10.1/10.1/9.6 001002300040000050300900007009700008000000000800006100600008004050000020004300700 10.4/10.4/9.3

by **m_b_metcalf** » Tue Apr 08, 2008 3:14 pm

gsf wrote::::Comment:::
tarek wrote:It is amazing to see such hard DIAMONDS coming out following 2 days of searching a pattern. I recall 1.5 years ago when I posted my first 9.9. it took much longer than that

adding pearls/diamonds has made it interesting
finally in this round I've become competitive
but only after adding 8x2.6Ghz to do the searching/se-rating

In order to give those of us with only one processor available an occasional chance to compete again on algorithms, might it be possible to replace the LIGHTNING games with ones where one is honour bound to use only one CPU?

Thanks,

Mike Metcalf

by **gsf** » Tue Apr 08, 2008 4:10 pm

m_b_metcalf wrote:
gsf wrote::::Comment:::
tarek wrote:It is amazing to see such hard DIAMONDS coming out following 2 days of searching a pattern. I recall 1.5 years ago when I posted my first 9.9. it took much longer than that

adding pearls/diamonds has made it interesting
finally in this round I've become competitive
but only after adding 8x2.6Ghz to do the searching/se-rating

In order to give those of us with only one processor available an occasional chance to compete again on algorithms, might it be possible to replace the LIGHTNING games with ones where one is honour bound to use only one CPU?

good point
I will limit my play for all games to 2 cpus
one just for patterns game searching, the other for g.r.emlin
the recent games I had been using the 8 cpus dedicated to searching

by **gsf** » Tue Apr 08, 2008 4:21 pm

Pat wrote:the above shows as 12:12 GMT
as far as i can tell, the time now is not yet 12 GMT
i.e. timestamps in the Forum are ( again ? always ? )
off by a quarter-hour or so

that's a forum admin issue
g.r.emlin uses the post times to determine premature entries for non-lightning entries
but otherwise uses an ntp regulated clock

the idle and closed times are rounded to the next ntp hour
so the idle interval is 22+roundup hours and the closing interval is 2+roundup hours

this means that players can look at their (hopefully within a few seconds of accurate)
computer (or wrist if you own one) clocks for the idle and closing times
the actual idle and closing posts should appear close to the hour

by **m_b_metcalf** » Tue Apr 08, 2008 4:28 pm

gsf wrote:good point
I will limit my play for all games to 2 cpus
one just for patterns game searching, the other for g.r.emlin
the recent games I had been using the 8 cpus dedicated to searching

That's very sporting of you. I've made continual improvements to my programs, even today, spurred on by the competition, but the best I can hope for with regard to GHz is to upgrade my laptop from 1.6 to 2.6, something I'm planning for this summer. Your exhaustive searches have, of course, been very impressive, even inspiring.

Thanks,

Mike Metcalf

by **gsf** » Tue Apr 08, 2008 4:51 pm

m_b_metcalf wrote:That's very sporting of you. I've made continual improvements to my programs, even today, spurred on by the competition, but the best I can hope for with regard to GHz is to upgrade my laptop from 1.6 to 2.6, something I'm planning for this summer. Your exhaustive searches have, of course, been very impressive, even inspiring.

I wish they were exhaustive
simply the old brute force pseudo-random*N

to compensate for one cpu I'm mulling minlex canonicalization only considering the pattern clue positions to guide/prune the search
but that's a few games off

by **gsf** » Tue Apr 08, 2008 5:00 pm

I think it might make sense to not include the more entries in the scoring
that way after a game closes players with extra horsepower could fill in the ratings blanks
without affecting fair play -- there is ~ 1 extra day before the final tally, the extra cpus could
be unleashed then

this would provide a more complete record for the pattern in the final tally
and also might help players tune their 1-cpu search strategies

by **Mauricio** » Tue Apr 08, 2008 6:37 pm

:::more:::

Code: Select all: 001002300030000040500600007002400008000000000800009500700008006010000030004100200 ED=5.2/2.5/2.5 001002300030000040500600007005800003000000000600007200700009004040000060008400100 ED=6.7/6.6/3.4

by **Mauricio** » Tue Apr 08, 2008 6:54 pm

m_b_metcalf wrote:That's very sporting of you. I've made continual improvements to my programs, even today, spurred on by the competition, but the best I can hope for with regard to GHz is to upgrade my laptop from 1.6 to 2.6, something I'm planning for this summer. Your exhaustive searches have, of course, been very impressive, even inspiring.

gsf wrote:I will limit my play for all games to 2 cpus
one just for patterns game searching, the other for g.r.emlin
the recent games I had been using the 8 cpus dedicated to searching

In my experience, the part that uses more cpu is the rating part. I prerated my puzzles with my own software and then elected the easiest ones for ER, that way I could rate ~100puzzles/min, all with ER<7.7, I could not compete against gsf if I rated hundreds of puzzles with rating>8.5.

I discovered a new searching method in the latest game I won, but only works advantageously when the percentage of heavy diamonds is low (gsf: or you did use only 2 cpu's that time?). So in a way, this patterns game has helped me to improve my programs too.

by **Sudo fanatic** » Tue Apr 08, 2008 7:51 pm

Where can I find screenshots for this sudoku game?

by **gsf** » Tue Apr 08, 2008 10:19 pm

Mauricio wrote:In my experience, the part that uses more cpu is the rating part. I prerated my puzzles with my own software and then elected the easiest ones for ER, that way I could rate ~100puzzles/min, all with ER<7.7, I could not compete against gsf if I rated hundreds of puzzles with rating>8.5.

I've been using

Code: Select all: -gtpr -e 'minimal==1&&(%#Hq)==(%#Pq)&&uniq()'

in my solver to bias towards pearls, changing Pq to Dq to bias towards diamonds later in the game
(those are only estimates -- only SE will determine the actual rating)

I discovered a new searching method in the latest game I won, but only works advantageously when the percentage of heavy diamonds is low (gsf: or you did use only 2 cpu's that time?). So in a way, this patterns game has helped me to improve my programs too.

8 cpus came in around game 0020

The New Sudoku Players' Forum

Patterns Game 1.5

timestamps in the Forum

Re: timestamps in the Forum