The New Sudoku Players' Forum

by **m_b_metcalf** » Thu Jan 24, 2008 5:05 am

Code: Select all: 7 . . . . . . . 1 . 6 . . . . . 2 . . . . 5 . 9 . . . . . 4 6 . 1 5 . . . . . . 4 . . . . . . 3 2 . 7 8 . . . . . 8 . 4 . . . . 1 . . . . . 6 . 2 . . . . . . . 7 SE 10.1

m_b_metcalf's solver says it's a strong Pearl; Simple Sudoku has only one hint and can solve no cell; Sudoku Explainer reaches SE 7.2 to solve the first cell. I live in hope.

Regards,

Mike Metcalf

by **g.r.emlin** » Thu Jan 24, 2008 5:07 am

::: adjustment :::

Code: Select all: 2008-01-24+04:05:00-0000 m_b_metcalf dup-puzzle -50

by **m_b_metcalf** » Thu Jan 24, 2008 5:15 am

g.r.emlin wrote:::: adjustment :::
Code: Select all
2008-01-24+04:05:00-0000 m_b_metcalf dup-puzzle -50

One versus three. Wins.

Regards,

Mike Metcalf

by **TTHsieh** » Thu Jan 24, 2008 6:04 am

Code: Select all: 1 . . | . . . | . . 2 . 3 . | . . . | . 1 . . . . | 2 . 4 | . . . - - - + - - - + - - - . . 5 | 4 . 6 | 7 . . . . . | . 8 . | . . . . . 9 | 3 . 2 | 5 . . - - - + - - - + - - - . . . | 7 . 9 | . . . . 6 . | . . . | . 3 . 2 . . | . . . | . . 6 EP=8.7/2

TTHsieh

by **gsf** » Thu Jan 24, 2008 6:09 am

m_b_metcalf wrote:m_b_metcalf's solver says it's a strong Pearl; Simple Sudoku has only one hint and can solve no cell; Sudoku Explainer reaches SE 7.2 to solve the first cell. I live in hope.

its isomorphic to the 11th entry

anyone know how to coax a per-move trace from the command line Sudoku Explainer?

by **m_b_metcalf** » Thu Jan 24, 2008 2:46 pm

gsf wrote:
TTHsieh wrote:
gsf wrote:...adding a pearl bonus could add a bit of strategy...

Just like Sudoku Explainer evaluating the SE ratings.
If there is a simple and unique way to evaluate the pearl, then we can add the pearl's rule.

gsf, I think you can do it.

with a bit of tweaking my solver can efficiently filter out simple sudoku constraint set pearls
...

Well, I accept that this is an isomorph but, once again, I was misled. In good faith, I too tuned my solver to SS! I think we should drop trumping until such time as a pearl rating can be obtained easily from SE. Let's wait for the author's reply. The pearl bonus can remain as before. I'll take pot luck on that.

Regards,

Mike Metcalf

by **gsf** » Thu Jan 24, 2008 4:32 pm

m_b_metcalf wrote:Well, I accept that this is an isomorph but, once again, I was misled. In good faith, I too tuned my solver to SS! I think we should drop trumping until such time as a pearl rating can be obtained easily from SE. Let's wait for the author's reply. The pearl bonus can remain as before. I'll take pot luck on that.

misled because a difference in pearl rating made you think it was a different puzzle?
if so, you're right, that shouldn't happen when computed by the same program
but
do I recall seeing two isomorphic puzzles with different ERs?
anyway, just to be sure the referee does the isomorphism checks first
besides, in searching it is much more efficient to weed out the isomorphs before checking with SE

by **gsf** » Thu Jan 24, 2008 7:07 pm

Code: Select all: 1 . . | . . . | . . 6 . 9 . | . . . | . 1 . . . . | 7 . 4 | . . . ------+-------+------ . . 5 | 9 . 8 | 4 . . . . . | . 4 . | . . . . . 3 | 2 . 7 | 5 . . ------+-------+------ . . . | 5 . 9 | . . . . 2 . | . . . | . 8 . 6 . . | . . . | . . 4 EP=2.29/2

by **TTHsieh** » Thu Jan 24, 2008 7:33 pm

Code: Select all: 1 . . | . . . | . . 2 . 3 . | . . . | . 4 . . . . | 5 . 6 | . . . - - - + - - - + - - - . . 7 | 1 . 3 | 6 . . . . . | . 8 . | . . . . . 5 | 2 . 9 | 8 . . - - - + - - - + - - - . . . | 8 . 5 | . . . . 2 . | . . . | . 1 . 3 . . | . . . | . . 4 EP=4.0/2

TTHsieh

by **gsf** » Thu Jan 24, 2008 8:46 pm

Code: Select all: 1 . . | . . . | . . 2 . 5 . | . . . | . 4 . . . . | 5 . 6 | . . . ------+-------+------ . . 6 | 7 . 1 | 8 . . . . . | . 6 . | . . . . . 7 | 3 . 8 | 5 . . ------+-------+------ . . . | 9 . 5 | . . . . 8 . | . . . | . 1 . 2 . . | . . . | . . 4 EP=5.2/2

by **gsf** » Thu Jan 24, 2008 9:02 pm

sometimes impatience is a good thing
I learned how to create jar files
and with that have posted a modified SE that also lists the pearl rating
I started editing the first post, but won't finish that until this game closes
here are the details from the rules
---
Pearl ratings are determined by a modified Sudoku Explainer 1.2.1. This version provides two new command line entry points similar to the Tester entry point described in Ravel's post. Both new entry points truncate ratings to the hundreths digit. The entry points are:

ER: This entry point takes one or more 81-character puzzle arguments and lists the explainer-rating for each puzzle on a separate line.
EP: This entry point takes one or more 81-character puzzle arguments and lists the explainer-rating/pearl-rating for each puzzle on a separate line. The pearl-rating is the explainer-rating of the most difficult method leading to the first cell placement.

---
the ER and especially the EP entry points need independent testing (meaning you the patterns game participant)
if that goes well then the next game (0014) will use EP ratings
and once again only Sudoku Explainer will be on the hook

note that both ER and EP truncate ratings to the hundreths digit

if EP ratings are acceptable then the weak/strong distinction will be dropped
in favor of a bonus based on the pearl rating

remember, no experiment is a failure

by **tarek** » Thu Jan 24, 2008 9:25 pm

That is great gsf,

I will try to test it & confirm the results....

I have been away from this game but wil return hopefully soon (very soon)...

The modification will be even greater if it would display:

Code: Select all: ER/EP/ED

where the Diamond rating is the ER of the 1st move

tarek

by **gsf** » Thu Jan 24, 2008 9:30 pm

tarek wrote:I will try to test it & confirm the results....

I have been away from this game but wil return hopefully soon (very soon)...

The modification will be even greater if it would display:
Code: Select all
ER/EP/ED
where the Diamond rating is the ER of the 1st move

thanks

EP is the highest ER leading to the first placement
ED would be the ER of the first method (placement or candidate elimination)?

can you point me to a puzzle where ER/EP/ED are all different

by **tarek** » Thu Jan 24, 2008 10:02 pm

gsf wrote:EP is the highest ER leading to the first placement
ED would be the ER of the first method (placement or candidate elimination)?

can you point me to a puzzle where ER/EP/ED are all different

Correct,

Code: Select all: .4.8.6.2.9..3.5.6.6..2.15..764952318.3.7642952..1836743.65.7...41.638.52...4.9.36

Source: modified from a puzzle posted by wychwood at http://forum.enjoysudoku.com/viewtopic.php?t=5860
ED 2.6 Pointing
EP 4.2 XY-Wing
ER 6.6 Bidirectional Y-Cycle
ER/EP/ED = 6.6/4.2/2.6

tarek

by **gsf** » Thu Jan 24, 2008 10:15 pm

tarek wrote:
gsf wrote:EP is the highest ER leading to the first placement
ED would be the ER of the first method (placement or candidate elimination)?

can you point me to a puzzle where ER/EP/ED are all different

Correct,

Code: Select all
.4.8.6.2.9..3.5.6.6..2.15..764952318.3.7642952..1836743.65.7...41.638.52...4.9.36

Source: modified from a puzzle posted by wychwood at http://forum.enjoysudoku.com/viewtopic.php?t=5860
ED 2.6 Pointing
EP 4.2 XY-Wing
ER 6.6 Bidirectional Y-Cycle
ER/EP/ED = 6.6/4.2/2.6

thanks
I added an ED entry point and it confirms 6.6/4.2/2.6
you'll have to download the jar file again

I'm generating ER/EP/ED for the current game entries and will post when done

it won't be a problem retooling the referee for ER/EP/ED
trumps would sort from the left: ER - EP - ED

new games should produce interesting taxonomies
especially if coloin keeps submitting deals
right now we stand at 60 entries, 4 with ER >= 10.0

The New Sudoku Players' Forum

Patterns Game 1.5