The SUPERIOR thread

Everything about Sudoku that doesn't fit in one of the other sections

Postby tarek » Wed May 10, 2006 12:54 am

Apparantly, Pappocom software rates puzzles that require only doubles to solve as hard & not Very hard......
Therefore although beatiful puzzles....the following are all non Superior puzzles:

Puzzle#002 (Ruud)
Puzzle#004 (ab/tarek)
Puzzle#006 (vidarino)
Puzzle#012 (gsf)
Puzzle#019 (vidarino)
Puzzle#022 (Ruud)
Puzzle#030 (vidarino)
Puzzle#036 (Ruud)
Puzzle#084 (JPF/tarek)
....

These will be replaced mostly by puzzles posted by the same compiler in the thread.........

tarek
User avatar
tarek
 
Posts: 2612
Joined: 05 January 2006

re: Pappocom ratings

Postby Pat » Thu May 11, 2006 7:49 am

tarek wrote:Apparantly, the Pappocom software rates puzzles that require only doubles to solve as Hard – not Very Hard

that's true for some puzzles
but it is certainly not a general rule,
we've seen Pappocom puzzles rated Very Hard which could be solved by a duo.


some examples:
The Sunday Times wrote:
  • { www.timesonline.co.uk/article/0,,18209-2075701,00.html -- broken link } #33 (2006.Mar.19) [·:·:· ]
  • { www.timesonline.co.uk/article/0,,18209-2132071,00.html -- broken link } #39 (2006.Apr.30)
User avatar
Pat
 
Posts: 3425
Joined: 18 July 2005

re: ratings

Postby Pat » Sun May 14, 2006 9:32 am

tarek (2006.Apr.25) wrote:This thread started nicely,
but ended causing a headache.

My rankings---


tarek, you are discovering that rating the puzzles is tougher than generating them:
gfroyle (2006.Jan.20) wrote:I think the confidential part of Pappocom's software is the RATING of puzzles.

Generating a valid puzzle is easy---

The tricky part is to then ensure that the resulting puzzle is of reasonable difficulty and reasonably interesting for solvers.


there is no simple answer on ratings.

~ Pat
User avatar
Pat
 
Posts: 3425
Joined: 18 July 2005

re: rating of Ocean's recent puzzle

Postby Pat » Fri May 19, 2006 9:42 am

Ocean wrote:20 clues
Code: Select all
 . . . | . . . | . . .
 . 1 . | . . . | . 2 .
 . . 3 | 4 . 5 | 6 . .
-------+-------+------
 . . 4 | . 5 . | 3 . .
 . . . | 2 . 7 | . . . 
 . . 6 | . 8 . | 4 . .
-------+-------+------
 . . 5 | 6 . 3 | 1 . .
 . 7 . | . . . | . 9 .
 . . . | . . . | . . .

tarek wrote:it scores 3


tarek, evidently i don't understand your method of rating the puzzles.

i'd think Ocean's recent puzzle should rate quite high:
1 X-wing + 2 trios.

but i'm not a computer -
is there perhaps a simpler solution-path??
User avatar
Pat
 
Posts: 3425
Joined: 18 July 2005

Postby tarek » Fri May 19, 2006 10:50 am

Hi Pat,

The score has nothing to do with difficulty, it has to do with consistancy & reproducability..........

for reproducability, scritinising the puzzle against the Quads (& possiblu UR) will insure reproducability.......

the puzzle posted by Ocean has many Quads (Hidden & naked), but only the x-wing remains...

Removing the Quads scrutiny.....

It needs hidden triple/naked quad & a Naked triple/Hidden triple & an x-wing totalling 11 on the consistancy score (Excellent)........

It brought to my attention that it had the same pattern as one of my replacement puzzles for #84 which also scored the same as Ocean's, however Ocean's puzzle overall is more difficult needing an extra box-line interaction.....

after replacing all puzzles, I'll post the rankings sorted according to my normal solver, consistancy scores would then be next to each....

I have been trying to generate some puzzles following the fully symmetric puzzles, & this pattern was the easiest to generate.....

I will therefore just change the configuration slightly of my puzzle to an equivelant & post it as a replacement, here it is
Code: Select all
 3 . . | 4 . 5 | . . 7 
 . . . | . . . | . . . 
 . . 2 | . . . | 6 . . 
-------+-------+------
 5 . . | . 1 . | . . 9 
 . . . | 6 . 8 | . . . 
 9 . . | . 7 . | . . 4 
-------+-------+------
 . . 3 | . . . | 2 . . 
 . . . | . . . | . . . 
 4 . . | 9 . 7 | . . 5 

if you tried solving Ocean's puzzle, then this would be almost identical.

& my replacement for puzzle 4 is just a joy to solve, I chose it although it wasn't as difficult as other possible replacements because it showed consistant variability it scored 7 without the scrutiny of the quads & scored 4 when they were present, No hidden triples & no x-wings, it NEEDS 4-5 different non single/non Box line interaction steps, & what a nice start:D
Code: Select all
 8 . . | . . . | . . 1 
 . . . | 9 3 4 | . . . 
 . . 6 | . . . | 3 . . 
-------+-------+------
 . 5 . | . 1 . | . 8 . 
 . 4 . | 6 . 2 | . 1 . 
 . 1 . | . 7 . | . 3 . 
-------+-------+------
 . . 9 | . . . | 2 . . 
 . . . | 2 4 8 | . . . 
 5 . . | . . . | . . 7


tarek
User avatar
tarek
 
Posts: 2612
Joined: 05 January 2006

Postby tarek » Fri May 26, 2006 1:00 pm

I've looked through the thread & found only theses replacements:

Code: Select all
Puzzle #006 (vidarino)
 . 1 5 | . . . | . 3 . 
 . . . | 6 1 5 | . . . 
 8 . . | . . 9 | . 7 . 
-------+-------+------
 . . . | . . . | 7 1 2 
 . . . | . 6 . | . . . 
 3 4 7 | . . . | . . . 
-------+-------+------
 . 7 . | 3 . . | . . 6 
 . . . | 5 2 7 | . . . 
 . 2 . | . . . | 3 4 .

will be replaced by:

 . 8 . | . . . | . . 3 
 . . . | 3 1 . | . . . 
 . 4 . | . . 5 | . 1 9 
-------+-------+------
 . 5 1 | 7 . . | . . 4 
 . . 2 | . . . | 6 . . 
 7 . . | . . 6 | 1 9 . 
-------+-------+------
 9 2 . | 8 . . | . 5 . 
 . . . | . 2 3 | . . . 
 5 . . | . . . | . 4 . 

Puzzle #019 (vidarino)
 . 8 . | 3 . 5 | . 2 . 
 3 . . | . . . | . . 5 
 . 5 . | . . . | . 1 . 
-------+-------+------
 . . 7 | . 1 . | 3 . . 
 2 . . | 5 . 7 | . . 8 
 . . 6 | . 4 . | 7 . . 
-------+-------+------
 . 3 . | . . . | . 6 . 
 1 . . | . . . | . . 4 
 . 6 . | 9 . 2 | . 3 . 

will be replaced by:

 1 9 . | 6 . 8 | . 4 3 
 6 8 . | . . . | . 9 7 
 . . . | . . . | . . . 
-------+-------+------
 2 . . | 7 . 4 | . . 6 
 . . . | . 3 . | . . . 
 3 . . | 5 . 1 | . . 2 
-------+-------+------
 . . . | . . . | . . . 
 4 3 . | . . . | . 5 1 
 8 5 . | 1 . 7 | . 6 4 

Puzzle #030 (vidarino)
 9 4 . | . 8 . | . 7 2 
 7 . . | 9 . 4 | . . 8 
 . . . | . 2 . | . . . 
-------+-------+------
 . 9 . | . . . | . 2 . 
 6 . 7 | . . . | 4 . 1 
 . 8 . | . . . | . 6 . 
-------+-------+------
 . . . | . 3 . | . . . 
 5 . . | 1 . 8 | . . 6 
 1 6 . | . 4 . | . 5 9 

will be replaced by:

 . . 7 | . 3 . | 5 . . 
 . 6 . | 9 . 7 | . 3 . 
 . 1 . | . 6 . | . 9 . 
-------+-------+------
 . 2 3 | . 8 . | 6 7 . 
 . . 9 | . . . | 2 . . 
 . 8 6 | . 2 . | 9 1 . 
-------+-------+------
 . 5 . | . 9 . | . 6 . 
 . 3 . | 5 . 4 | . 2 . 
 . . 2 | . 1 . | 4 . . 


Puzzle #002 (Ruud)
 . . 5 | 7 . . | . . . 
 . . . | . 9 3 | . . . 
 . . 7 | . . 1 | 6 . 4 
-------+-------+------
 2 . . | . . 8 | . 6 . 
 . . . | . 2 . | . . . 
 . 1 . | 4 . . | . . 5 
-------+-------+------
 5 . 4 | 8 . . | 3 . . 
 . . . | 6 3 . | . . . 
 . . . | . . 2 | 9 . .

will be replaced by:

 . 5 . | . . . | . . 7 
 3 . . | 2 . 6 | 5 . 9 
 . 4 . | . 5 . | 6 . . 
-------+-------+------
 . . 7 | . 6 8 | . . . 
 . . . | . . . | . . . 
 . . . | 4 2 . | 8 . . 
-------+-------+------
 . . 5 | . 9 . | . 2 . 
 6 . 2 | 8 . 3 | . . 4 
 9 . . | . . . | . 1 . 


I still need one to replace gsf's #012 & Ruud's #022 & #036.......Could the corresponding compilers provide replacements (any that can't be solved using doubles only but NEED 1 or more of the following to solve [Any form of triple, X-wing]), I'll wait for 7 days for a reply....

Also as some of the puzzles posted by compilers have the same clue distribution (some of Ocean's, myself & others), would it be possible for them to repost their puzzles with some row,column,band permutations so that no 2 puzzles look exactly the same (It saves me some time), I'll also wait for 7 days for that.

Thanx,

tarek
User avatar
tarek
 
Posts: 2612
Joined: 05 January 2006

Postby gsf » Fri May 26, 2006 3:36 pm

tarek wrote:I still need one to replace gsf's #012

(edit) just finished a scan of ~7M fresh SM puzzles
Code: Select all
# these puzzles solve with singles and one x-wing

# 25 clues diagonal symmetric minimal (VII)

. . .  7 6 .  . . .
. 4 .  . . 1  3 . .
. . 8  . 5 .  . 6 .

. . .  2 . .  . 8 .
. 3 2  . . .  7 . 1
. . 7  . . 6  . . 2

. 8 .  9 1 .  6 . .
. 9 3  . 2 .  . 5 .
. . .  . . .  . . .

# 28 clues pi rotational (180 degrees) symmetric minimal (V)

. 2 .  . 1 6  . . 3
3 9 .  . 5 .  . . .
. . 6  9 . .  . . .

9 8 .  . 3 5  . . 1
. . .  . . .  . . .
4 . .  7 9 .  . 5 6

. . .  . . 1  4 . .
. . .  . 8 .  . 2 5
5 . .  4 6 .  . 1 .

# 24 clues vertical symmetric minimal (VI)

. . .  3 . 8  . . .
. 2 .  . . .  . 9 .
. 5 1  . . .  3 6 .

. . 8  7 . 4  1 . .
3 . .  . . .  . . 9
. . 5  . . .  7 . .

. 8 .  . 6 .  . 3 .
2 . .  . . .  . . 1
. 9 .  . 8 .  . 5 .

# 28 clues diagonal and antidiagonal symmetric minimal (IV)
3 . .  . 6 8  . 7 .
. . .  . 3 .  4 . 9
. . .  5 . .  . 3 .

. . 7  6 . .  . . 4
4 2 .  . . .  . 6 5
5 . .  . . 7  1 . .

. 6 .  . . 9  . . .
1 . 2  . 5 .  . . .
. 4 .  8 7 .  . . 3

# 30 clues horizontal and vertical symmetric minimal (III)

1 4 .  3 . 5  . 8 2
. . 5  . . .  1 . .
2 . .  . 6 .  . . 7

3 . .  8 . 6  . . 1
. . .  . . .  . . .
5 . .  4 . 9  . . 6

7 . .  . 4 .  . . 3
. . 3  . . .  7 . .
8 6 .  2 . 7  . 9 5

# 28 clues full rotational (90 180 270 degrees) symmetric minimal (II)

. 7 .  2 . .  . 6 .
4 . .  . 3 .  . . 1
. . 9  4 . 6  8 . .

. . 7  . . .  1 . 6
. 3 .  . . .  . 5 .
5 . 1  . . .  4 . .

. . 8  7 . 4  2 . .
1 . .  . 2 .  . . 5
. 9 .  . . 3  . 4 .

# 28 clues full dihedral symmetric minimal (I)

. . 8  4 . 2  7 . .
. . .  . 7 .  . . .
7 . .  3 . 6  . . 8

8 . 9  . . .  1 . 5
. 5 .  . . .  . 6 .
2 . 6  . . .  8 . 3

3 . .  8 . 1  . . 6
. . .  . 4 .  . . .
. . 4  5 . 7  9 . .
gsf
2014 Supporter
 
Posts: 7306
Joined: 21 September 2005
Location: NJ USA

Postby Ocean » Sat May 27, 2006 11:29 pm

tarek wrote:Also as some of the puzzles posted by compilers have the same clue distribution (some of Ocean's, myself & others), would it be possible for them to repost their puzzles with some row,column,band permutations so that no 2 puzzles look exactly the same (It saves me some time), I'll also wait for 7 days for that.

Thanx,

tarek


I have changed one:
Code: Select all
Puzzle #098 Variant (Ocean)
 . 2 . | 6 . 7 | . 1 . 
 . . 6 | . 8 . | 2 . . 
 . . . | 4 . 5 | . . . 
-------+-------+------
 3 . 2 | . . . | 5 . 9 
 . 4 . | . . . | . 6 . 
 6 . 5 | . . . | 1 . 2 
-------+-------+------
 . . . | 7 . 1 | . . . 
 . . 7 | . 9 . | 6 . . 
 . 6 . | 5 . 2 | . 8 . 


Didn't spot any others with identical patterns, but will have a second look later.
Ocean
 
Posts: 442
Joined: 29 August 2005

Postby tarek » Sun Jun 04, 2006 1:49 pm

gsf wrote:these puzzles solve with singles and one x-wing

Ocean wrote:I have changed one:
Code: Select all
Puzzle #098 Variant (Ocean)

Didn't spot any others with identical patterns, but will have a second look later


thanx gsf & Ocean,

You puzzles gsf indeed need an x-wing to solve even when scrutinised with Quads.....

As you well know type VI & Type VII are not 180 degree symmetrical, the rest of symmetry tupes have 180 degree symmetry as default.

so Puzzle #012 gsf should now be:
Code: Select all
 . 2 . | . 1 6 | . . 3 
 3 9 . | . 5 . | . . . 
 . . 6 | 9 . . | . . . 
-------+-------+------
 9 8 . | . 3 5 | . . 1 
 . . . | . . . | . . . 
 4 . . | 7 9 . | . 5 6 
-------+-------+------
 . . . | . . 1 | 4 . . 
 . . . | . 8 . | . 2 5 
 5 . . | 4 6 . | . 1 . 


Puzzle #98 Ocean would now be:
Code: Select all
 . 2 . | 6 . 7 | . 1 . 
 . . 6 | . 8 . | 2 . . 
 . . . | 4 . 5 | . . . 
-------+-------+------
 3 . 2 | . . . | 5 . 9 
 . 4 . | . . . | . 6 . 
 6 . 5 | . . . | 1 . 2 
-------+-------+------
 . . . | 7 . 1 | . . . 
 . . 7 | . 9 . | 6 . . 
 . 6 . | 5 . 2 | . 8 .


As there was no response in 7 days regarding #22 & #36, one of gsf's puzzles replaces #22 as follows:
Code: Select all
 . 7 . | 2 . . | . 6 . 
 4 . . | . 3 . | . . 1 
 . . 9 | 4 . 6 | 8 . . 
-------+-------+------
 . . 7 | . . . | 1 . 6 
 . 3 . | . . . | . 5 . 
 5 . 1 | . . . | 4 . . 
-------+-------+------
 . . 8 | 7 . 4 | 2 . . 
 1 . . | . 2 . | . . 5 
 . 9 . | . . 3 | . 4 . 
& one of mine replace #36
Code: Select all
 . . 8 | . . . | 6 . . 
 . . . | 9 3 1 | . . . 
 5 . . | . . . | . . 3 
-------+-------+------
 . 2 . | . 1 . | . 9 . 
 . 8 . | 4 . 5 | . 3 . 
 . 6 . | . 7 . | . 1 . 
-------+-------+------
 9 . . | . . . | . . 4 
 . . . | 1 4 2 | . . . 
 . . 7 | . . . | 5 . . 


I will now update the list & then hopefully spot 2 puzzles that look the same & change one [Edit: I just performed a face-lift on my puzzles to insure no 2 look exactly the same, Also included was one of Ocean's, If any still remain, inform me & LASER surgery should be possible:D ......

I will update the (if I can) the list the ronk supplied at the end, & I'll post rankings for all as follows: General difficulty, Consistantcy points (up to level of Hidden triple), Consistantcy points (up to level of Quads), Possibly Consistancy score up to level of colouring (includ. UR & Fish & finned fish)

tarek

tarek
User avatar
tarek
 
Posts: 2612
Joined: 05 January 2006

very superior: Sunday Times #44 (2006.June.4)

Postby Pat » Wed Jun 07, 2006 7:42 am

— and to celebrate the closing of the "superior" list, { www.TimesOnline.co.UK/article/0,,18209-2202623,00.html — broken link } The Sunday Times #44 (2006.June.4) is very superior — seems way above their usual level — i suspect it would rate quite high in our "superior" Topic

Code: Select all

 . . . | 8 2 . | . 7 .
 3 . . | 5 . . | . . 1
 . 2 4 | . . 1 | . . .
-------+-------+------
 . 7 1 | . . . | 5 . .
 2 . . | . . . | . . 8
 . . 5 | . . . | 6 2 .
-------+-------+------
 . . . | 7 . . | 3 6 .
 1 . . | . . 9 | . . 7
 . 5 . | . 3 6 | . . .



·:· — an X-wing and 2 trios
— ·:· — or is there a simpler way ??
User avatar
Pat
 
Posts: 3425
Joined: 18 July 2005

Re: very superior: Sunday Times #44 (2006.June.4)

Postby tarek » Wed Jun 07, 2006 9:00 am

Pat wrote:— and to celebrate the closing of the "superior" list The Sunday Times #44 (2006.June.4) is very superior — seems way above their usual level — i suspect it would rate quite high in our "superior" Topic

Indeed a nice superior, the consistancy marks (4) show that it needs a naked triple (with a hidden triple counterpart) & a hidden double (with a naked triple counterpart) The rest is variable but an x-wing is definitely not needed to solve.

tarek
User avatar
tarek
 
Posts: 2612
Joined: 05 January 2006

re: Sunday Times #44 (2006.June.4)

Postby Pat » Wed Jun 07, 2006 11:33 am

tarek wrote:Indeed a nice superior; the consistancy marks (4) show that it needs a naked triple (with a hidden triple counterpart) + a hidden double (with a naked triple counterpart), the rest is variable but an X-wing is definitely not needed.


thanks, tarek!
i did use an X-wing,
but i had a feeling i was doing things the hard way -
will look again when not so tired---
User avatar
Pat
 
Posts: 3425
Joined: 18 July 2005

Postby doduff » Thu Jun 15, 2006 1:26 am

This is kind of off topic, but when I look at the inlin list of superior puzzles it looks like a quilt pattern. Maybe that is just because all the puzzles are symetric in some way. I'll have to try some out.
doduff
 
Posts: 32
Joined: 29 May 2006

Re: very superior: Sunday Times #44 (2006.June.4)

Postby Archangel » Tue Jan 02, 2007 10:41 am

Pat wrote:— and to celebrate the closing of the "superior" list, The Sunday Times #44 (2006.June.4) is very superior — seems way above their usual level — i suspect it would rate quite high in our "superior" Topic


I am a newbie. I do not know anything about X-Wings etc.

But I applied my amateur's program in Fortran 77 (!) which simply applies the fundamental rules (just detecting the digits appearing only once in a row or a column or a block, without considering doubles etc); it stopped at a scheme like this:

Code: Select all
569 1 69 | 8 2 34 | 49 7 34569
3 689 6789 | 5 4679 47 | 2 489 1
56789 2 4 | 369 679 1 | 89 3589 3569
---------------------------------------------
4689 7 1 | 3469 4689 2348 | 5 349 349
2 3469 369 | 3469 4569 345 | 7 1 8
489 3489 5 | 1 4789 3478 | 6 2 349
---------------------------------------------
489 489 289 | 7 1 2458 | 3 6 2459
1 3468 2368 | 24 458 9 | 48 458 7
4789 5 2789 | 24 3 6 | 1 489 249


Then simply trying r1c3<>6 my program went on and solved the puzzle.
So what ?
Archangel
 
Posts: 5
Joined: 30 December 2006

Re: very superior: Sunday Times #44 (2006.June.4)

Postby ab » Tue Jan 02, 2007 4:09 pm

Archangel wrote:
Pat wrote:— and to celebrate the closing of the "superior" list, The Sunday Times #44 (2006.June.4) is very superior — seems way above their usual level — i suspect it would rate quite high in our "superior" Topic


I am a newbie. I do not know anything about X-Wings etc.

But I applied my amateur's program in Fortran 77 (!) which simply applies the fundamental rules (just detecting the digits appearing only once in a row or a column or a block, without considering doubles etc); it stopped at a scheme like this:

Code: Select all
569 1 69 | 8 2 34 | 49 7 34569
3 689 6789 | 5 4679 47 | 2 489 1
56789 2 4 | 369 679 1 | 89 3589 3569
---------------------------------------------
4689 7 1 | 3469 4689 2348 | 5 349 349
2 3469 369 | 3469 4569 345 | 7 1 8
489 3489 5 | 1 4789 3478 | 6 2 349
---------------------------------------------
489 489 289 | 7 1 2458 | 3 6 2459
1 3468 2368 | 24 458 9 | 48 458 7
4789 5 2789 | 24 3 6 | 1 489 249


Then simply trying r1c3<>6 my program went on and solved the puzzle.
So what ?


I presume you mean r1c3=6. That's what's known as a backdoor single. Basically it's a guess and you see where it leads. If you guess right then that's fine, but if you're solving by hand and you guess wrong, then it can be a mess.

I wrote a solver that uses backdoor singles to finish puzzles that it can't do by logical deduction, but there are some puzzles that don't fall to backdoor singles. Look here at Ocean's puzzle for example.
ab
 
Posts: 451
Joined: 06 September 2005

PreviousNext

Return to General

cron