The New Sudoku Players' Forum

[Pat]

here's a splendid example -- try solving it on paper, no help from the computer --

1 hour. No computer

hi Archangel,

was it tricky as promised? what did it take to solve it?

P.S. please do not post the answer, i'm only interested in the solution-path

[Pat]

This puzzle is harder than 5-star puzzles on my newspaper

well, GoSimon, i don't know about the "stars" in the newspaper -- so how about you trying the following example and tell me how many stars it rates?

[ 35 clues, no symmetry ]

Code: Select all

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



~ Pat

[GoSimon]

This puzzle is harder than 5-star puzzles on my newspaper

well, GoSimon, i don't know about the "stars" in the newspaper -- so how about you trying the following example and tell me how many stars it rates?

[ 35 clues, no symmetry ]

Code: Select all

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



~ Pat

Well, this one is definitely off the star-chart. I cannot even start it...

[udosuk]

Well, this one is definitely off the star-chart. I cannot even start it...

A grid of pencilmarks would help heaps to solve that one (unless you're RW)...

Here is such a grid, drawn up with the help of a computer program:

Code: Select all

 78     6      5      | 23     237    4      | 9      1      28     
 178    278    4      | 9      1267   167    | 3      5      268   
 3      279    179    | 5      1267   8      | 2467   247    246   
----------------------+----------------------+---------------------
 16789  789    16789  | 126    4      1567   | 256    289    3     
 5      349    169    | 1236   8      136    | 246    249    7     
 2      3478   678    | 36     9      3567   | 456    48     1     
----------------------+----------------------+---------------------
 4689   489    689    | 7      36     2      | 1      34     5     
 67     5      2      | 4      136    136    | 8      37     9     
 47     1      3      | 8      5      9      | 247    6      24     

Although ideally we should draw it manually on paper using a pen...

[Pat]

Well, this one is definitely off the star-chart.
I cannot even start it...

GoSimon, you clearly don't know about box-line interactions

angusj has a good description, if you wish to learn

After generating candidates and checking naked 2/3/4, I got stuck --

hi GoSimon.

it seems many people have managed to learn all about "naked 2/3/4" while skipping over box-line interactions (or "Locked Candidates" as angusj calls them)

so you're missing much of the fun

[RW]

A grid of pencilmarks would help heaps to solve that one (unless you're RW)...

Oh, come on! You need pencilmarks for that...? Two locked candidates and the rest is hidden singles... Nice ulterior puzzle!

[Pat]

A grid of pencilmarks would help heaps to solve that one (unless you're RW)...

Oh, come on! You need pencilmarks for that...?
Two locked candidates and the rest is hidden singles...
Nice ulterior puzzle!

i'm sure GoSimon knows all about "pencilmarks" --
and does not know about "locked candidates"

the puzzle was posted in order to confirm the above diagnosis
+ perhaps convince him it's worth learning
just one "locked candidates" is actually needed for solving this puzzle (rubylips - 22 clues - which i recently re-posted in the "ulterior" Topic).

the puzzle was chosen because it cannot be solved by "singles" + subsets.

[udosuk]

RW is right... You need only 2 locked candidates/box-line interactions to reduce the puzzle to hidden singles only...

Pat is right too... If you considered a puzzle with naked singles and hidden singles left as being "solved" then just 1 locked candidate move is needed...

Code: Select all

7 in b7 locked in c1 => r124c1<>7

Afterwards, only naked singles and hidden singles are required to reach the solution...

But if like RW you don't like pencilmarks, you might not want to do any naked single... So one more move to do:

Code: Select all

7 in b3 locked in r3 => r3c235<>7

Now you can reach the solution even if the only one move you ever know is called "hidden single"...

And to start this puzzle without any pencilmarks:

Code: Select all

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

Just notice the ## cells contain the only possible cells holding the 7 within the corresponding box... As a result, all the xx cells cannot hold a 7... Which leaves only 1 possible cell to hold 7 in box 1... Afterwards, you can fill in more cells in box 1 because there is only one possible placement for some digits within that box...

[Pat]

just one box-to-line exclusion is needed
box7 7 is in c1 ==>
everything is "hidden singles"

probably "3 stars"

too bad GoSimon is showing no interest --
but then again, he never asked for this advice, i'd better shut up.

[udosuk]

just one box-to-line exclusion is needed

box7 7 is in c1 ==>
everything is "hidden singles"

I missed the "well-hidden" hidden single of 7 on r1 after the first move...

Anyway I blame RW (the pencilmarkless solving expert) who inexcusably also missed that, and the lack of the feature in Simple Sudoku to shift the priority of hidden singles up above naked singles...