Help spotting swordfish?

Advanced methods and approaches for solving Sudoku puzzles

Help spotting swordfish?

Postby rbnn » Sat Feb 16, 2008 8:51 pm

Can anyone help me spot swordfish?

I've been working through the six-star puzzles in More Sudoku and the last several have used swordfish. I always carefully check for swordfish when I'm at an impasse, but it seems like I can never see them. (I know they are there afterwards when I run the solutions in SadMan.)

It's also very time-consuming for me to check - I usually find trial-and-error is faster than looking for swordfish.

Oh, I am using pen-and-paper. Online, I can highlight each number successively in Sadman and see the swordfish more easily (even there it's not trivial).
rbnn
 
Posts: 3
Joined: 16 February 2008

Postby Bigtone53 » Sat Feb 16, 2008 9:50 pm

Can anyone help me spot swordfish?


Hi, rbnn. Welcome to the forum.

If you go to the section called 'Help with particular puzzles' you will see a very recent chain called HELP, started by Dogguk. His initial question was on colouring but it developed into an explanation of swordfishes. Not only did Cec give a careful explanation, but I gave 3 puzzles which are most simply solved via swordfishes.

Knowing that they are there may make it easier to spot them. Good luck:D
Bigtone53
 
Posts: 413
Joined: 19 September 2005

swordfish spotting techniques?

Postby rbnn » Sat Feb 16, 2008 11:01 pm

I understand what swordfish are and why the work.

I am asking whether there are specific techniques, usable with pencil-and-paper, actually to find a swordfish in a reasonable amount of time.

I mean, what process do you use to look for swordfish in a diagram - this is the part I do not do well or efficiently. Is there some trick to find them faster?

Bigtone53 wrote:
Can anyone help me spot swordfish?


Hi, rbnn. Welcome to the forum.

If you go to the section called 'Help with particular puzzles' you will see a very recent chain called HELP, started by Dogguk. His initial question was on colouring but it developed into an explanation of swordfishes. Not only did Cec give a careful explanation, but I gave 3 puzzles which are most simply solved via swordfishes.

Knowing that they are there may make it easier to spot them. Good luck:D
rbnn
 
Posts: 3
Joined: 16 February 2008

Re: swordfish spotting techniques?

Postby Mauricio » Sat Feb 16, 2008 11:28 pm

rbnn wrote:I mean, what process do you use to look for swordfish in a diagram - this is the part I do not do well or efficiently. Is there some trick to find them faster?
Practice.
Mauricio
 
Posts: 1174
Joined: 22 March 2006

Postby tarek » Sun Feb 17, 2008 10:42 am

if you are a pencil & paper solver & you do not use a pencilmark grid.... then it will be extremely difficult to spot one without practice. these puzzles should be designed to have swordfish following singles. the pure swordfish collection thread is a good start.

if you are a pencilmark grid user then my suggestion in addition to what Mauricio said is to FOCUS on single digits & how they are arranged within the grid.

The FISH are part of what we call X-cycles which means that deductions based on the pattern of a single digit would help us eliminate some candidtaes of that very same digit.

tarek
User avatar
tarek
 
Posts: 2622
Joined: 05 January 2006

Postby rbnn » Sun Feb 17, 2008 10:57 am

I have practiced, I still can't seem to spot them in the field. I do use a penciled candidates.

Can you say in more detail how you look for swordfish?

Say you are stuck on a puzzle.

Do you say: let's look for swordfish in the 1s, row by row then column by column; then in the twos, and so on?

But even knowing there is a swordfish in say, the 7s, it just seems very difficult even to find the fish if there are a lot of candidate 7s in the grid. There are so many different possible row and column combinations.
rbnn
 
Posts: 3
Joined: 16 February 2008

Postby tarek » Sun Feb 17, 2008 11:11 am

rbnn wrote:a lot of candidate 7s in the grid. There are so many different possible row and column combinations.
this is a good observation....

For a swordfish you need 3 rows & 3 columns without PLACEMENT for that digit.

if you have 7 digits of that candidtae already placed then you will not find a swordfish.

What I just said is probably very trivial.

I've read somwhere else that how much effort you put into a technique should have an equal reward.

The method heirachy thread tells you which techniques yield more KILLS, Swordfish & Jellyfish tend to be RARER & therefore some people avoid using them all together & replaced them with MORE DIFFICULT TO SPOT techniques that yield more kills.

tarek
User avatar
tarek
 
Posts: 2622
Joined: 05 January 2006

Postby Bigtone53 » Mon Feb 18, 2008 5:55 am

Can you say in more detail how you look for swordfish?

Say you are stuck on a puzzle.

Do you say: let's look for swordfish in the 1s, row by row then column by column; then in the twos, and so on?

But even knowing there is a swordfish in say, the 7s, it just seems very difficult even to find the fish if there are a lot of candidate 7s in the grid. There are so many different possible row and column combinations


As a purely mechanical way of finding xwings, swordfish etc, Arcilla's method takes some beating.

Unfortunately, I am sitting in a hotel room 3000 miles from home and cannot access his explanation but I am sure that one of the experts can point you to this. You may even be able to google it,

[edit] try this

http://forum.enjoysudoku.com/viewtopic.php?t=5017
Bigtone53
 
Posts: 413
Joined: 19 September 2005

Postby Myth Jellies » Sun Feb 24, 2008 7:35 am

Note that the vast majority of swordfish by far will have at least one of the nine possible digits missing. What this means is that you can use sort of a finned x-wing as a marker for finding nearly all swordfish.

The progression is bi-located pair in a row or column...
Code: Select all
 1   /   /   1   /   /   /   /   /

...to almost an x-wing (hey you are probably looking for x-wings as well, right)...
Code: Select all
 1   /   /   1   /   /   /   /   /
 .   .   .   .   .   .   .   .   .
 .   .   .   .   .   .   .   .   .
 .   .   .   .   .   .   .   .   .
 1   /   /   1   /   /   /   1   /
 .   .   .   .   .   .   .   .   .
 .   .   .   .   .   .   .   .   .
 .   .   .   .   .   .   .   .   .
 .   .   .   .   .   .   .   .   .

...then look to complete the swordfish...
Code: Select all
 1   /   /   1   /   /   /   /   /
 *   .   .   *   .   .   .   *   .
 *   .   .   *   .   .   .   *   .
 *   .   .   *   .   .   .   *   .
 1   /   /   1   /   /   /   1   /
 *   .   .   *   .   .   .   *   .
 *   .   .   *   .   .   .   *   .
 *   .   .   *   .   .   .   *   .
 1   /   /   1   /   /   /   1   /
Last edited by Myth Jellies on Tue Feb 26, 2008 3:57 am, edited 1 time in total.
Myth Jellies
 
Posts: 593
Joined: 19 September 2005

Re: Help spotting swordfish?

Postby denis_berthier » Sun Feb 24, 2008 1:21 pm

rbnn wrote:Can anyone help me spot swordfish?

Arcilla introduced a new view of fish here: http://forum.enjoysudoku.com/viewtopic.php?t=5017&start=0&postdays=0&postorder=asc&highlight
I wasn't aware of this when I wrote my book and my first posts on the rn and cn spaces, which can be considered a formalisation and a generalisation of Arcilla's view.
The rn and cn- spaces are defined formally in my book and less formally in these threads:
http://www.sudoku.org.uk/SudokuThread.asp?fid=4&sid=9235&p1=2&p2=10
http://forum.enjoysudoku.com/viewtopic.php?t=5555

They can be used for finding much more than fish, e.g. hidden xy, xyt and xyzt chains.
On my web pages, you can find an extended Sudoku board including these spaces. It can be helpful for finding fish and other patterns.

rn and cn spaces are implemented in Ruud's Sudocue program.
denis_berthier
2010 Supporter
 
Posts: 1253
Joined: 19 June 2007
Location: Paris

Postby Luke » Wed May 14, 2008 5:16 am

Some thoughts on swordfishing for rbbn….

1.) Patterns
2.) Filtering
3.) The fin factor
4.) Expand the horizon

1.) Patterns
You know, I feel your pain. You can’t find a swordfish. I can’t find a so-called “almost locked set” to save my supper. I guess we all have our nemeses! One thing in your favor is that if there is a swordfish, there might be a pattern to help you find it. Look for the pattern.

Some 4’s in this puzzle fall in exactly the same positions in two different rows, 5 and 9.
Code: Select all
 *--------------------------------------------------------------------*
 | 379    8      3679   | 2379   3679   2367   | 14679  5      1469   |
 | 4      1      5679   | 579    8      67     | 679    2      3      |
 | 2      367    35679  | 4      35679  1      | 679    689    689    |
 |----------------------+----------------------+----------------------|
 | 357    367    8      | 1      357    4      | 2      69     569    |
 | 135    9      13456  | 35     2      8      | 1456   7      1456   |
 | 157    47     2      | 6      57     9      | 3      48     1458   |
 |----------------------+----------------------+----------------------|
 | 1379   347    13479  | 8      13679  5      | 469    469    2      |
 | 6      5      39     | 239    4      23     | 8      1      7      |
 | 8      2      1479   | 79     1679   67     | 4569   3      4569   |
 *--------------------------------------------------------------------*

Thumbnail:
Image

That’s 2/3 of a swordfish and a definite pattern. When I see something like this I try to complete the pattern.

2,) Filtering
I heard of a savant who took an airplane ride over a city, just once. Back on terra firma he was able to remember and draw a map of every street, building, church, etc. that he had seen on his ride. Just for the record, that savant was NOT me:D . (I have only mastered the first half of idiot savant….)

I certainly cannot look at a puzzle and readily sort all of one candidate for analysis, savant-like. So I say, filter ‘em. The computer makes this easy, but you mention you are a paper and pencil solver. Fine! So am I. I use a small 9 by 9 grid at the top of the puzzle and lightly circle the candidate I’m filtering. Never takes more than abt 20 seconds to do the imput. When I’m done with it, I erase the circles. This almost always ends up revealing something nice, like maybe a swordfish.

Image

3.) The Fin Factor
Once your candidate is filtered and there’s nothing fishy, it’s time to look for the fin. Or fins. I find far more finned fishes than naked fishes. I'm not really sure swords are all that common to begin with. I haven’t got databases of puzzles to crunch like the savants around here, but I’d bet for every naked fish you hook in super hard puzzles, there are eight or ten finned ones waiting to be consumed! I’m including finned X-wings in this claim. This makes me wonder, why even bother to sweat the swordfish? If it’s there, you’ll find it while plying more productive waters.

And don’t forget your sashimi. Fish don’t have to have your candidate where you want it to allow an elimination. Solved cells can hold a spots in the pattern where you’d like to see your candidate. This is esp useful in bigger fish.

4.) Expand the Horizon
When you’ve got a particular candidate filtered, no reason to stop with fishes. Take a second and connect all the strong links with a pencil line. Now seek out things like skyscrapers, empty rectangles, X-wings, kites, X-cycles, grouped X-cycles, hinges, AIC’s, and on and on. Pick your favorites, the ones you’re good at.

This should take some of the drudgery out of fishing for things that are hard to catch. To me, the real fun in this game is not solving the dang puzzle. It’s solving it in a satisfying fashion.
User avatar
Luke
2015 Supporter
 
Posts: 435
Joined: 06 August 2006
Location: Southern Northern California

Postby ab » Wed May 14, 2008 9:42 pm

Arcilla's method is good. A different approach is Ruud's fishing with lines. Look here:
http://www.sudocue.net/guide.php
ab
 
Posts: 451
Joined: 06 September 2005

Postby Pat » Thu May 15, 2008 1:44 pm

Luke451 wrote:
3.) The Fin Factor

I’d bet for every naked fish you hook,
there are eight or ten finned ones waiting to be consumed!



yes, definitely many more finned than unfinned
User avatar
Pat
 
Posts: 3438
Joined: 18 July 2005

Re: Help spotting swordfish?

Postby wintder » Fri May 16, 2008 9:48 pm

rbnn wrote:Can anyone help me spot swordfish?

I always carefully check for swordfish when I'm at an impasse, but it seems like I can never see them.


I like and use x-wing, skyscraper and swordfish.

I look for them at the same time.

Typically, I look for a long strong link. Then I check the row (or column), at right angle to the first, for another strong link on the same digit. If there is a second and it does not reach far enough for a sky-x pattern I look across for another strong OR weak link that hooks up to be a sword, finned sword, or sashimi sword.

I usually look in low candidate cells to start.

Swordfish usually get lost in smaller patterns so it is often true they they are not robust enough to survive until a point where they are required.
wintder
 
Posts: 297
Joined: 24 April 2007

MS Access Solution

Postby Bear » Fri May 30, 2008 5:11 pm

I am solving all the SudokuSolver very difficult grids in about 30 minutes.

I only look for x-wing & swordfish after all the simplier patterns have been cleaned up -- a la the methods described on SudokuSolver.

At 1st i was using a screenshot utility & its highlighter to make digit patterns stand out. That was cumbersome so i wrote an MS Access application that highlights cells w/ the target digit. When a digit is double-clicked, all cells w/ the digit are highlighted.

Then i made a 2nd form that just shows rows w/ 2 or 3 occurrences of the target digit. And a 3rd form for columns w/ 2 or 3 occurrences. This quickly reveals x-wing & swordfish patterns & also aides w/ exclusive pairs (2-cell chains).

The application records each cell change & records the date & time. This is helpful as chains eliminate digits, rows & columns can be re-examined quickly for newly revealed patterns for the eliminated digits.

The application does NOT solve the puzzles, it is a helper ONLY & just implements my screenshot method w/o using screenshots. I saved each change record for all 19 of the very difficult grids on SudokuSolver -- for study as time permits. I still find new strategies from time-to-time.

Requires MS Access 2000 or above. It was not designed to address grids that are not logic bound -- requiring guesses.
Bear
 
Posts: 2
Joined: 30 May 2008


Return to Advanced solving techniques