Solving Hierarchy

Ask your support questions for SudoCue here
Post Reply
MorganNY
Regular
Regular
Posts: 17
Joined: Thu Jan 04, 2007 8:48 am
Location: Long Island, NY

Solving Hierarchy

Post by MorganNY »

Sensei Ruud,

Status: v. 2.2.0.0, all solving tools enabled. Illustration included:

I got to a point in an 'unfair' puzzle. The Vague Hint reported my next step would be a Medusa Bridge, which means (to me) start coloring a [bivalue cell] candidate and follow the event chain, etc. A step I consider should be a virtually last resort to solving a puzzle. However, I spotted what appeared to be a finned X-wing, obviously an easier and more basic elimination than massive coloring. After the elimination, I was left with 4 single candidates, which I filled. F7 still reported a Medusa Bridge when in fact there was a pair of locked candidates, the removal of which revealed a chain of singles that solved the puzzle. Usually, I resort to F7 if I'm stuck and often find there was a naked quad, kite, etc. that I missed. I always assumed the program would offer the easiest next step.
So, either I misidentified the finned x-wing (I'm just starting to get 'fins') and got lucky on the 50/50 chance, or the program identified a more difficult (and unnecessary) pattern than indeed existed. Even if I did get lucky, 'cue still should have reported the resulting locked candidates. I made an image of the game state for your examination.

Image

I appreciate your time, as always. Hope I didn't waste it.
As always I remain,
Your humble student,
Morgan
rep'nA
Hooked
Hooked
Posts: 49
Joined: Fri Jan 19, 2007 11:37 am
Location: Union City, California

Post by rep'nA »

Your finned x-wing in r15c15 with fin r2c1 would eliminate 6's from r1c2 and r1c3. The way I think of a finned x-wing is the following. If the fin wasn't there you would have an x-wing and you could make certain eliminations. If the fin was true then you could make other eliminations. Any candidate eliminated in either case is safely eliminated. So the eliminations need to be both in the same block as the fin and in the same row (or column) as one side of the x-wing.
"Obviousness is always the enemy to correctness."-Bertrand Russell
MorganNY
Regular
Regular
Posts: 17
Joined: Thu Jan 04, 2007 8:48 am
Location: Long Island, NY

Post by MorganNY »

rep'nA, I appreciate your interest and consideration. However, even converting your answer to the 'Cue default row/column numbering(I changed my row/column numbering scheme to conform with a site I frequent, http://sudoku.com.au), I'm not sure I understand your answer. r1c3 doesn't contain a 6 in either numbering schemes. In any event, my question to Ruud was less about my perceived finned x-wing and more about the hierarchy of the solver regarding the technique suggested for the next 'step'. That's why it's in the software forum. Either my fx-wing was right and the program offered a more complicated elimination than a more simple one already available, or I was wrong and I got lucky, which led to an available 'locked candidate' step, which also should have been suggested by the solver before medusa coloring, assuming there is an actual hierarchy of techniques. It's been my experience that the solver, or in this case, the hinter, would offer the least advanced method of progress. Naked pairs before Nishio, for instance. The correctness of my applied technique, or my lucky misapplication of same that didn't invalidate the puzzle, was secondary to the question, as the topic states, the "Solving Hierarchy" of the SudoCue program.
Still, I appreciate your desire to help.
rep'nA
Hooked
Hooked
Posts: 49
Joined: Fri Jan 19, 2007 11:37 am
Location: Union City, California

Post by rep'nA »

MorganNY wrote:rep'nA, I appreciate your interest and consideration. However, even converting your answer to the 'Cue default row/column numbering(I changed my row/column numbering scheme to conform with a site I frequent, http://sudoku.com.au), I'm not sure I understand your answer. r1c3 doesn't contain a 6 in either numbering schemes.
I'm sorry I wasn't more explicit. When I said that
rep'nA wrote:Your finned x-wing in r15c15 with fin r2c1 would eliminate 6's from r1c2 and r1c3.
I meant that these are the possible eliminations in general, even though in this case no such elimination is possible. In any case, the elimination you made is valid in the puzzle, but is not a valid finned x-wing exclusion.
MorganNY wrote: In any event, my question to Ruud was less about my perceived finned x-wing and more about the hierarchy of the solver regarding the technique suggested for the next 'step'. That's why it's in the software forum. Either my fx-wing was right and the program offered a more complicated elimination than a more simple one already available, or I was wrong and I got lucky, which led to an available 'locked candidate' step, which also should have been suggested by the solver before medusa coloring, assuming there is an actual hierarchy of techniques.
So, the first answer is that your finned x-wing exclusion is incorrect. You are however right that you got lucky and revealed a locked candidate step. Actually you revealed something easier, namely a hidden single in r6c2. As to why SudoCue doesn't notice this, Ruud can answer better than I. However, I have noticed that SudoCue does not update the Analyzer after a move is made outside of its solution path, at least not enough to catch this.
"Obviousness is always the enemy to correctness."-Bertrand Russell
Ruud
Site Owner
Site Owner
Posts: 601
Joined: Fri Dec 30, 2005 10:21 pm

Post by Ruud »

rep'nA wrote:However, I have noticed that SudoCue does not update the Analyzer after a move is made outside of its solution path, at least not enough to catch this.
That is the main reason. SudoCue completely solves a puzzle when you load it into the program, and builds a list of hints from the moves it made. When you ask the program for a hint, it will pick the first hint from this list where placements/eliminations have not been (completely) performed by the player.

When you follow a different solving path, easier moves may be unlocked which are not in the hint list. To refresh this list, use the Solve from here menu option. The solver will then build a new solving path with hints starting from the current state of the puzzle. The difficulty rating will be updated to show the difficulty for completing the remaining puzzle.

Ruud
“If the human brain were so simple that we could understand it, we would be so simple that we couldn't.” - Emerson M Pugh
MorganNY
Regular
Regular
Posts: 17
Joined: Thu Jan 04, 2007 8:48 am
Location: Long Island, NY

Post by MorganNY »

Thanks folks. Those fins drive me nuts. The solver points them out and I try to understand the valid conditions. Once of these days...
I did have fun creating the image in photoshop and figuring out how to imbed it. So learning did occur :)
Cheers to rep'nA and a bow to Sensei Ruud
Pete
Gold Member
Gold Member
Posts: 113
Joined: Sat May 13, 2006 6:12 pm
Location: Cincinnati Ohio, USA

Post by Pete »

"That is the main reason. SudoCue completely solves a puzzle when you load it into the program, and builds a list of hints from the moves it made. When you ask the program for a hint, it will pick the first hint from this list where placements/eliminations have not been (completely) performed by the player.

When you follow a different solving path, easier moves may be unlocked which are not in the hint list. To refresh this list, use the Solve from here menu option. The solver will then build a new solving path with hints starting from the current state of the puzzle. The difficulty rating will be updated to show the difficulty for completing the remaining puzzle."
thanks Ruud this has answered a lot of my questions.

when i get stuck i load the puzzle into your program get a hint and go on to find some other path then the answers/hints start getting strange

sumocue has no "solve from here " button and that could explain a lot

i start a lot of killers with the triples that are in the 3cages
sumocue seems to look first for innies and outies

so the error messages when i take out a wrong digit are sometimes real strange i just assumed i did something wrong but then the answer is correct.

I seem to find xwings that arent there too. but they seem to be valid

my last killerjigsaw had 3 examples of this...


when i try to do walk thrus and skip a step or two i can never seem to get back to the posistion in the last picture...

In short the solver goes down some path that i dont.

so perhaps ALL the solvers should maybe have a "solve from here"
as an optional step after any manual intervention.
That would certainly be a boon to the learning process.
"It gets dark at night" - Olbers
MorganNY
Regular
Regular
Posts: 17
Joined: Thu Jan 04, 2007 8:48 am
Location: Long Island, NY

Post by MorganNY »

Pete, I've found, as I use 'cue to learn to recognize kites and fish etc and expecially coloring, creating a second SudoCue instance and using that to experimentally place something without making the puzzle 'invalid' or losing my coloring (as with Solve from Here).
You can run as many instances as you like of SudoCue
Post Reply