How do different people solve Sudoku !!

Advanced methods and approaches for solving Sudoku puzzles

How do different people solve Sudoku !!

Postby dipen01 » Sun Oct 23, 2005 10:12 am

Hi everybody....


Well today is miserable Sunday for me.. I am just coming from a competition called Times Sudoku...there were aroung 500 people in a big auditorium. we were given a close post card in which our puzzle was printed.. there were four types of card labelled A,B,C,D each having same puzzle but rotated 90 degrees in each case (i came to know this after everything was over) so as to avoid copying. On a specific time we were asked to open it then count began.

I was very tensed. so tensed that i even coudn't concentrate on my own puzzle. was wandering how much my felow has completed. Am i lagging am i leading. then within 4 min. one person completed. that again distracted me.. In all this distractions i made a mistake somewhere which spoiled my puzzle and i couldnt finish it either... So am really disappointed over that. After everythin was over they declared that puzzle was of Medium difficulty ... i cursed myself as i am able to solve medium puzzles in nearly 7-8 min. and i couldn't do it even in 15...

Thats why thought of sharing this with you guys... so that i can understand some of many things...

-------------------------------------------------------------------------------------

i am sure lot of experts are between us... so i would like them to share something here....

Well...since a long time i have been curious about how other people solve sudoku i mean.

So i thought there would be no better place than our forum to share each others techniques and methods which eventually will help overbody overcome thier weaknesses..



1) With which method do you start solving sudoku..

2) How is the flow of your puzzle..i mean which sequence of methods you people try..e.g first single sets then hidden....and....

3) What do you do if u are stuck at one point that no options are visible..

4) VVIMP question : Do you use pencil-eraser and pen or only pen is sufficient.. if only pen then how do you manage.


I hope everybody will co-operate here...:D
dipen01
 
Posts: 15
Joined: 26 September 2005

Postby Max Beran » Sun Oct 23, 2005 1:45 pm

I'll start the ball rolling even though it makes my toes curl and hair stand on end to even think about the horrors you put yourself through. But like you I tend to use the puzzles in the papers as a speed trial, albeit personal, and am probably of comparable standard. Actually I find that I get them wrong 25-50% of the time when I use the paper itself and I deliberately rush. Resizing the puzzles to an A4 sheet dramatically improves my getting-it-right rate.

However the real reason why I'm putting finger to keypad is because I'm a keen advocate of minimal candidate pencilling and I'm not sure if this is a common strategy. I'm not theological about writing absolutely nothing but limit entries strictly to candidates that are in a row or a column in a box. I even eschew pairs in a box unless they are in a line. So my initial speed-scanning is directed to filtering for these possibilities.

To this end I head straight for the rows or columns of boxes that visually have the most amount of black in them rather than taking them in an orderly fashion. As others have noted, horizontal scanning is preferred to vertical so I go for rows ahead of columns. I'm on the lookout for completed or nearly completed lines in a box as these most often translate into filled cells in that and neighbouring boxes. I'm especially keen on two in a line with candidates at right angles that preclude those values from the missing cell of the line. This provides the paired candidates (and sometimes even filled cells) that I'm on the lookout for. One habit I've got into which I fancy may be suboptimal is whenever I complete a cell I can't resist checking what it's implications are in the other direction to the current scan. The same applies when chance has it that the lie of the candidates create a pinned situation. It may be better to get on with the current scan and pick up the implications later when performing that scan.

Another rule which might or might not be condusive to speed is that whenever a row, column or box has six filled cells I switch from scan-mode to full analysis-mode to complete the candidate list for the incomplete cells. This is a departure from my "line" rule mentioned above and I distinguish them with a pencil stroke around the candidates. Again it might be more efficient to stay with the scan and do this job later.

One reason why I feel confident that my lean and mean approach to pencilling in candidates is a good one is when I compare it with say the Simple Sudoku program. It's quite a rarity that I have to eliminate candidates due to naked pairs and triples or because of pinning as such candidates seldom get written in in the first place.

At later stages when analysis takes over from scanning then the capability to scan a row, column or box and almost immediately "know" what are the missing candidates is an obvious asset. It's probably the point at which I tend to make most mistakes. I also like to keep an eye out for intersections where the candidate list will be shortest. Obviously this increases the chances of spotting a naked triplet or whatever at an early point with obvious benefits for clearing the decks. Surprisingly often the dearth of pencilled-in candidates is a clue prior to enumerating.

You ask about implements - I never use a pen however I do get through rubbers (whoops; erasors - there may be Americans listening in) like a dose of salts.

I too look forward to picking up hints and tips from others.
Max Beran
 
Posts: 57
Joined: 17 August 2005

Postby dipen01 » Sun Oct 23, 2005 3:08 pm

Max Beran wrote:
One habit I've got into which I fancy may be suboptimal is whenever I complete a cell I can't resist checking what it's implications are in the other direction to the current scan. The same applies when chance has it that the lie of the candidates create a pinned situation. It may be better to get on with the current scan and pick up the implications later when performing that scan.


Our approach till here is same..even i cannot resist myself from checking implications..:)

By the way it must be taking a long time for you to solve a puzzle by your methods.. as u scan 2-3 times the whole puzzle...and scanning is the only thing which takes time in solving puzzle.. the writing job is quickier...

Thanx a lot for your reply...

Lets hear from others...
dipen01
 
Posts: 15
Joined: 26 September 2005

Postby dipen01 » Mon Oct 24, 2005 5:20 am

Well.. a little surprise ..

No Replies..... Dunno why...
------------------------------------------------------------------------------------

The competition in which i participated and i lost in the first round of it.. So the person who won it... solved it in precisely 70 secs:!:
His name is flashed in newspapers today.. Co-incidently he was the winner of my batch also...

How come people solve in 70 secs....Do u have some much time to apply so many techniques and methods...(ofcourse if needed)...
dipen01
 
Posts: 15
Joined: 26 September 2005

Postby Guru » Tue Oct 25, 2005 12:53 pm

I use only a pen when solving su dokus.
I usually start with the 1 digits, then the 2 digits etc., writing all candidates in the top of the cells whenever i find them. ( this takes some time, but i find it rewarding when solving harder puzzles)

If there are any naked singles, i write them in.
Then i immediately starts looking for hidden subsets, as this usually eliminates most of the candidates i otherwise would have eliminated with hidden singles/ block/block interactions, naked subsets and block/column/row interaction.

Then i immediately starts looking for grids with n columns and n rows with a single candidate, as this rules out the fish cycles and XY wings.

After this i checks thoroughly for coloring, although this takes much time...

Then forcing chains, nishio and BB is the next steps to completion.

Sorry if this post is unclear, it's hard to describe the techniques used...
Guru
 
Posts: 5
Joined: 25 October 2005

Postby Bigtone53 » Tue Oct 25, 2005 2:57 pm

Readers of The Times will have had the dubious benefit of seeing how Carol Vordemann goes about solving, namely by writing in every possibility for every square.

I can see that this will lead to a solution in most cases (although the more advanced solving techniques are not mentioned) but I find this mentally unsatisfying. Speed is not my primary object - I prefer the mental challenge of not writing in anything other than final 'for certain' numbers. Admittedly I go over what I have thought before many times but it is a bit like chess; you tend to see the patterns after a while without writing them down.

When I fall from these high standards, as I usually have to for a Sunday Times Superior, I pencil in fixed pairs, but nothing else.

I guess that it is each to their own.
Bigtone53
 
Posts: 413
Joined: 19 September 2005

Postby PaulIQ164 » Tue Oct 25, 2005 3:03 pm

Bigtone53 wrote:Readers of The Times will have had the dubious benefit of seeing how Carol Vordemann goes about solving, namely by writing in every possibility for every square.

I can see that this will lead to a solution in most cases (although the more advanced solving techniques are not mentioned) but I find this mentally unsatisfying. Speed is not my primary object - I prefer the mental challenge of not writing in anything other than final 'for certain' numbers. Admittedly I go over what I have thought before many times but it is a bit like chess; you tend to see the patterns after a while without writing them down.

When I fall from these high standards, as I usually have to for a Sunday Times Superior, I pencil in fixed pairs, but nothing else.

I guess that it is each to their own.


This pretty much sums up my philosophy too.
PaulIQ164
 
Posts: 533
Joined: 16 July 2005

Postby CathyW » Tue Oct 25, 2005 3:31 pm

Bigtone53 wrote:I guess that it is each to their own.


I would agree with that! I also prefer to keep candidate entries to a minimum when solving on paper although I haven't yet managed a Times Fiendish without any at all. Seems to help to see the pairs, triples and locked candidates more easily.

I use the software (Sadman or Pappocom) if I get stuck or if pasting in a puzzle from the forum or elsewhere. Tend to use Sadman more these days because of being able to paste directly from clipboard and the automatic candidate entry is useful when trying to follow/learn some of the more advanced techniques.
CathyW
 
Posts: 316
Joined: 20 June 2005

Postby QBasicMac » Tue Oct 25, 2005 8:07 pm

dipen01 wrote:How come people solve in 70 secs


Heh - I once took a Rubik's cube to the office. I had a little cheat-sheet book which gave rules that eventually led to the solution. I had more-or-less memorized it and could amaze some colleages solving in, say, 15 minutes.

Well I gave it to one person who said (and I believe him) he had never tried one before. He picked it up, stared a few seconds, and immediately solved it. He did it again and again as we shuffled at random. On being asked for his technique, he looked puzzled and said "What technique? You just turn it until the colors match!". Like a famous sculptor who said "You take a block of stone and chip away anything that does not look like (whatever he was sculpting)."

Bottom line: some people have a natural talent for some things. You can read all the techniques you want, but you can't play a violin like Jascha Heifetz unless you have that particular talent.

Accept it and just enjoy solving puzzles, is my advice.

Mac
QBasicMac
 
Posts: 441
Joined: 13 July 2005

Postby dipen01 » Wed Oct 26, 2005 7:41 am

Completely agree with you...QbasicMac...
dipen01
 
Posts: 15
Joined: 26 September 2005

Postby emm » Wed Oct 26, 2005 8:29 am

I agree with Mac too. And you might be being too hard on yourself dipen01 - I imagine that doing Sudoku in a competition is very different from doing them on your own. I could easily freeze up if I was trying to beat someone else's time.

Float like a leaf on the river of life.
emm
 
Posts: 987
Joined: 02 July 2005

Postby Guru » Wed Oct 26, 2005 2:58 pm

I would agree with that! I also prefer to keep candidate entries to a minimum when solving on paper although I haven't yet managed a Times Fiendish without any at all. Seems to help to see the pairs, triples and locked candidates more easily.


Of course solving without writing candidates is preferred, but this is only easy/possible up to a certain point, some fiendish/extreme su dokus will require visible candidates, and many of the puzzles here on the net are way harder.

So when solving easy magazine/newspaper su dokus i never write down candidates, but they are usually boring to solve:(
Guru
 
Posts: 5
Joined: 25 October 2005


Return to Advanced solving techniques