## Help with my first very hard please.

Advanced methods and approaches for solving Sudoku puzzles
hi paulf ,thanks for your interest, i have already done the cross matching up and down the columns as well as the cross referencing block by block and still end up with all theses darn candidates in multple cells all over the darn puzzle - sorry i am not asking for help to solve it directly , if i make myself clear, but a hint as to how to look at a block/row/column of cells with multiple candidates and get to learn a technique to reduce them
, once i can do that , and can arrive at a cell(s) with only a single candidate, i can make my own way if you get what i mean? thnks
clara5218

Posts: 11
Joined: 06 June 2005

Clara,

Its very difficult without seeing the puzzle. There are certainly no 'groups' in the three rows you have given us. Here would be my summary of the advanced strategies to think through.

Groups: as discussed previously in this forum, within any row, any column or any 3x3 block, looking for either a group of (say) three cells that only have some combination of three possible answers; or (say) three numbers that only fit (in some combination) into three cells. Groups of two are very common and usually easier to spot if you have any. Given any group you can eliminate other possibilities from your row/column/block as per the discussion above.

Restricted cells: This usually applies to a number that can only fit within one column or row within a particular 3x3 block. This means that this number cannot fit within this particular column or row in the neighbouring 3x3 blocks.

X-Wings, Swordfish etc: There are some more advanced strategies which I will be surprised you need to implement. X-wings are the most useful. This involves finding two rows that only have the same two 'column options' for a particular number e.g. in row one, 4 can only go in column three and seven and in row five, 4 can also only go in column three and seven. As there are only two options within both of these rows, if you place one of the numbers (say the 4 in R1C3) then that will automatically eliminate the possibility in the corresponding row (R5C3 in our example). This means that the number must go in the 'diagonal' cell (R5C7 becomes a 4 in the example). The number must go in both cells of one of the two diagonal pairs. This means that other cells in that column cannot possibly have that number (4) in them, thus eliminating some possibilities for you. Clearly, X wings also work if you find two columns that have the same two 'row options'. You can find examples with probably a better explanation here: http://www.simes.clara.co.uk/programs/sudokutechnique6.htm

Swordfish are an extension of X wings into three columns and three rows. I've done a lot of hard puzzles and never had to use one yet, even some of the puzzles listed on the swordfish page: http://www.simes.clara.co.uk/programs/sudokutechnique7.htm

Hope that helps,

Andrew
abailes

Posts: 27
Joined: 02 June 2005

hi abailes,you make me feel a little better, so there were no groups in the rows that i posted? i was busy going cross eyed trying to find some -ok here is the whole grid with my pencil marks in and hard numbers in brackets:

row 1 - 269-2679-679-(1)-(3)-2467-678-46789-(5)

row 2 - 13569-(4)-1679-5678-578-567-(2)-36789-13789

row 3 - (8)-123567-1267-(9)-47-24567-1367-3467-1347

row 4 - 146-1678-1478-23478-(5)-1247-(9)-23678-12578

row 5 - 1569-15679-(2)-378-789-179-(4)-3678-1378

row 6 - 1459-15789-(3)-2478-(6)-1247-18-278-1278

row 7 - 249-289-489-457-(1)-(3)-2578-245789-(6)

row 8 - 23469-23689-(5)-467-479-4679-2378-(1)-234789

row 9 - (7)-1369-1469-456-(2)-(8)-35-3459-349

ok i can see a group of three in row 8 ( 239 ) but ive no clue what to do with it? any tips would be appreciated
clara5218

Posts: 11
Joined: 06 June 2005

so sorry bailes, my original post was incorrect in that i had my row numbers wrong - i dont think it was relevant as there were no groups in any of the rows anyway - but this last post has all the rows correct! my apologies for any inconvenience caused - clara8
clara5218

Posts: 11
Joined: 06 June 2005

Clara,

Sorry, 239 is not a group in row 8. Sadly it does not either fit the rule of being the only three numbers in three cells (there are no cells with only some combination of 2, 3 or 9 in), nor does it fit the rule that there are only three cells with 2, 3 and 9 in (there are six cells with 2, 3 or 9 in).

At first glance I can't see any other groups in the rows.

Anyway, I will reconstruct the grid using the hard numbers you have given and see where we get to. There may be some groups in the columns or blocks or other possibilities. Give me a little while.

Cheers,

Andrew
abailes

Posts: 27
Joined: 02 June 2005

OK.

Here are the steps I used beyond the normal four strategies of only place for number in row, column and block and the only number left in a particular cell. There may be simpler ways and I'm not going to tell you how many additional numbers you need to add between each of the steps (by the normal four strategies). That way you still have a puzzle left.

1) First, there is a 3,5 group in the Top-Left 3x3 block
2) Then, focus on the 1s in the Top-Left 3x3 block and their effect on the placement of the 1 in the Bottom-Left 3x3 block
3) Then, look at the placement of the 6s in the Centre-Right 3x3 block and its effect on the 6s in the Top-Right Block.
4) Then find the X-wing in the 6s

And that's all you need to do

Andrew
abailes

Posts: 27
Joined: 02 June 2005

hey andrew ,thanks a lot for your interest and effort to assist me
, i got this puzzle from the list of xwing examples on the website for solving techniques for soduko puzzles - its listed as - xwing.1.sdk - there are a few examples given so i just took the first one - i am just a little worried that i might have confused you and put in a candidate (s) that should not be there - but the original puzzle is as per the example given on the web site - wow!!! no wonder im battling if it gives you guys trouble!
i dont see any wxing either but thats to be expected because right now i am so confused i couldnt see my face in the mirror lol! cheers
clara5218

Posts: 11
Joined: 06 June 2005

I thought I recognised the puzzle! I've done that one before... The great thing about Sudokus is that you cannot possibly remember the placements of particular numbers, so after a while doing old puzzles is no less of a joy.

By the way, you said in your first post that you had just progressed from v. easy to easy. The x-wing puzzles are about as hard as it gets, so I'm worried that you may have gone to the other extreme... Don't let the complexity of this one put you off Sudoku - if I were you (of course, feel free to ignore me at this point), I would focus on puzzles that are either Difficult or Fiendish in the Times. Even Fiendish is easier than this puzzle. You may find those more 'doable', less frustrating and certainly more enjoyable.

Cheers,

Andrew
abailes

Posts: 27
Joined: 02 June 2005

ok andrew will do, but bear in mind that i am in africa so the times isnt readily available - however we do have a sudoku running in our local paper and yes !!! they are rated "fiendish" thats why i left them strictly alone - wow, now i am gonna get stuck into them and see just how i go for a month or so ! thanks very much for your patience and assistance you are a star!!!! cheers clara8
clara5218

Posts: 11
Joined: 06 June 2005

Try also this:

http://www.timesonline.co.uk/section/0,,18209,00.html

That gives you the last week's worth to get started
abailes

Posts: 27
Joined: 02 June 2005

Firstly, a belated thanks to abailes et al for the explainations of this. I think I now have it, as well as the technique I was looking for to handle these.

Given the puzzle at the start of this topic, am I right in thinking that column 1 (which equates to the following as far as I can tell):

{8} - 267 - 2456 - 127 - 1267 - 126 - 145679 - {3} - 4679

also has a similar grouping in rows 2, 4, 5 and 6 with the numbers 1, 2, 6 and 7? This allows the possible values the other rows to become

{8} - 267 - 45 - 127 - 1267 - 126 - 459 - {3} - 49

(My technique also throws up some spurious 'matches', especially in there is a cell with all of the possible values of all the other unfilled cells in the row/col/block but the end effect is harmless).

Cheers,

Don