I've received a lot of interesting comments and questions from Sudoku fans over the last few years and this page
is where I try to answer them. Please feel free to drop me a note on the side of the page or try the Facebook
comment box. Or you can email me directly at andrew@str8ts.com.
Post a Comment or Question here...
... by: Bruce K, Portland, OR
I was playing about in the Killer solver (which is most excellent!) with this puzzle. It got to this step: http://i.imgur.com/EKyYUQb.png And it just doesn't make sense. A 12-in-4 cage can be 1236 or 1245, but it says "only 1/2/3/4/6 is possible". What happened to 5? 1245 seems perfectly possible in that situation. Is there something I'm missing here? Thanks. Andrew Stuart writes (19-Apr-2014):
That’s a good one. You are correct that 1245 will fit a 4-cage with 12 as the total. But There is no 4 available in G1 or G7 - So it would have to occupy G5 or G6. Then if you wanted the other cell in G5/G6 to be a 5 you couldn't place the 1 and 2 in G1 and G7. Won't fit. So there is no combination which will permit a 5 in G5/G6.
As well as considering all the cage combinations the solver prunes the candidates according to those that can fit.
... by: Bernd Mielke, Germany
Hi Andrew,
many thanks for the puzzles! I like the Killer Sudoku most, and I like to select the few strategies which I allow the solver to apply on my behalf.
With the version 1.99 in my (outdated...) Firefox 3.03 deselecting the tough, diabolical and extreme strategies works opposite as in the past: unmarking the group toggles all settings there. The group selector is tickes, and the elements of the group are deselected (or selected when they were before ...). Probably related to the new option with the cookies which I value - it saves me to de-select the same strategies every time.
If possible, I'd like to unselect also the Pointing Pairs and Box/Line Reduction strategies because I want to train my brain to find them by myself ...
Hope this helps to improve the solver further!
Cheers, Bernd..
Andrew Stuart writes (19-Apr-2014):
I don’t have Firefox 3 (I've on v28 ;) so I cant test the issue. But the right side tick boxes will toggle the left hand groups - I'm sure they did before. However, left hand ticks won't e saved in the cookie until you press "take step". Something I need to change next update. If you don’t get the behaviour I might ask for a screen shot or two, see if I can debug it.
I can't offer uncheck boxes for point pairs and LBR - this is because some of the later strategies will produce illogical conclusions (sometimes) if candidates haven't been cleared out using the basic strategies. However, if you want to print off a puzzle with the candidates cleared all you need to do is "take step" for just the first few steps.
Hope that helps
... by: Charles, Spain / UK
There is a problem with Sudoku Solver - the page seems to have become corrupted
Andrew Stuart writes (14-Apr-2014):
Seems ok to me. Any change today?
... by: Meg, Florida
Andrew:
Thanks to you I have a new strategy in my arsenal!! Simple coloring solved the puzzle I was stuck on. And I know there are even more strategies to learn on your site, so THANKS!!
But here's my problem. I solve on my iPad. There's no Sudoku app that I know of that will let me color a cell individually. Your new app isn't for iPad. I'll try the 5 daily puzzles, but I wanted to ask you if you knew of an app that would let me somehow mark a cell. If I'm running just one number, I don't have to mark the little number, just the cell.
After so many years it's great to have a new way to get unstuck!
Meg
Andrew Stuart writes (14-Apr-2014):
So glad you like the site, many thanks for the kind words.
True, I would like to see a solver that allows full coloring. I've started that on the website solver but its automatic rather than manual, and only does bi-location links. I am looking at adding something more sophisticated for the SudokuWiki solver app. Its would be very difficult to do for the cell phone version but possible on an Android tablet. Also a good idea to add cell marking. You can currently highlight any sets of numbers of you click on a cell with those numbers in it. But a pallet of numbers you can turn on/off would be nice as well
... by: Pat, Indiana
Puzzle 2294 has 2 solutions and is therefore not unique.
Andrew Stuart writes (14-Apr-2014):
I'd love to see both solutions. Never had one with a double solution in all these years, but I can help you detect the problem if I can see them
... by: Cathy, Edmonton, Alberta
Hi, how do I change the level to EASY - I am just starting and I like the Jigsaw idea but I don't want to start at 5 of 6 stars...
Thank You,
Cathy
Andrew Stuart writes (4-Mar-2014):
Talking about http://www.sudokuwiki.org/Daily_Jigsaw_Sudoku? The 30 day archive has puzzles on Monday which are the easiest. Also, there is an easy jigsaw on http://www.sudokuwiki.org/5_Daily_Puzzles_for_the_iPadevery day
... by: Nat Sothanaphan, Thailand
Load Sudoku: CLICK TO LOAD
Hi,
I'm doing a project on Sudoku solver myself and I found your site a very great resource! However I think I found a bug in your program that you should know.
I got this Sudoku from http://norvig.com/sudoku.html. It has no solution. Your solver cannot solve it (or prove that there is no solution), but it deduces a naked pair 5,6 in H5, J5. However, when I plug in both possibilities, it can quickly determine that there is contradiction.
So I stared at the puzzle to see if there is a logical deduction without guessing. And I found it - row 8 actually has no cell with number 1 as a candidate! I think your solver should be able to detect this right away, but you probably missed it because the focus of the solver is not on the "no solution" puzzle. However this should be nice and easy to add :)
Andrew Stuart writes (3-Mar-2014):
That is an interesting situation in that puzzle. I do tack on checks for validity now and again when I see a hole in the validation. Looks like another hole. I can see that 1/5/6 want to fit in only two cells. Any puzzle that doesn’t have exactly one solution will not produce a credible solution path on my solver, for logical reasons.
... by: Michael, UK
I use your killer solver on my iPad. It is brilliant! - however one small complaint - the number font when I enter the grid is too large to see in the cells - for example, if I enter 12 in a group of boxes this just displays as 1. Is there anyway of reducing the font size so that the totals display in full?
Andrew Stuart writes (7-Mar-2014):
Ok fixed it. iPad was defaulting to quite a large padding, set that to zero. Made the font smaller. Also, I've changed the input box type to number so the right keyboard popups up on the ipad
... by: Dan, UK
hi there :)
the solver is perfect, also handling of solver. but I miss one (for me important) thing: a simple reset button for resetting puzzle to 'first given', mostly defined by url. because if I am doing tests of techniques, or want to start over in same puzzle, I have to refresh page. a simple 'reset puzzle' button, which wuld do a reset of puzzle would be very nice :)
thankx
Dan
Andrew Stuart writes (18-Feb-2014):
The solver is designed to tackle the puzzle at any given stage without memorizing any previous steps. Such memory retention would be extremely complex to implement on the server side with so many different users and on the client side when you could potentially have many solvers open. But in fact is doesn't need a memory because each state contains all the information necessary to proceed to the next step. Also, in Sudoku, it doesn’t need to distinguish between cells that are clues and those which are previously solved cells. I do allow this distinction in the small board with "enter clues or solutions" which gives you the two colors for large numbers. But as far as solving that is cosmetic.
So given that, a reset button is impossible to implement because there is nothing in the puzzle itself that says "this is the start". That is equivalent to asking "is this puzzle minimal" in the sense that all the clues are necessary. But some published puzzles are easier because they have extra clues and the minimal requirement is not a requirement for publishing.
But, we can brush all that aside by having the user decide the puzzle is at the start. So what you need is the "Save" and "Re-load" buttons which are above the small board. That does what you ask, but you need to remember to hit "save". I suppose if the puzzle information was in the URL a "restart" would be possible but I'd have to hide the button if the URL wasn't being used. I'll try that out.
... by: Wayne, US
2/15-21/14 Weekly 'Unsolvable' : Multiple Nested Dynamic Forcing Chains...nice!
You don't normally go this far out...did you unintentionally omit a solved value?
My trial and error algorithm takes a full 30 seconds to solve (Escargot "only" takes 9 seconds).
Andrew Stuart writes (18-Feb-2014):
David Filmer has kindly provided me with some of his puzzles that he's created. I'm finding it a very long process to get decent unsolvables now and he's been working on very hard puzzles for some time with a lot of success. Check the name under the page title where it says "guest compiler" so you know it's one of his.
|
