View previous topic :: View next topic |
Author |
Message |
| Simes
| Joined: 08 Apr 2005 | Posts: 71 | : | Location: North Yorkshire, UK | Items |
|
Posted: Tue Jul 05, 2005 5:33 pm Post subject: SadMan Software: Sudoku |
|
|
For all flavours of 32-bit Windows.
* Can solve puzzles completely automatically if you wish, and copy the result to the clipboard ready for pasting into those competition entry emails.
* It solves using exactly the same techniques that a human would. Only if this doesn't result in a solution does it resort to brute-force trial-and-error, and you can selectively disable this, or any other, solving technique.
* When solving automatically, it can create a log that shows the reason for entering each number in each cell, and the logic that allows for elimination of cell candidates.
* If a puzzle has multiple solutions, it can determine how many there are, and then show them to you.
* Can show the allowed entry "pencil-marks" for each cell.
* It can automatically update the pencil-marks as you enter numbers. This removes the tedium, but leaves the fun!
* Can prevent and/or highlight invalid entries.
* Can show the limitations imposed by the numbers already entered.
* When pasting from the clipboard, it will accept a wide variety of styles. This is extremely useful if you want to copy from discussion forums, websites etc.
* It can print the puzzle, optionally including pencil-marks, enabling you to solve away from your computer if you prefer.
* You can associate sudoku puzzle files, so that double-clicking one will automatically open it in Sudoku
* You can drag Sudoku puzzle files from Windows Explorer and drop them onto Sudoku.
* Small and extremely simple to use.
And, dare I say it, as used by Pappocom when he needs to determine the number of solutions for a puzzle.
Available for download from SadMan Software _________________ Simes
www.sadmansoftware.com/sudoku
Last edited by Simes on Sun Oct 12, 2008 11:58 am; edited 1 time in total |
|
Back to top |
|
|
| barleyfarmer
| Joined: 12 Jul 2005 | Posts: 3 | : | | Items |
|
Posted: Wed Jul 13, 2005 1:05 am Post subject: Hat's off to you |
|
|
A good all-round app, does all the things I want. I wish I had more time to carry on developing my own. But it's Summer and I want out of the office/house!
Steve |
|
Back to top |
|
|
| Simes
| Joined: 08 Apr 2005 | Posts: 71 | : | Location: North Yorkshire, UK | Items |
|
Posted: Thu Jul 28, 2005 4:47 am Post subject: and now it can create puzzles as well as solve them |
|
|
As of version 2.0, released today:
* Added the ability to create new puzzles based on the required solving techniques or degree of difficulty.
* Added the ability to rate a puzzle.
* Added the ability to rotate or reflect the entire puzzle, or substitute numbers.
Feedback welcome. _________________ Simes
www.sadmansoftware.com/sudoku |
|
Back to top |
|
|
| rfb
| Joined: 09 Apr 2005 | Posts: 26 | : | | Items |
|
Posted: Thu Jul 28, 2005 8:43 pm Post subject: What is a 'hard' puzzle |
|
|
I dowloaded the new version today and it looks good.
May I suggest you post the current version on the download page so that it is easy to tell when a newer version is available.
How does the puzzle grading work?
I tried generating a hard puzzle which took a long time (the counter went past 80) before it produced a puzzle.
When I tried to solve it I got stuck so I tried the solver and it required the 'Forcing Chains' technique to solve.
I was surprised to find such an advanced technique required - what on earth do you put in Fiendish and Diabolicals!
Is there an explanation of what the grades mean
Code: |
+---+---+---+
|7..|.42|3.9|
|.1.|7..|...|
|..4|..8|...|
+---+---+---+
|.2.|.3.|..5|
|14.|...|.23|
|8..|.2.|.7.|
+---+---+---+
|...|4..|1..|
|...|..5|.8.|
|5.8|29.|..4|
+---+---+---+
|
Bob _________________ Bob |
|
Back to top |
|
|
| Simes
| Joined: 08 Apr 2005 | Posts: 71 | : | Location: North Yorkshire, UK | Items |
|
Posted: Thu Jul 28, 2005 9:07 pm Post subject: |
|
|
rfb wrote: | I tried generating a hard puzzle ... when I tried to solve it I got stuck | Well that's the nature of hard!
I suspect the grading of puzzles is going to be tweaked for quite a while yet. It's difficult to be objective, as a puzzle that is hard when you first start Sudokus will probably seem a lot easier after a bit of practice. It's always going to be a subjective experience - some "easy" puzzles will seem hard and some "hard" puzzles will seem easy.
The grade is based on the techniques that are required to solve the puzzle - the more advanced the technique (IMHO), or the more times it is required, the higher the grade. Here's the current grading scheme:
Diabolical
Trial and Error (only if explicitly enabled)
or more than one of (Disjoint Subsets size 4 or Unique Subsets size 4 or swordfish)
or more than 4 of (XWing or ForcingChains)
Fiendish
any of Swordfish or Disjoint Subsets size 4 or Unique Subsets size 4
or more than 2 of (XWing or Forcing Chains)
Very Hard
more than 1 of (XWing or Forcing Chains)
or more than 1 of (Disjoint Subsets size 3 or Unique Subsets size 3)
Hard
any of XWing or Forcing Chains or Disjoint Subsets size 3 or Unique Subsets size 3
Moderate
(Block/Block Interactions or Block and Column/Row Interactions)
and (Disjoint Subsets size 2 or Unique Subsets size 2)
Mild
Disjoint Subsets size 2 or Unique Subsets size 2
Easy
Unique Candidate
Trivial
Sole Candidate
Does that make sense? This scheme won't agree with many published puzzles. They rarely get above moderate on the above scale.
Do you think it needs toning down a bit? _________________ Simes
www.sadmansoftware.com/sudoku |
|
Back to top |
|
|
| rfb
| Joined: 09 Apr 2005 | Posts: 26 | : | | Items |
|
Posted: Thu Jul 28, 2005 9:30 pm Post subject: |
|
|
I suppose I expected 'Forcing chains' to be considered a Fiendish level technique. I will try generating by Solving Technique until I ready to face 'Forcing chains'. At times I can't spot a simple pair staring me in the face
On the subject of Forcing chains I know you consider both values of a cell where there are two candidates but do you consider cases where a candidate number occurs twice in a row/column/box? _________________ Bob |
|
Back to top |
|
|
| Simes
| Joined: 08 Apr 2005 | Posts: 71 | : | Location: North Yorkshire, UK | Items |
|
Posted: Fri Jul 29, 2005 6:10 am Post subject: |
|
|
rfb wrote: | On the subject of Forcing chains I know you consider both values of a cell where there are two candidates but do you consider cases where a candidate number occurs twice in a row/column/box? | No, it only looks at cells with two candidates, and tries to form chains of them. _________________ Simes
www.sadmansoftware.com/sudoku |
|
Back to top |
|
|
| chuckfresno
| Joined: 16 Jun 2005 | Posts: 39 | : | | Items |
|
Posted: Sat Jul 30, 2005 12:03 am Post subject: |
|
|
It can take a long time to generate a puzzle, even easyier ones. Until and unless the generation routine can be accelerated, maybe one could select a range of levels at once. Or generate puzzles directly to disk files, automatically naming them according to level and/or tactics required. Let it run for a few hours, you've got all you need. I think I just solved a diabolical faster than it generated one! By comparision, "unreasonables" are created in a fraction of a second by the open source SOLO (http://www.chiark.greenend.org.uk/~sgtatham/puzzles/). I just put two of those into your program -- it rated one as "fiendish" and the other "diabolical"; the first requireing forcing chains, the second T&E to solve. So they *can* be generated much, much faster.
Simes wrote: |
I suspect the grading of puzzles is going to be tweaked for quite a while yet. It's difficult to be objective, as a puzzle that is hard when you first start Sudokus will probably seem a lot easier after a bit of practice. It's always going to be a subjective experience - some "easy" puzzles will seem hard and some "hard" puzzles will seem easy.
The grade is based on the techniques that are required to solve the puzzle - the more advanced the technique (IMHO), or the more times it is required, the higher the grade. Here's the current grading scheme:
...
Does that make sense? This scheme won't agree with many published puzzles. They rarely get above moderate on the above scale.
...
Do you think it needs toning down a bit?
|
I'm with you all the way. Some of the Japanese magazines give their puzzles a dual rating -- 1 to 5 on difficulty and the number of minutes a hypthetical solver would take. Obviously an easy 25x25 will take longer than a hard 6x6. As far as your ratings not matching up with anyone else's -- I don't think you should even try. It isn't possible anyway. As long as your ratings are internally consistant, the users will be able to make her own judgement and comparisons. I'm for as many ratings as you can justify without slowing down puzzle generation.
A short forcing chain can be child's play -- a long one can be mind bending. Most people until have ignored this and lumped them all together.
Also, the solution log could be a lot more verbose on request. Something like: "(4,1) = 9 : Forced from either candidate of (7,1)" could be spelled out with complete implication chains. |
|
Back to top |
|
|
| nj3h
| Joined: 29 Jul 2005 | Posts: 9 | : | Location: Virginia - USA | Items |
|
Posted: Mon Aug 01, 2005 2:05 am Post subject: |
|
|
Greetings,
Can the program key be entered without having to be on-line? I have a laptop that I like to put my games on that I never hook up to the internet. Of course, I have no problem getting an email on my desktop computer and transferring the key over to the laptop. Some programs require the user to be on the internet when activating with the purchased key. I was just wondering if this was one such program.
Thanks,
nj3h |
|
Back to top |
|
|
| Simes
| Joined: 08 Apr 2005 | Posts: 71 | : | Location: North Yorkshire, UK | Items |
|
Posted: Mon Aug 01, 2005 6:33 am Post subject: |
|
|
No problem. The registration key is delivered via email, and there are no on-line checks. _________________ Simes
www.sadmansoftware.com/sudoku |
|
Back to top |
|
|
| Simes
| Joined: 08 Apr 2005 | Posts: 71 | : | Location: North Yorkshire, UK | Items |
|
Posted: Tue Aug 02, 2005 7:21 pm Post subject: V2.1 is a bit faster |
|
|
People - thanks for the feedback.
I've just released V2.1 - this has some speed improvement when generating puzzles, but still needs to be a lot faster.
chuckfresno - I looked at the open source generator you mentioned, and we're both generating using very similar techniques (so why is mine so much slower then? ). Simon Tatham makes a very good point though - the more levels you have, the harder it is to find a puzzle that fits into one of them. So that's an argument for having fewer, broader difficulty categories. Listing the forcing chains is on the to-do list btw. _________________ Simes
www.sadmansoftware.com/sudoku |
|
Back to top |
|
|
| MickJ27
| Joined: 07 Aug 2005 | Posts: 1 | : | | Items |
|
Posted: Sun Aug 07, 2005 4:19 pm Post subject: Sadman Sudoku v2.2 |
|
|
Just downloaded and registered. Excellent program Simes, thanks.
I almost posted a message sayign that I'd found a fault with a forcing chain. Actually I was wrong (of course ) but it made me realise that sometimes following the chains actually leads you to one of the candidates being impossible. Does the program use this as a solving technique ? |
|
Back to top |
|
|
| Simes
| Joined: 08 Apr 2005 | Posts: 71 | : | Location: North Yorkshire, UK | Items |
|
Posted: Sun Aug 07, 2005 6:35 pm Post subject: |
|
|
Hi Mick,
MickJ27 wrote: | Excellent program Simes, thanks. |
Glad you like it.
MickJ27 wrote: | ...but it made me realise that sometimes following the chains actually leads you to one of the candidates being impossible. Does the program use this as a solving technique ? |
Unless the puzzle has more than one solution, following the chain from one of the candidates will always eventually lead to a dead end, but no, the forcing chain technique doesn't make use of that. I'd regard that more as trial and error - something I personally want to avoid unless all else fails. _________________ Simes
www.sadmansoftware.com/sudoku |
|
Back to top |
|
|
| rfb
| Joined: 09 Apr 2005 | Posts: 26 | : | | Items |
|
Posted: Tue Aug 23, 2005 10:58 pm Post subject: |
|
|
Life has been quiet in this thread recently.
I was pleased to see you had released Version 2.2 and that it supported colouring. Unfortunately though this is only as a computer solving technique - not as a technique that a user can employ.
Having used it in Angus J's program it is a very useful technique and I wouldn't want to attempt to manually solve a puzzle requiring colouring or forcing chains without it.
While I still prefer the interface of your program Angus has improved his program so much over the past few weeks that there isn't much to choose between them.
Although you are improving generation times Angus seems to generate new puzzles almost instaneously.
Another useful feature of his program is the ability to select multiple cells by holding ctrl whilst clicking on cells - this is useful when removing candidates from multiple cells after identifying pairs etc.
A very minor improvement would be to have the Puzzle generation, solving techniques and print options menus as additional tabs in the main options menu rather than as separate menus (though you could retain the direct links in the pulldown menus)
Another minor change would be to assume that pasted numbers are clues rather than having to use 'Change all numbers to clues'
Bob _________________ Bob |
|
Back to top |
|
|
| Simes
| Joined: 08 Apr 2005 | Posts: 71 | : | Location: North Yorkshire, UK | Items |
|
Posted: Wed Aug 24, 2005 7:12 pm Post subject: |
|
|
Hi Bob,
Thanks for the feedback.
rfb wrote: | I was pleased to see you had released Version 2.2 and that it supported colouring. Unfortunately though this is only as a computer solving technique - not as a technique that a user can employ. |
Good point. I've just taken a peak at SS, and manual colouring could be a useful aid, so I'll look at adding some colouring options.
rfb wrote: | Although you are improving generation times Angus seems to generate new puzzles almost instaneously. | Yes, I wonder what the trick is? I think I must be missing something fundamental in the generation technique.
rfb wrote: | Another useful feature of his program is the ability to select multiple cells by holding ctrl whilst clicking on cells - this is useful when removing candidates from multiple cells after identifying pairs etc. | I'll look into this one too.
rfb wrote: | A very minor improvement would be to have the Puzzle generation, solving techniques and print options menus as additional tabs in the main options menu rather than as separate menus (though you could retain the direct links in the pulldown menus) | Hmmm. I think you may be right.
rfb wrote: | Another minor change would be to assume that pasted numbers are clues rather than having to use 'Change all numbers to clues' | I think I'll probably do this too, but I suspect I'll then get requests to change it back. _________________ Simes
www.sadmansoftware.com/sudoku |
|
Back to top |
|
|
|