The thread for discussing: How To Make Clueless Explosions.
The thread for discussing: How To Make Clueless Explosions.
Thanks to a lot of extremely good and valuable information and input from Glyn, I have hopefully found a way for us to make our own Clueless Explosions. But it is only doable as a group collaboration. I have of course tried to make a complete Clueless Explosions myself today, but after 2 hours of hard work I only had 3 out 9 grids completed. Here is my suggestion:
Using SudoCue, SadMan or some other software for generating standard 9x9 Sudokus, I will generate a Sudoku and have it solved. I will then manually enter the complete solution (81 cells) into the 10th merged grid of the clueless grid in a TUSP worksheet for Clueless Explosions. When entering into the 10th merged grid, the clues are automatically duplicated to their twin cells in the nine 9x9 Sudokus.
I then post the codes for the 9 Sudokus, each with 9 clues here in a thread, see “Substitute Clueless Explosion number 1 (SCE #1)” for an example. Now several people, for instance 9 different ones, each grab the code for a single 9x9 Sudoku and using whatever means at their disposal create a solvable 9x9 Sudoku having EXACTLY 1 SOLUTION, AND INCLUDING the 9 original clues as given. Then each post their code string here or send it to me via PM or email.
Then I can paste the 9 code strings directly into TUSP, erase all the 81 clues in the clueless (blue) cells and (try to) solve the Clueless Explosion and see if I can find more than 1 solution. A DLX based brute force solver for making sure that there is exact one solution would be extremely helpful (J-C are you reading?).
If everything looks OK, I will then post the Clueless Explosion code and a paper version, which can be downloaded.
Does this sound like plan?
Using SudoCue, SadMan or some other software for generating standard 9x9 Sudokus, I will generate a Sudoku and have it solved. I will then manually enter the complete solution (81 cells) into the 10th merged grid of the clueless grid in a TUSP worksheet for Clueless Explosions. When entering into the 10th merged grid, the clues are automatically duplicated to their twin cells in the nine 9x9 Sudokus.
I then post the codes for the 9 Sudokus, each with 9 clues here in a thread, see “Substitute Clueless Explosion number 1 (SCE #1)” for an example. Now several people, for instance 9 different ones, each grab the code for a single 9x9 Sudoku and using whatever means at their disposal create a solvable 9x9 Sudoku having EXACTLY 1 SOLUTION, AND INCLUDING the 9 original clues as given. Then each post their code string here or send it to me via PM or email.
Then I can paste the 9 code strings directly into TUSP, erase all the 81 clues in the clueless (blue) cells and (try to) solve the Clueless Explosion and see if I can find more than 1 solution. A DLX based brute force solver for making sure that there is exact one solution would be extremely helpful (J-C are you reading?).
If everything looks OK, I will then post the Clueless Explosion code and a paper version, which can be downloaded.
Does this sound like plan?
Last edited by Børge on Fri Mar 14, 2008 5:59 pm, edited 1 time in total.
A minor correction. The uniqueness of the 9 outer sudokus must be reliant on a subset of the solution values in the Clueless grid, perhaps 2 -3 should be required as givens. Also using those values alone in the the Clueless grid, it should not be immediately solvable or the puzzle will be trivial. Several passes are required as some must be inferred from the implications of the other outer grids.
We really need DLX to get grids that are not only unique but to find the one which is in a subset that is not so convoluted as to be to impracticable to solve or too trivial. It shouldn't necessarily require a dedicated DLX solver, a master program running 10 standard ones running in parallel and sharing the overlapping data should suffice.
"I have a truly marvelous proof of this proposition which this margin is too narrow to contain."
We really need DLX to get grids that are not only unique but to find the one which is in a subset that is not so convoluted as to be to impracticable to solve or too trivial. It shouldn't necessarily require a dedicated DLX solver, a master program running 10 standard ones running in parallel and sharing the overlapping data should suffice.
"I have a truly marvelous proof of this proposition which this margin is too narrow to contain."
I have 81 brain cells left, I think.
Following my own suggestion to the letter I produced my first hopefully solvable Clueless Explosion with hopefully one solution only. It took me approx. 5 hours. I am in the process of solving it, and I am having far more problems than with the ones Ruud makes.
For the code and a downloadable paper version, please see the upcoming thread “SCE #2. A hopefully real Clueless Explosion”.
For the code and a downloadable paper version, please see the upcoming thread “SCE #2. A hopefully real Clueless Explosion”.
-
- Addict
- Posts: 92
- Joined: Mon Apr 23, 2007 12:50 pm
- Location: Belgium
Re: A suggestion for making our own Clueless Explosions.
Here Ruud explains how he's generating Clueless Could be outdated, thoughtBørge wrote:Thanks to a lot of extremely good and valuable information and input from Glyn, I have hopefully found a way for us to make our own Clueless Explosions. But it is only doable as a group collaboration. I have of course tried to make a complete Clueless Explosions myself today, but after 2 hours of hard work I only had 3 out 9 grids completed. Here is my suggestion:
Using SudoCue, SadMan or some other software for generating standard 9x9 Sudokus, I will generate a Sudoku and have it solved. I will then manually enter the complete solution (81 cells) into the 10th merged grid of the clueless grid in a TUSP worksheet for Clueless Explosions. When entering into the 10th merged grid, the clues are automatically duplicated to their twin cells in the nine 9x9 Sudokus.
I then post the codes for the 9 Sudokus, each with 9 clues here in a thread, see “Substitute Clueless Explosion number 1 (SCE #1)” for an example. Now several people, for instance 9 different ones, each grab the code for a single 9x9 Sudoku and using whatever means at their disposal create a solvable 9x9 Sudoku having EXACTLY 1 SOLUTION, AND INCLUDING the 9 original clues as given. Then each post their code string here or send it to me via PM or email.
Then I can paste the 9 code strings directly into TUSP, erase all the 81 clues in the clueless (blue) cells and (try to) solve the Clueless Explosion and see if I can find more than 1 solution. A DLX based brute force solver for making sure that there is exact one solution would be extremely helpful (J-C are you reading?).
If everything looks OK, I will then post the Clueless Explosion code and a paper version, which can be downloaded.
Does this sound like plan?
As for a DLX solver, I would suggest first to read various threads @ programmers' forum. Here are some links
Once you have some DLX working for vanilla sudoku, it's easy to adapt for Clueless or other overlapping variants.
Hope this helps
Re: A suggestion for making our own Clueless Explosions.
Jean-Christophe: Thanks for all the info.
One of the many inputs I got from Glyn. Even so, thanks a lot.Jean-Christophe wrote:Here Ruud explains how he's generating Clueless Could be outdated, thought
I need some help regarding info on other solvers than SadMan. SadMan is the only solver that I’m pretty familiar with. SudoCue is IMHO by far the best solver around, but I am not on the friendliest of terms with its GUI (Graphical User Interface)
First I’ll need to explain the first two steps I do to create a Clueless Explosion.
STEP1:
Take the complete valid solution for a vanilla 9x9 Sudoku, or generate a Sudoku and solve it using your favourite solver. The difficulty rating is of no relevance. This Sudoku becomes the 10th grid, i.e. the clueless one.
Here an example:
STEP 2:
For each of the nine blocks in the Sudoku generated in step 1, I repeat the following:
Enter the block into SadMan as the clueless (blue) cells for one of the nine grids. Then I ask SadMan to solve this Sudoku. Since it has only 9 givens it likely has billions of solutions but SadMan gives me one of them. Which one I don’t care.
Here an example for block 1:
What I need to know is:
Can any freeware solvers (SudoCue, JSudoku, SudokuSolver, Simple Sudoku, Sudoku Susser, Chisai, QSudoku, Yudoku, etc.) do this, i.e. based on only 9 clues produce one of its solutions or create a puzzle with the 9 clues intact and then solve it. If yes, which ones and with which settings. I do not have the time to install and familiarise me with a bunch of solvers. So everybody, please help out, or you will not get further Clueless Explosions from me.
STEP 3:
After step 2 I have a complete solved Clueless Explosion. Now the trick is to generate a puzzle for this solved Clueless Explosion that fulfils the following two criteria:
1) Is solvable and has one solution only.
2) Is not too simple nor too difficult.
This step is of course the crucial one because it determines if the puzzle is solvable and if it is too simple or too difficult. I am currently working on improving this step.
STEP 4:
Enter the puzzle created in step 3 into SudokuSolver (with only a limited number of the solver capabilities turned on) and see if the puzzle is solvable and has one solution only. If yes, then try to solve it manually, to determine its difficulty rating.
First I’ll need to explain the first two steps I do to create a Clueless Explosion.
STEP1:
Take the complete valid solution for a vanilla 9x9 Sudoku, or generate a Sudoku and solve it using your favourite solver. The difficulty rating is of no relevance. This Sudoku becomes the 10th grid, i.e. the clueless one.
Here an example:
STEP 2:
For each of the nine blocks in the Sudoku generated in step 1, I repeat the following:
Enter the block into SadMan as the clueless (blue) cells for one of the nine grids. Then I ask SadMan to solve this Sudoku. Since it has only 9 givens it likely has billions of solutions but SadMan gives me one of them. Which one I don’t care.
Here an example for block 1:
What I need to know is:
Can any freeware solvers (SudoCue, JSudoku, SudokuSolver, Simple Sudoku, Sudoku Susser, Chisai, QSudoku, Yudoku, etc.) do this, i.e. based on only 9 clues produce one of its solutions or create a puzzle with the 9 clues intact and then solve it. If yes, which ones and with which settings. I do not have the time to install and familiarise me with a bunch of solvers. So everybody, please help out, or you will not get further Clueless Explosions from me.
STEP 3:
After step 2 I have a complete solved Clueless Explosion. Now the trick is to generate a puzzle for this solved Clueless Explosion that fulfils the following two criteria:
1) Is solvable and has one solution only.
2) Is not too simple nor too difficult.
This step is of course the crucial one because it determines if the puzzle is solvable and if it is too simple or too difficult. I am currently working on improving this step.
STEP 4:
Enter the puzzle created in step 3 into SudokuSolver (with only a limited number of the solver capabilities turned on) and see if the puzzle is solvable and has one solution only. If yes, then try to solve it manually, to determine its difficulty rating.
Last edited by Børge on Fri Mar 21, 2008 2:11 pm, edited 1 time in total.
I see that JC is turning these out on an assembly line now.
With regard to step 2, there is a possibility that Sudoku Architect written by Havard may be able to deal with this step, it is not officially released yet. the current version is called Sudoku Assistenten. Not much about it on the Programmers site, but some of its latest features have been added to Glenn Fowler's solver and generator programs, which is a pretty good seal of approval.
To find Glenns program (C source code is available) do a search for Sudoku and att.
For Sudoku Architect a search for posts by Havard on the Programmers site will give you an initial contact. By the way he is in Norway.
Hope this is useful.
With regard to step 2, there is a possibility that Sudoku Architect written by Havard may be able to deal with this step, it is not officially released yet. the current version is called Sudoku Assistenten. Not much about it on the Programmers site, but some of its latest features have been added to Glenn Fowler's solver and generator programs, which is a pretty good seal of approval.
To find Glenns program (C source code is available) do a search for Sudoku and att.
For Sudoku Architect a search for posts by Havard on the Programmers site will give you an initial contact. By the way he is in Norway.
Hope this is useful.
I have 81 brain cells left, I think.
Ruud's SuMoCue does what you want (I think).Børge wrote:What I need to know is:
Can any freeware solvers ..based on only 9 clues produce one of its solutions or create a puzzle with the 9 clues intact and then solve it.
1) open SuMoCue
2) click in a cell
3) put in number
4) after all the cells you want are filled go to "File...Save"
5) after save, F10 and it will give you one of more than 999 solutions
Hope this helps.
Ed
How to make clueless explosions
In response to the challenge posed by Glyn some days ago I have designed a program that creates clueless puzzles (both explosion and special). The program is based on principles similar (but in reverse) to those I applied last year when my one minute solving times created so much commotion (see thread "Clueless Explosion #42").
Unlike Ruud's game generator, which seems to be based on a few templates (nowadays I feel that I have played every new game before) my generator is fully random. One kink that had to be worked out was to guarantee that each generated game had only one and unique solution.
My only problem that I am working on right now is to create a clu-file that can be imported into Clueless Helper. The problem is the last line in the file which appears to be an encryption in hexadecimal of various check sums from the 9+1 9x9 Sudoku matrices. Until I have cracked the code (or get the key from Ruud) I will have to type every game into Clueless Helper.
The generator did in a short order (a few minutes) generate a few hundred games with various degrees of difficulty and I am now set for life with a new explosion every thursday regardless of what happens to this website.
Rgds,
Tomas
Unlike Ruud's game generator, which seems to be based on a few templates (nowadays I feel that I have played every new game before) my generator is fully random. One kink that had to be worked out was to guarantee that each generated game had only one and unique solution.
My only problem that I am working on right now is to create a clu-file that can be imported into Clueless Helper. The problem is the last line in the file which appears to be an encryption in hexadecimal of various check sums from the 9+1 9x9 Sudoku matrices. Until I have cracked the code (or get the key from Ruud) I will have to type every game into Clueless Helper.
The generator did in a short order (a few minutes) generate a few hundred games with various degrees of difficulty and I am now set for life with a new explosion every thursday regardless of what happens to this website.
Rgds,
Tomas
@ Jean-Christophe & TKollen:
As announced HERE I will post SCE #3 on Thursday 2008-03-20. I presume that one of you will post a CE on Thursday 2008-03-27 and then the other on Thursday 2008-04-03. IMHO it is a good idea to announce the CE a couple of days ahead.
@TKollen:
If you crack the .CLU format what about making a Sudoku-Code to .CLU converter. A simple DOS command line program should be sufficient, reading the Sudoku-Code file and outputting the .CLU file.
My method for generating a Clueless Explosion (CE) is fairly manual. The generation itself only requires SadMan and SudoCue. Instead of SadMan, SumoCue can probably be used, but I have not looked into that yet. For evaluating the result I use SudokuSolver. Generating a CE takes approx 20 minutes. Since it is so simple I will write a small guide on how to do it and post is here in this thread.
When I have published my guide and people see how simple it is, I hope that many others will feel inclined to make CE puzzles post them here.
As announced HERE I will post SCE #3 on Thursday 2008-03-20. I presume that one of you will post a CE on Thursday 2008-03-27 and then the other on Thursday 2008-04-03. IMHO it is a good idea to announce the CE a couple of days ahead.
@TKollen:
If you crack the .CLU format what about making a Sudoku-Code to .CLU converter. A simple DOS command line program should be sufficient, reading the Sudoku-Code file and outputting the .CLU file.
My method for generating a Clueless Explosion (CE) is fairly manual. The generation itself only requires SadMan and SudoCue. Instead of SadMan, SumoCue can probably be used, but I have not looked into that yet. For evaluating the result I use SudokuSolver. Generating a CE takes approx 20 minutes. Since it is so simple I will write a small guide on how to do it and post is here in this thread.
When I have published my guide and people see how simple it is, I hope that many others will feel inclined to make CE puzzles post them here.
How to make a paper copy of a Clueless puzzle
How to make a paper copy of a Clueless puzzle
When support was added for Samurai, a new post including the previous text of this post was added to the end of this thread.
When support was added for Samurai, a new post including the previous text of this post was added to the end of this thread.
Last edited by Børge on Thu Apr 17, 2008 7:10 pm, edited 1 time in total.
-
- Expert
- Posts: 143
- Joined: Wed Nov 15, 2006 1:45 pm
- Location: London
Hi Børge
If you want a printout of the full puzzle for a standard Clueless, Clueless Explosion and Samurai, you can load them into my program and use the print function to print the Composite Grid.
You an screen shot the print preview or - as I do - use a free PDF printer driver to "print" the puzzles to .pdf format. This saves having to copy and paste between multiple applications and paint packages.
As Mike guessed on another post, I'm extending support in SudokuSolver for these type of puzzles - including a revamp of the navigation and a full puzzle viewer.
Rgds
If you want a printout of the full puzzle for a standard Clueless, Clueless Explosion and Samurai, you can load them into my program and use the print function to print the Composite Grid.
You an screen shot the print preview or - as I do - use a free PDF printer driver to "print" the puzzles to .pdf format. This saves having to copy and paste between multiple applications and paint packages.
As Mike guessed on another post, I'm extending support in SudokuSolver for these type of puzzles - including a revamp of the navigation and a full puzzle viewer.
Rgds
Hi Richard,
What you are writing sounds great.
Personally I do not need printouts, but others seem to need them. I freely admit that using the print option in SudokuSolver did not occur to me . Well as a result there are now at least two methods to choose from for getting printouts of non-Ruud generated Clueless puzzles. [And mine is a bit more customizable ]
Rgds
Børge
What you are writing sounds great.
Personally I do not need printouts, but others seem to need them. I freely admit that using the print option in SudokuSolver did not occur to me . Well as a result there are now at least two methods to choose from for getting printouts of non-Ruud generated Clueless puzzles. [And mine is a bit more customizable ]
Rgds
Børge
In light of Nouggie’s finding that a .CLU file saved by the Clueless Helper on one PC cannot be read by a Clueless Helper running on another PC, I think it is reasonable to assume that the last line a .CLU file is not only a checksum of the rest of the file but also of some information identifying the PC it runs on (CPU, disk-drive, main-board serial number or something stored in the registry). If you search the registry for “Ruud” you’ll find at least two entries for the Clueless Helper. One directly in a folder with a (probably random generated) GUID (Globally Unique Identifier) name: http://en.wikipedia.org/wiki/GUID
On my PC the name of this folder is:
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Uninstall\{4B83D37C-0B99-4E71-B6DB-95F41510BD89}
I’ve spent some time looking into the matter without finding any clues to cracking the final line; including looking for a “hidden” way to import the Sudoku code for a clueless puzzle, since you can import Samurais, but I’ve had no success, and I’m not going to pursue this further. Only changing a single pencil mark, generates a complete different checksum (final line) in a .CLU file. IMHO the chances of figuring out Ruud’s algorithm are extremely slim.
One way around could be to write a proxy, fooling the Clueless Helper into thinking that it gets its data from SudoCue.net, but really gets them from the proxy. The proxy could either communicate with SudoCue.net, substituting the Sudoku code it receives from SudoCue.net with another code from a file or just simulate the complete communication session to the Clueless Helper. But who knows, Ruud could be using his own encrypted protocol.
On my PC the name of this folder is:
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Uninstall\{4B83D37C-0B99-4E71-B6DB-95F41510BD89}
I’ve spent some time looking into the matter without finding any clues to cracking the final line; including looking for a “hidden” way to import the Sudoku code for a clueless puzzle, since you can import Samurais, but I’ve had no success, and I’m not going to pursue this further. Only changing a single pencil mark, generates a complete different checksum (final line) in a .CLU file. IMHO the chances of figuring out Ruud’s algorithm are extremely slim.
One way around could be to write a proxy, fooling the Clueless Helper into thinking that it gets its data from SudoCue.net, but really gets them from the proxy. The proxy could either communicate with SudoCue.net, substituting the Sudoku code it receives from SudoCue.net with another code from a file or just simulate the complete communication session to the Clueless Helper. But who knows, Ruud could be using his own encrypted protocol.