The New Sudoku Players' Forum

[dyitto]

How are killer sudokus created?

What I'm trying now is:

- define some cages with undefined totals
- generate a grid such that no number occurs twice in such a cage
- compute the cage totals from the grid
- empty the grid
- see if the current cages and cage totals allow at most one solution

There are always more solutions. Now my program is running a scenario and still no answer after an hour.

[dyitto]

I tried a more 'gradual' approach:

Define cages with unknown totals.
(A valid grid is supposed to have at least unique numbers in each cage.)

For each cage:

Generate some valid grids and calculate the corresponding totals for that cage.
If some total appears to be higher, lower or more rare then average, then stick to that total for that cage.

Then go on to the next cage etc.

I stumbled upon this one:



It needs few guesses. I've also checked it with JSudoku.
For three of the cages, knowledge of the total is not essential for uniqueness of the solution.

[tarek]

Hi dyitto,

Redundant cages can be removed ... but it is not a must .. you can make a similar comparison with minimal and non minimal vanilla sudoku. both are playable & legal.

A minimal killer:
If ANY sum constraint is made unknown then the puzzle would have multiple solutions

IMO, Killer puzzle designers should attempt the following if possible to achieve a better design:
1. All cells should be covered with sum constraint cages.
2. No single cell cage sum (i.e. No killer with vanilla style given(s))

tarek

[HATMAN]

Of course if you disobey the first rule Tarek mentions it is called a zero killer, see:

http://www.rcbroughton.co.uk/sudoku/for ... ?f=3&t=848

[dyitto]

Hi Hatman, your link doesn't seem to work.

[Pat]

"Zero Killer"

[HATMAN]

Thanks Pat