Proposed by Peter Nightingale
Killer Sudoku is a puzzle played on a $\{9\times 9\}$ grid containing 81 cells. The cells are filled in with numbers from the set $\{1\ldots 9\}$. Each row and column must contain all numbers $\{1\ldots 9\}$. Each of the 9 non-overlapping $3\times 3$ subsquares (named boxes) must also contain all numbers $\{1\ldots 9\}$.
Each Killer Sudoku puzzle has a set of cages. A cage is a set of contiguous cells and a total; the numbers in the cells must add up to the total. Also, the cells in a cage cannot contain the same number more than once. The cages do not overlap, and they cover all cells. Cages typically contain two to four cells. Typically a Killer Sudoku puzzle will have exactly one solution.
An example Killer Sudoku puzzle is shown below. Each cage is shown as an area of one colour.
The solution of the above puzzle is shown below.
There is a straightforward generalisation of Killer Sudoku. For any $n$ that has an integer square root, we have an $n \times n$ grid and each cell takes any value in $\{1\ldots n\}$. In a solution each row and column contains all numbers $\{1\ldots n\}$, and the $n$ non-overlapping $\sqrt{n} \times \sqrt{n}$ boxes also contain all numbers $\{1\ldots n\}$. Cages function in the same way in the generalised problem as in the $\{9\times 9\}$ problem.