Magnetism is played by two players on a strip of squares, who take turns placing + and – tokens onto the strip. The only rule is that no two tokens with the same parity can be placed next to each other. For instance, there are three legal moves in the following position:
The player who moves last, wins. This makes Magnetism an impartial game, so that each position is equivalent to a Nim-pile. It turns out that Magnetism is very simple.
First we notice that any position is the sum of simpler positions that have tokens just at the end of a strip. (A sum of games is played by first choosing a game summand, and then making a move in that summand).
Therefore we will know everything about Magentism if we can determine the size of the Nim-piles (the “nimbers”) of the 9 elementary positions:
Things get even simpler. Because of the symmetry of things, there are only four truly different boards to consider.
Let’s denote the nimbers of a board of n empty squares (thus not counting the tokens at the end when present) by G(n), G+(n), G++(n), and G+-(n).
| n | 0 | 1 | 2 | 3 | 4 | 5 | 6 |
|---|---|---|---|---|---|---|---|
| G(n) | 0 | 1 | 0 | 1 | 0 | 1 | 0 |
| G+(n) | 0 | 1 | 2 | 3 | 4 | 5 | 6 |
| G++(n) | – | 1 | 1 | 1 | 1 | 1 | 1 |
| G+-(n) | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
Now you can win in a position with a positive nimber by moving to a position with zero nimber. For instance, on a board with a single + at one end, one possible winning move is to put a – at the other end.
The Inner Frame 