| p |
|
|
| p + r |
|
(≥ 0) |
| r |
|
(≥ 0) |
| -p |
|
(≤ 0) |
| p - r |
|
(≥ 0) |
| 2∙p - r |
|
(≥ 0) |
| p |
|
(≥ 0) |
| r - p |
|
(≤ 0) |
| - r |
|
(≤ 0) |
| p |
|
(≥ 0) |
| p + r |
|
|
|
From (0) we conclude (i) that the sequence contains a nonnegative element, (ii) that one of its neighbours is nonnegative, and (iii) that at least one of the two elements adjacent to a pair of nonnegative neighbours is nonnegative. More precisely: the sequence contains in some direction a triple of adjacent elements of the form (p, p+r, r) with 0 ≤ r ≤ p. To the left we have extended the sequence with another 8 elements. From (0) we further conclude that the whole sequence is determined by a pair of adjacent values; hence, the repetition of the pair (p, p+r) at distance 9 proves the theorem. [The above deserves recording for its lack of case analyses.] |