Cycle check Implementation
Based on the thought that if the thief can escape to a cycle, then, the chasing becomes infinite.
- Only cycles that contain more than 3 nodes can be available for the thief.
- Even in a 4 nodes cycle, if all thief's next move is within the police's range, the thief can't escape.
Backtracking Implementation
Assumption: the police is always heading to the thief by the shortest path.
- Map. Space complexity:
N*w - Trie. Space complexity:
T*N ~ T*N*w
When T<w, a Trie could possible win a Map.
T= expansion way degreeN= distinct key countw= average key length
R-way Trie implementation
Ts Trie implementation
Memory measure: Digest 1,000,000 random alphabet String of W width. Max distinct key count: Permutation(26,W)
- W=3: 15,600
- W=4: 358,800
- W=5: 1,000,000(7,893,600)
See test
A typical result(MB):
| Implementation | Width of 3 | Width of 4 | Width of 5 |
|---|---|---|---|
| 256-way Trie | 716 | NA | NA |
| HashMap_0.25 | 26 | 83 | 89 |
| HashMap_0.75 | 13 | 71 | 71 |
| HashMap_1.00 | 13 | 59 | 65 |
| TreeMap | 13 | 59 | 65 |
| Ts Trie | 7 | 45 | 85 |
(Tested serval times on HotSpot 1.8.0_152 64bit)
Memory usage of TsTrie is pretty much stable. TreeMap sizes should be affected by random key count. A HashMap's size also relates to its load factor.