# | User | Rating |
---|---|---|
1 | ecnerwala | 3649 |
2 | Benq | 3581 |
3 | orzdevinwang | 3570 |
4 | Geothermal | 3569 |
4 | cnnfls_csy | 3569 |
6 | tourist | 3565 |
7 | maroonrk | 3531 |
8 | Radewoosh | 3521 |
9 | Um_nik | 3482 |
10 | jiangly | 3468 |
# | User | Contrib. |
---|---|---|
1 | maomao90 | 174 |
2 | awoo | 164 |
3 | adamant | 163 |
4 | TheScrasse | 159 |
5 | nor | 158 |
6 | maroonrk | 156 |
7 | -is-this-fft- | 152 |
8 | SecondThread | 147 |
9 | orz | 146 |
10 | pajenegod | 145 |
Name |
---|
The algorithm basically reduces to this: the result for a position i = v[i-k] & v[i-k+1] & ... &v[i] & ... & v[i+k-1] & v[i+k], not taking into account the fact that the array is cyclical. The algorithm that you propose is O(log(maximum value) * N), as you do a pass for each bit in the values of the array. However, we can achieve O(N * log(N)) by using a interval tree which calculates the & value for arbitrary subsegments for the array. By using this, we can do a single pass through the results, and, for each result, calculate the value of v[i-k] & ... & v[i+k] (we have a special case in which either i-k or i+k reaches past the end of the array. Then we have v[0] & ... & v[(i+k)%n] & v[(i-k)&n] & ... & v[n-1], if we index the array v from 0). Since log(N) ~= 15, whereas log(maximum value) ~= 31, with this technique we can halve the execution time.
(for example source code you can look here http://pastebin.com/vD8renXr . I implemented the segment tree stupidly in this solution, so it might be a bit hard to understand -- sorry :( )