Given an array A of non-negative integers, where . The task is to count number of distinct possible results obtained by taking the bitwise OR of all the elements in all possible Subarrays.
Input: A = [1, 2] Output: 3 Explanation: The possible subarrays are , , [1, 2]. These Bitwise OR of subarrays are 1, 2, 3. There are 3 distinct values, so the answer is 3. Input: A = [1, 2, 4] Output: 6 Explanation: The possible distinct values are 1, 2, 3, 4, 6, and 7.
The Naive approach is to generate all possible subarrays and take bitwise OR of all elements in the subarray. Store each result in set and return length of the set.
We can make the above approach better. The Naive approach is to calculate all possible result where, res(i, j) = A[i] | A[i+1] | … | A[j]. However we can speed this up by taking note of the fact that res(i, j+1) = res(i, j) | A[j+1]. At the kth step, say we have all of the res(i, k) in some set pre. Then we can find the next pre set (for k -> k+1) by using res(i, k+1) = res(i, k) | A[k+1].
However, the number of unique values in this set pre is atmost 32, since the list res(k, k), res(k-1, k), res(k-2, k), … is monotone increasing, and any subsequent values that are different from previous must have more 1’s in it’s binary representation which can have maximum of 32 ones.
Below is the implementation of above approach.
Time Complexity: O(N*log(K)), where N is the length of A, and K is the maximum size of elements in A.