Question

I have a program to determine the largest contiguous sum in an array, but want to extend it to work with circular arrays. Is there an easier way to do that than doubling the single array and calling my function to find the largest sum over all n-length arrays in the 2n length array?

Answer 1

I assume you are using the O(n) algorithm that continues adding to the sum, keeping track of the maximum, only restarting if you sum to a negative number. The only thing you need to do to capture the case of circular arrays is to apply the same principle to the circular aspect. When you reach the end of the array in the original algorithm, keep looping around to the start until you go below the maximum or hit the beginning of the current range (I think this is impossible, though, because if the solution was the full array, we sould have seen this on the first pass), in which case you’re done.

max_start=0; max_end =0; maxv = 0; sum 0;
for i in range(arr):
    sum+= arr[i];
    if sum<0:
       sum=0; max_start =i;
    if maxv<sum:
       maxv=sum; max_end = i;

#seocnd pass
for i in range(max_start):
    sum+= arr[i];
    if sum<0:
       break;
    if maxv<sum:
       maxv=sum;max_end = i;