Question

I am looking to store my data in a data structure which allows me to do the following efficiently:

I have 10 arrays of type 1 and 10 arrays of type 2 with 100 elements each. Each of these array store values of 100 variables. Accompanying these I have 20 arrays containing the corresponding variable IDs.
In total arrays of type 1 store values of 1000 variables and arrays of type 2 store different values of the same 1000 variables.

Now, I need to take a difference of the values of variables in arrays of type 1 from that of arrays of type 2. Let A be the one of the array of type 1. The problem is that variables in A would be spread 10 each across all 10 arrays of type 2.

A hash map would not be useful as I also need to iterate over the values in each array. Any ideas?

Answer 1

If you are intending to do such operations often you might want to construct and index at the program startup (the construction has O(n^2) time complexity) once and then refer to it later.

struct LookupEntry {
   A *item1;
   B *item2;
};

std::map< IdType, LookupEntry > index;

This would allow you to linearly go through arrays of type A and get the values from arrays of type B in a constant time. You can also subtract B items from A in a similar fashion.

Alternatively, you can go through the index items and subtract one item from another.