C++ case multimap

Question

I have a multimap defined by

typedef std::pair<int, int> au_pair; //vertices
typedef std::pair<int, int> acq_pair; //ch qlty specified by C
typedef std::multimap<int, acq_pair> au_map;
typedef au_map::iterator It_au;

The no. of simulations depend on the size of the au_map . For eg: if the au_map.size() = 5 I will have C1, C2, C3, C4, C5. and therefore 2^5 = 32 cases.

For example: If the au_map.size()=4 , I need to simulate my algorithm for 16 cases.

for(size_t i = 0; i != 16; ++i)
{
  for(It_au it = a_map.begin(); it != a_map.end();)
  {
    acq_pair it1 = it->second;
    //case 0:
    //C1 = 0, C2 = 0, C3 = 0, C4 = 0
    //@Matthieu M 's suggestion http://stackoverflow.com/questions/3110975/c-case-declaration-closed
    //bool const c1 = i & 1;
    //bool const c2 = i & 2;
    //bool const c3 = i & 4;
    //bool const c4 = i & 8;
    //Update it1.second with corresponding C values
    it->second.second = C1;
    it++;
    it->second.second = C2;
    it++;
    it->second.second = C3;
    it++;
    it->second.second = C4;
    it++;
  }
  //simulate algorithm
}

How can I automate this process, where the size of C changes according to the au_map.size() ? Thus, I will have C1, C2, C3, C4 when au_map.size() = 4 and C1, C2, C3, C4, C5 when au_map.size() = 5 .

Also, what's preferred a vector with these values or add this to a pair inside multimap? Vector lookup time is lesser than multimap.

Also, if I keep inserting values to a multimap, will the new/updated values be passed to the algorithm?

Answer 1

Like others, I'm not certain I completely understand your question. It looks like all you need is a representation of the binary of each integer from 0 to 2 ^bits -1 that you can conveniently reference (in the cases you mention, bits is either 4 or 5, but you want to generalize). If that's the case, a simpler to manage and access structure will be a vector of bool vectors. That is, rather than using a std::multimap , for general values of bits , replace the std::multimap with a std::vector<std::vector<bool> > ... something like:

std::vector<std::vector<bool> > c_flags(1 << bits);

for (size_t i = 0; i < c_flags().size(); ++i)
{
    for (size_t j = 0; j < bits; ++j)
        c_flags[i].push_back( (i & (1 << j)) > 0);
}

At this point, c_flags[i] contains a vector of bools representing the binary digits of i where true and false correspond to 1 and 0 respectively.

You could also use a std::map<std::vector<bool> > instead of the std::vector<std::vector<bool> > which may reduce memory requirements (if you don't need all of the possible binary representations) at the expense of being more expensive computationally. I don't see why you need to use a std::multimap , but then I don't have a lot of insight into the specifics of the problem you're trying to address either.

Answer 2

What are C1, C2, and so on? Are they just ints, or strings? In that case you could automatically generate them by keeping a counter variable.
Why would you want a pair<int,int> inside the multimap?
I don't understand the last question.

Answer 3

No disrespect intended, but you ask the most confusing questions. I had to dig up your previous question to understand this one a little better and I'm not sure if I understood the previous question either.

Originally I have 4 inputs C1, C2, C3, C4. This means I have a total of 16 combinations:

0000 0001 . . . 1111

How can I automate this process, where the size of C changes [...]

Normally the simplest way is to write nested loops to generate the combinations (I know this isn't what you want, keep reading):

for (int a=0; a < 2; ++a)
{
    for (int b=0; b < 2; ++b)
    {
        for (int c=0; c < 2; ++c)
        {
            for (int d=0; d < 2; ++d)
            {
                // I'm just printing the values here but
                // you could insert them to a container if 
                // you want.
                cout << a << b << c << d << endl;
            }
        }
    }
}

However, if you can't determine the number of nested loops we need to write in advance (ex: if the size of C is based on runtime conditions), then consider a recursive solution to generate the combinations.

void generate_combinations(int depth, int max_depth, string str)
{
    if (depth < max_depth)
    {
        generate_combinations(depth + 1, max_depth, str + "0");
        generate_combinations(depth + 1, max_depth, str + "1");
    }
    else
        cout << str << " ";
}

int main()
{
    generate_combinations(0, 3, "");
}

This outputs:

000 001 010 011 100 101 110 111

While this:

generate_combinations(0, 4, "");

Outputs:

0000 0001 0010 0011 0100 0101 0110 0111 1000 1001 1010 1011 1100 1101 1110 1111

... and so on. You can control C however you like based on runtime conditions and very easily so.

Also, what's preferred a vector with these values or add this to a pair inside multimap? Vector lookup time is lesser than multimap.

If your data is dense (ex: indices ranging from 0 to N with no gaps), then there's no reason to use a map with int keys. Using a map with integral keys is only useful if the data you want to represent is sparse. Otherwise consider std::vector or std::deque.