C ++类变量：向量的初始化与赋值和初始化

Question

I am working on a C++ program that has a series of class variables that contain vectors on some or all of the member variables. My question is three-fold:

1. Is it straight-forward to use constructors to initialize vector variables that are part of a class (see sample class definition below)? Could someone post an example constructor for the class below (or for at least the single and two-dimension vector variables)?

2. Is there a problem with simply initializing the variables myself in my code (ie, iterating through each element of the vectors using loops to assign an initial value)?

3. Along the same lines, if the variables need to be initialized to different values in different contexts (eg, zero in one instance, some number in another instance), is there a way to handle that through constructors?

Sample class definition:

class CreditBasedPoolLevel {
public:
    int NumofLoans; 
    int NumofPaths; 
    int NumofPeriods; 
    double TotalPoolBal;

    vector<int> NumofModeled;
    vector<double> ForbearanceAmt;
    vector<double> TotalAmtModeled;

    vector<vector<int>> DefCountPoolVector;
    vector<vector<double>> TermDefBalPoolVector;
    vector<vector<double>> BalloonDefBalPoolVector;
    vector<vector<double>> TermDefBalPoolVectorCum; 
    vector<vector<double>> TermSeverityAmt;
    vector<vector<double>> TermELAmtPoolVector;
    vector<vector<double>> BalloonELAmtPoolVector;
    vector<vector<double>> TermELAmtPoolVectorCum;

};

Answer 1

As for question number three, simply add a constructor with an argument that is the value you want to initialize the vectors with.

And if you just want the vectors to be default constructed, then there's nothing that needs to be done.

Answer 2

Constructor may look something like this:

CreditBasedPoolLevel::CreditBasedPoolLevel()
{
   const int numDefCountPools = 13;
   const int numDefCountPoolEntries = 25;

   for(int i = 0; i < numDefCountPools; i++)
   {
      vector<int> v;
      for(int j = 0; j < numDefCountPoolEntries; j++)
      {
          v.push_back(j + i * 5);  // Don't know what value you ACTUALLY want to fill here
      }
      DefCountPoolVector.push_back(v);
   }
}

Note that this is ONE solution, it really depends on what values you want, how you went them organized, etc, what is the "right" solution for your case.

Answer 3

1. In C++, initializing a variable calls its constructor. In a vector's case, this means it creates an instance of a vector with whatever the initial capacity is (10 I believe), with no values. At this point, you need to use push_back in order to fill the vector - even though it has a capacity, it will cause undefined behavior if you try to access unfilled areas directly (such as with NumofModeled[0]). You can also initialize it with some amount of space by using vector NumofModeled(x) (x being the number of spaces), but generally because vectors have dynamic size, it's easier to use push_back unless there is some reason you need to enter your data out of order.

2. Relates to the capacity part of one, if you try to access unfilled space in a vector you will get undefined behavior. It's pretty standard practice to fill a vector with a loop though, such as:

    vector<int> v; int in = 0; while (cin) { cin >> in; v.push_back(in); } 

3. Yes, but remember that like functions, constructors only differentiate by the type of input parameters. So, for example, you could have CreditBasedPoolLevel(int level) and CreditBasedPoolLevel(vector<int> levels) , but not another with the definition CreditBasedPoolLevel(int otherint) , because it would conflict with the first. If you want to be able to take different contextual input of the same type, you can use another variable to define the constructor type, such as CreditBasedPoolLevel(int input, string type) and use a switch block to define the initialization logic based on the type.