Vectors as arguments and C++ vs. Visual C++

Ivan · #1 28-May-2009, 13:33

The following code successfully compiles and produces the expected result in Visual C++ 2008; in Linux, although it still compiles (with g++ -g ...), the result of the execution is “Segmentation fault”. Could anyone, please, explain why this is happening and what is the problem with the code?

Supposedly, VectorFunction takes in the function f of one variable x and a vector v,
and computes f(v) component-wise.

CPP / C++ / C Code:

#include<vector>
#include<iostream>
#include<cmath>

using namespace std;



vector<double> VectorFunction( double (*f)(double x), vector<double> v)

{
  vector<double> a;

  for (int i=0; i<v.size(); i++){

    a[i]= f(v[i]);

}

  return a;

}

double sq(double x) {return x*x;}



int main()

{
     vector<double> b(2);
      b[0]=1; b[1]=2; 
 
      vector<double> c; c =  VectorFunction<double>( sq, b);
      cout<< c[0]<< endl;

  return 0;

}

davekw7x · #2 28-May-2009, 14:44

Quote:

Originally Posted by Ivan

The following code successfully compiles

Well, the following statement is invalid and keeps it from compiling. See Footnote.

CPP / C++ / C Code:

int main()
{
.
.
.
      c =  VectorFunction<double>( sq, b);
.
.

Change it to

CPP / C++ / C Code:

    c = VectorFunction(sq, b);

Now, it compiles, but the function has undefined behavior.

The program crashes when I compile it with Visual C++ on Windows XP and also when I compile it with GNU g++ (Windows XP or Linux), but since the behavior is undefined, it is not guaranteed to crash. Since the code that you posted won't even compile, I would like to see what code you are actually using that compiles and seems to execute properly with Visual C++. Really.

Anyhow...

CPP / C++ / C Code:

vector<double>VectorFunction(double (*f) (double x), vector<double>v)
{
    vector < double >a; // An empty vector: size = 0

    for (int i = 0; i < v.size(); i++) {
        a[i] = f(v[i]); // Invalid, since a[i] does not exist
    }
    return a;
}

I can think of a couple of ways to change the function so that it works with no undefined behavior.

1. Declare the vector a to have the same size as the vector v and leave everything else as is:

CPP / C++ / C Code:

.
.
    vector < double >a(v.size()); // Same size as v
.
.

or

2. Leave the declaration as is, and change the statement inside the loop to push the values of v[i] into the vector:

CPP / C++ / C Code:


        vector < double >a; // Size = 0;
.
.
.
        a.push_back(f(v[i]));
.
.

Regards,

Dave

Footnote: I am having a hard time believing that Visual C++ (any version) or Linux GNU g++ (any version) or any other compiler accepts that first illegal statement that I showed. Are you sure that it compiled?

Ivan · #3 28-May-2009, 15:20

Thanks a lot, davekw7x. You are absolutely right that

CPP / C++ / C Code:

c =  VectorFunction<double>( sq, b);

is unvalid. The function was a template function originally, and I forgot to erase <double> while posting the message. I apologize for that.

Your suggestions with respect to declaration/initialization of a vector "a" work very well.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Rate This Thread
Rate This Thread:

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Vectors	ultrAslan	C++ Forum	8	15-Mar-2009 09:22
Command Line Arguments question	bostero22	Java Forum	2	10-Mar-2009 14:06
Major newbie problem	cynack	MS Visual C++ / MFC Forum	1	08-Apr-2007 12:25
main() arguments	crystalattice	C Programming Language	2	24-Oct-2004 01:05
Re: Command Line Arguments, Part 1	WaltP	C Programming Language	0	11-Jul-2004 00:34

Vectors as arguments and C++ vs. Visual C++

Re: Vectors as arguments and C++ vs. Visual C++

Re: Vectors as arguments and C++ vs. Visual C++