Using C++ for equation, Gaussian probability

[Simon Mitchell]

I'm trying to write a programme a c++ program to evaluate the Gaussian probability for x=0,1.5,3 where standard deviation =0.5 and the mean is 1.5 this is what I have

#include <cstdlib>
#include <iostream>
#include <cmath>
#include <cfloat>
using namespace std;
 
int main(int argc, char *argv[])
{double x1,x2,x3,u,o,g1,g2,g3,j1,j2,j3;
x1=0;
x2=3/2;
x3=3;
u=3/2;
o=1/2;
j1=x1-u;
j2=x2-u;
j3=x3-u;
g1=((exp(0)/2*pow(o,2))/(o*sqrt(2*M_PI)));
g2=(1/(o*sqrt(2*M_PI)))*exp(-pow(j2,2)/2*pow(o,2));
g3=(1/(o*sqrt(2*M_PI)))*exp(-pow(j3,2)/2*pow(o,2));
cout<< "for x equal to 0 the guassian ="<<g1<<endl;
cout<< "for x equal to 1.5 the guassian ="<<g2<<endl;
cout<< "for x equal to 3.0 the guassian ="<<g3<<endl;
    system("PAUSE");
    return EXIT_SUCCESS;
}

it returns for all of them 1.INF I have no clue what this means.

[donbock]

x1=0; 
x2=3/2; 
x3=3; 
u=3/2; 
o=1/2;

It is unwise to use rational constants to initialize floating-point variables. The compiler sees an expression with integer terms so it uses integer math and assigns the integer result to the variable. For example, the value of 1/2 is 0. It is much better to use floating point constants or floating point terms. Some people [not me] prefer to use casts. For example:

x2 = 1.5;    // floating point constant.
u = ((double)3)/((double)2);   // cast.
o = 1.0/2.0;   // floating point terms.

[Oralloy]

Simon,

The problem is that you're generating infinities. That's what the program is trying to tell you.

Don was exactly correct in observing that you can't use "rational" numbers in the way you have.

Basically, when you write

x2=3/2;

This means

x2=1;

In your program.

This is because you didn't include a decimal point or other floating point indicator in your constants three (3) and two (2). So the compiler did exactly what you told it to do, which is to divide one integer by the other.

What got you is that dividing integers returns an integer result. Thus, 3/2 is equal to the integer value one (1), and not the floating point value one-and-a-half (1.5).

So, if you want to write rational number style constants, you must include one of the floating-point forms that Don illustrated.

Good Luck!