C ++中的浮点数学运算输出

Question

I am trying to right a simple math program, and I am facing difficulty getting correct formatted output.

Here is a sample code:

#include <iostream>
#include <iomanip>
#include <cmath>
using namespace std;

const double PI = 3.14159265358979;

int main()
{
        double a,b,c,d;

        a = cos(30*PI/180);
        b = sin(30*PI/180);
        c = cos(90*PI/180);
        d = sin(90*PI/180);

        cout << a << endl;
        cout << b << endl;
        cout << c << endl;
        cout << d << endl;
}

which give the following output:

0.866025
0.5
1.61554e-015
1

I tried to use round as follows but round will round all of them

cout << round(a) << endl;
cout << round(b) << endl;
cout << round(c) << endl;
cout << round(d) << endl;  

which gives the following output:

1
0
0
1

at last I tried fixed but it is fixed for all

cout << fixed << a << endl;
cout << fixed << b << endl;
cout << fixed << c << endl;
cout << fixed << d << endl;

Output

0.866025
0.500000
0.000000
1.000000

What I am trying to get is output output like:

0.866025
0.5
0
1

I know floating-point is hard to deal with due to representing the unlimited by a limited storage.

I went through a lot of readings about floating-point numbers but have not find how can I get the required result using only the standard C++ libraries.

and because angles 30 and 90 is just a sample I cannot use different output techniques for each variable.

I would prefer sticking to cout rather than printf as long as possible.

I appreciate your help in advance.

Answer 1

I assume the offending output is the one for c using scientific notation. If so, you'll need to deal with the values separately: the flexibly format (neither std::ios_base::fixed nor std::ios_base::scientific is set and the format is equivalent to the use of printf() 's %g format) will print values in a fashion which represents values with the set precision as best as it can. This approach will use scientific notation when the number of digits needed to reasonably represent the value.

One way to deal with this situation is to explicitly detect "bad" values, eg, values below some threshold. If you want to avoid checking each output individually you can imbue() a custom std::locale object with a custom std::num_put<...> facet, eg:

#include <iostream>
#include <locale>
#include <iomanip>
#include <cmath>

const double PI = 3.14159265358979;

struct num_put: std::num_put<char> {
    iter_type do_put(iter_type to, std::ios_base& fmt, char_type fill, double v) const {
        if (v < 1e-4) {
            *to = '0';
            ++to;
            return to;
        }
        return std::num_put<char>::do_put(to, fmt, fill, v);
    }
};

int main()
{
    std::cout.imbue(std::locale(std::locale(), new num_put));

        double a,b,c,d;

        a = std::cos(30*PI/180);
        b = std::sin(30*PI/180);
        c = std::cos(90*PI/180);
        d = std::sin(90*PI/180);

        std::cout << a << '\n';
        std::cout << b << '\n';
        std::cout << c << '\n';
        std::cout << d << '\n';
}

This quick demo also fixed a few unrelated problems:

• do not use std:endl : if you mean to flush the stream explicitly use std::flush
• using directives are bad