5.2 Numerical evaluation

GiNaC keeps algebraic expressions, numbers and constants in their exact form. To evaluate them using floating-point arithmetic you need to call

ex ex::evalf(int level = 0) const;

The accuracy of the evaluation is controlled by the global object Digits which can be assigned an integer value. The default value of Digits is 17. See section Numbers, for more information and examples.

To evaluate an expression to a double floating-point number you can call evalf() followed by numeric::to_double(), like this:

{
    // Approximate sin(x/Pi)
    symbol x("x");
    ex e = series(sin(x/Pi), x == 0, 6);

    // Evaluate numerically at x=0.1
    ex f = evalf(e.subs(x == 0.1));

    // ex_to<numeric> is an unsafe cast, so check the type first
    if (is_a<numeric>(f)) {
        double d = ex_to<numeric>(f).to_double();
        cout << d << endl;
         // -> 0.0318256
    } else
        // error
}