Common operators

After having explained how mp++’s type coercion works, we can now move on to describe how mp++’s classes can be used to perform mathematical computations.

All of mp++’s multiprecision classes support the four basic arithmetic operations (\(+\), \(-\), \(\times\) and \(\div\)) via overloaded binary operators:

auto r1 = int_t{4} + int_t{3};
assert(r1 == 7);

auto r2 = rat_t{4, 3} - rat_t{2, 5};
assert(r2 == rat_t{14, 15});

auto r3 = real128{5} * real128{2};
assert(r3 == 10.);

auto r4 = real{5} / real{2};
assert(r4 == 2.5);

It is of course possible to mix operands of different types, and the type of the result will be determined by the type coercion rules explained earlier:

auto r1 = int_t{4} + 3;        // r1 will be of type int_t.
assert(r1 == 7);

auto r2 = -0.75 + rat_t{1, 2}; // r2 will be of type double.
assert(r2 == -.25);

auto r3 = real128{5} * 2;      // r3 will be of type real128.
assert(r3 == 10.);

auto r4 = real{5} / int_t{2};  // r4 will be of type real.
assert(r4 == 2.5);

The behaviour of the division operator varies depending on the types involved. If only integral types are involved, division truncates, and division by zero throws a zero_division_error exception:

auto r1 = int_t{5} / 2; // Integral division truncates.
assert(r1 == 2);

int_t{1} / 0;           // This will throw a zero_division_error exception.

If floating-point types are involved, division by zero is allowed and results in infinity:

auto r1 = real128{1} / 0; // Floating-point division by zero generates an infinity.
assert(isinf(r1))

Rational division is always exact, unless the divisor is zero:

auto r1 = rat_t{3} / 4;
assert(r1 == rat_t{3, 4});

rat_t{2} / 0;              // This will throw a zero_division_error exception.

The in-place versions of the binary operators are available as well. Given a binary operator, its corresponding in-place counterpart behaves as a binary operation followed by assignment:

int_t r1{4};
r1 += 5;                    // Equivalent to: r1 = r1 + 5
assert(r1 == 9);

rat_t r2{4, 3};
r2 -= 1.5;                  // Equivalent to: r2 = r2 - 1.5
assert(r2 == rat_t{-1, 6});

real128 r3{5};
r3 *= rat_t{1, 2};          // Equivalent to: r3 = r3 * rat_t{1, 2}
assert(r3 == 2.5);

real r4{42};
r4 /= real128{0};           // Equivalent to: r4 = r4 / real128{0}
assert(isinf(r4));

It is also possible to use fundamental C++ types on the left-hand side of in-place operators:

int n = 5;
n += int_t{5};
assert(n == 10);

n -= rat_t{3, 4}
assert(n == 9);

double x = 1.5;
x *= real128{2};
assert(x == 3.);

x /= real{3};
assert(x == 1.);

The identity, negation, pre/post increment/decrement operators are also supported for all of mp++’s multiprecision classes:

int_t n;
assert(++n == 1);
n++;
assert(n == 2);
assert(--n == 1);
n--;
assert(n == 0);

rat_t q{1, 2};
assert(+q == q);
assert(-q == rat_t{-1, 2});

All of mp++’s multiprecision classes are contextually convertible to bool, following the usual rule that nonzero values convert to true and zero values convert to false:

int_t n{3};
if (n) {
   std::cout << "n is nonzero!\n";
}

rat_t q{0};
if (!q) {
   std::cout << "q is zero!\n";
}

real r{1.23};
if (!!r) {
   std::cout << "r is nonzero!\n";
}

In addition to the common arithmetic operators, all of mp++’s multiprecision classes support the relational operators \(=\) and \(\neq\). Real-valued types also support the comparison operators \(>\), \(\geq\), \(<\) and \(\leq\). Any combination of multiprecision and numerical C++ types is supported:

assert(int_t{42} == 42);
assert(3 != rat_t{1, 3});
assert(0.9 < int_t{1});
assert(real128{15} <= int_t{15});
assert(real{"inf", 100} > rat_t{123, 456});
assert(int_t{4} >= rat_t{16, 4});

The comparison operators treat NaN values in the standard way: comparing NaN to any other value returns always false, apart from the \(\neq\) operator which always returns true when NaN is involved. For the floating-point multiprecision classes, custom comparison functions with special NaN handling are also available (e.g., real_lt(), real128_equal_to(), etc.). These functions can be used as replacements for the comparison operators in facilities of the standard library such as std::sort(), std::set, etc.