부동 소수점: f32와 f64

When you need decimals, Zig offers floating-point types. They store fractional values like 3.14 instead of whole numbers only.

The f64 type holds a double-precision float using 64 bits. It is the usual default when you want accuracy and plenty of range.

An f32 uses 32 bits. It saves memory and can be faster, but it stores fewer significant digits than an f64.

Beyond f32 and f64, Zig also supports f16 and f128. You choose the width that balances precision against memory and speed.

A float literal includes a decimal point, like 2.0. Writing just 2 gives an integer, which is a different kind of value.

You can write very large or small floats with an exponent. The form 1.5e3 means 1.5 times ten to the third power, or 1500.

The usual operators apply to floats. Division of two floats keeps the fractional part instead of throwing it away.

Floats cannot store every decimal exactly. Tiny rounding errors mean 0.1 plus 0.2 may not equal exactly 0.3.

Because of rounding, avoid testing floats for exact equality. Check that the difference is smaller than a tiny threshold instead.

To turn an integer into a float you ask explicitly with @floatFromInt. Zig never mixes the two number kinds silently.

Floats can also represent infinity and not-a-number. These appear from operations like dividing by zero, and the std library names them.

You divide one float by another and want to keep the fractional result. What is true about floating-point math in Zig?

Use f64 for general decimals and f32 to save space. Remember floats are approximate, so compare with a tolerance and cast on purpose. 🎯