How Can I Avoid Floating-Point Errors When Converting Floats to Integers in Go?

Golang: Avoiding Floating-Point Errors in Float-to-Int Conversions

Floating-point operations can introduce errors when converting between different numerical types. In Go, converting a float64 to an int may lead to unexpected results due to floating-point representation limitations.

Understanding Floating-Point Representation

Computers typically store numeric data in binary format. However, decimal numbers like 100.55 cannot be represented as a finite number in a binary system.

IEEE-754 is the standard used by Go to represent float64 values. It employs 53 bits for storing digits and 11 bits for the exponent, allowing for a finite range of numbers but inevitably introducing approximations.

Decimal to Binary Conversion

When converting a decimal number like 100.55 to the float64 internal representation, it cannot be expressed precisely. Instead, the nearest binary number is used, resulting in a slightly different value from the original.

Error in Float-to-Int Conversion

In the code example, subtracting float64(int(x)) should yield 0.55, but the output is closer to 0.5499999999999972. This error occurs because the subtraction is performed between two different representations of the same number, with the float64(int(x)) representing 100.0, not 100.55.

Solutions

To avoid precision errors, consider the following approaches:

• Round the Result: Use fmt.Printf with an appropriate format string to round the output to the desired number of decimal places.
• Avoid Floating-Point Numbers: For applications involving currency or precise calculations, consider representing values as integers or rational numbers to eliminate floating-point errors.

The above is the detailed content of How Can I Avoid Floating-Point Errors When Converting Floats to Integers in Go?. For more information, please follow other related articles on the PHP Chinese website!