System.Single.Equals method
This article provides supplementary remarks to the reference documentation for this API.
The Single.Equals(Single) method implements the System.IEquatable<T> interface, and performs slightly better than Single.Equals(Object) because it does not have to convert the obj parameter to an object.
Widening conversions
Depending on your programming language, it might be possible to code an Equals method where the parameter type has fewer bits (is narrower) than the instance type. This is possible because some programming languages perform an implicit widening conversion that represents the parameter as a type with as many bits as the instance.
For example, suppose the instance type is Single and the parameter type is Int32. The Microsoft C# compiler generates instructions to represent the value of the parameter as a Single object, and then generates a Single.Equals(Single) method that compares the values of the instance and the widened representation of the parameter.
Consult your programming language's documentation to determine if its compiler performs implicit widening conversions of numeric types. For more information, see Type Conversion Tables.
Precision in comparisons
The Equals method should be used with caution, because two apparently equivalent values can be unequal because of the differing precision of the two values. The following example reports that the Single value .3333 and the Single returned by dividing 1 by 3 are unequal.
// Initialize two floats with apparently identical values
float float1 = .33333f;
float float2 = 1/3;
// Compare them for equality
Console.WriteLine(float1.Equals(float2)); // displays false
// Initialize two floats with apparently identical values
let float1 = 0.33333f
let float2 = 1f / 3f
// Compare them for equality
printfn $"{float1.Equals float2}" // displays false
' Initialize two singles with apparently identical values
Dim single1 As Single = .33333
Dim single2 As Single = 1/3
' Compare them for equality
Console.WriteLine(single1.Equals(single2)) ' displays False
One comparison technique that avoids the problems associated with comparing for equality involves defining an acceptable margin of difference between two values (such as .01% of one of the values). If the absolute value of the difference between the two values is less than or equal to that margin, the difference is likely to be an outcome of differences in precision and, therefore, the values are likely to be equal. The following example uses this technique to compare .33333 and 1/3, which are the two Single values that the previous code example found to be unequal.
// Initialize two floats with apparently identical values
float float1 = .33333f;
float float2 = (float) 1/3;
// Define the tolerance for variation in their values
float difference = Math.Abs(float1 * .0001f);
// Compare the values
// The output to the console indicates that the two values are equal
if (Math.Abs(float1 - float2) <= difference)
Console.WriteLine("float1 and float2 are equal.");
else
Console.WriteLine("float1 and float2 are unequal.");
// Initialize two floats with apparently identical values
let float1 = 0.33333f
let float2 = 1f / 3f
// Define the tolerance for variation in their values
let difference = abs (float1 * 0.0001f)
// Compare the values
// The output to the console indicates that the two values are equal
if abs (float1 - float2) <= difference then
printfn "float1 and float2 are equal."
else
printfn "float1 and float2 are unequal."
' Initialize two singles with apparently identical values
Dim single1 As Single = .33333
Dim single2 As Single = 1/3
' Define the tolerance for variation in their values
Dim difference As Single = Math.Abs(single1 * .0001f)
' Compare the values
' The output to the console indicates that the two values are equal
If Math.Abs(single1 - single2) <= difference Then
Console.WriteLine("single1 and single2 are equal.")
Else
Console.WriteLine("single1 and single2 are unequal.")
End If
In this case, the values are equal.
Note
Because Epsilon defines the minimum expression of a positive value whose range is near zero, the margin of difference must be greater than Epsilon. Typically, it is many times greater than Epsilon. Because of this, we recommend that you do not use Epsilon when comparing Double values for equality.
A second technique that avoids the problems associated with comparing for equality involves comparing the difference between two floating-point numbers with some absolute value. If the difference is less than or equal to that absolute value, the numbers are equal. If it is greater, the numbers are not equal. One way to do this is to arbitrarily select an absolute value. However, this is problematic, because an acceptable margin of difference depends on the magnitude of the Single values. A second way takes advantage of a design feature of the floating-point format: The difference between the mantissa components in the integer representations of two floating-point values indicates the number of possible floating-point values that separates the two values. For example, the difference between 0.0 and Epsilon is 1, because Epsilon is the smallest representable value when working with a Single whose value is zero. The following example uses this technique to compare .33333 and 1/3, which are the two Double values that the previous code example with the Equals(Single) method found to be unequal. Note that the example uses the BitConverter.GetBytes and BitConverter.ToInt32 methods to convert a single-precision floating-point value to its integer representation.
using System;
public class Example
{
public static void Main()
{
float value1 = .1f * 10f;
float value2 = 0f;
for (int ctr = 0; ctr < 10; ctr++)
value2 += .1f;
Console.WriteLine("{0:R} = {1:R}: {2}", value1, value2,
HasMinimalDifference(value1, value2, 1));
}
public static bool HasMinimalDifference(float value1, float value2, int units)
{
byte[] bytes = BitConverter.GetBytes(value1);
int iValue1 = BitConverter.ToInt32(bytes, 0);
bytes = BitConverter.GetBytes(value2);
int iValue2 = BitConverter.ToInt32(bytes, 0);
// If the signs are different, return false except for +0 and -0.
if ((iValue1 >> 31) != (iValue2 >> 31))
{
if (value1 == value2)
return true;
return false;
}
int diff = Math.Abs(iValue1 - iValue2);
if (diff <= units)
return true;
return false;
}
}
// The example displays the following output:
// 1 = 1.00000012: True
open System
let hasMinimalDifference (value1: float32) (value2: float32) units =
let bytes = BitConverter.GetBytes value1
let iValue1 = BitConverter.ToInt32(bytes, 0)
let bytes = BitConverter.GetBytes(value2)
let iValue2 = BitConverter.ToInt32(bytes, 0)
// If the signs are different, return false except for +0 and -0.
if (iValue1 >>> 31) <> (iValue2 >>> 31) then
value1 = value2
else
let diff = abs (iValue1 - iValue2)
diff <= units
let value1 = 0.1f * 10f
let value2 =
List.replicate 10 0.1f
|> List.sum
printfn $"{value1:R} = {value2:R}: {hasMinimalDifference value1 value2 1}"
// The example displays the following output:
// 1 = 1.0000001: True
Module Example
Public Sub Main()
Dim value1 As Single = .1 * 10
Dim value2 As Single = 0
For ctr As Integer = 0 To 9
value2 += CSng(.1)
Next
Console.WriteLine("{0:R} = {1:R}: {2}", value1, value2,
HasMinimalDifference(value1, value2, 1))
End Sub
Public Function HasMinimalDifference(value1 As Single, value2 As Single, units As Integer) As Boolean
Dim bytes() As Byte = BitConverter.GetBytes(value1)
Dim iValue1 As Integer = BitConverter.ToInt32(bytes, 0)
bytes = BitConverter.GetBytes(value2)
Dim iValue2 As Integer = BitConverter.ToInt32(bytes, 0)
' If the signs are different, Return False except for +0 and -0.
If ((iValue1 >> 31) <> (iValue2 >> 31)) Then
If value1 = value2 Then
Return True
End If
Return False
End If
Dim diff As Integer = Math.Abs(iValue1 - iValue2)
If diff <= units Then
Return True
End If
Return False
End Function
End Module
' The example displays the following output:
' 1 = 1.00000012: True
The precision of floating-point numbers beyond the documented precision is specific to the implementation and version of .NET. Consequently, a comparison of two numbers might produce different results depending on the version of .NET, because the precision of the numbers' internal representation might change.
