Formatting doubles for output in C#

user_1527 0
Viewed 137913

Running a quick experiment related to Is double Multiplication Broken in .NET? and reading a couple of articles on C# string formatting, I thought that this:

{
    double i = 10 * 0.69;
    Console.WriteLine(i);
    Console.WriteLine(String.Format("  {0:F20}", i));
    Console.WriteLine(String.Format("+ {0:F20}", 6.9 - i));
    Console.WriteLine(String.Format("= {0:F20}", 6.9));
}

Would be the C# equivalent of this C code:

{
    double i = 10 * 0.69;

    printf ( "%f\n", i );
    printf ( "  %.20f\n", i );
    printf ( "+ %.20f\n", 6.9 - i );
    printf ( "= %.20f\n", 6.9 );
}

However the C# produces the output:

6.9
  6.90000000000000000000
+ 0.00000000000000088818
= 6.90000000000000000000

despite i showing up equal to the value 6.89999999999999946709 (rather than 6.9) in the debugger.

compared with C which shows the precision requested by the format:

6.900000                          
  6.89999999999999946709          
+ 0.00000000000000088818          
= 6.90000000000000035527

What's going on?

( Microsoft .NET Framework Version 3.51 SP1 / Visual Studio C# 2008 Express Edition )

I have a background in numerical computing and experience implementing interval arithmetic - a technique for estimating errors due to the limits of precision in complicated numerical systems - on various platforms. To get the bounty, don't try and explain about the storage precision - in this case it's a difference of one ULP of a 64 bit double.

To get the bounty, I want to know how (or whether) .Net can format a double to the requested precision as visible in the C code.

ieee-754 floating-point formatting c#
10 Answers
Digits after decimal point
// just two decimal places
String.Format("{0:0.00}", 123.4567);      // "123.46"
String.Format("{0:0.00}", 123.4);         // "123.40"
String.Format("{0:0.00}", 123.0);         // "123.00"

// max. two decimal places
String.Format("{0:0.##}", 123.4567);      // "123.46"
String.Format("{0:0.##}", 123.4);         // "123.4"
String.Format("{0:0.##}", 123.0);         // "123"
// at least two digits before decimal point
String.Format("{0:00.0}", 123.4567);      // "123.5"
String.Format("{0:00.0}", 23.4567);       // "23.5"
String.Format("{0:00.0}", 3.4567);        // "03.5"
String.Format("{0:00.0}", -3.4567);       // "-03.5"

Thousands separator
String.Format("{0:0,0.0}", 12345.67);     // "12,345.7"
String.Format("{0:0,0}", 12345.67);       // "12,346"

Zero
Following code shows how can be formatted a zero (of double type).

String.Format("{0:0.0}", 0.0);            // "0.0"
String.Format("{0:0.#}", 0.0);            // "0"
String.Format("{0:#.0}", 0.0);            // ".0"
String.Format("{0:#.#}", 0.0);            // ""

Align numbers with spaces
String.Format("{0,10:0.0}", 123.4567);    // "     123.5"
String.Format("{0,-10:0.0}", 123.4567);   // "123.5     "
String.Format("{0,10:0.0}", -123.4567);   // "    -123.5"
String.Format("{0,-10:0.0}", -123.4567);  // "-123.5    "

Custom formatting for negative numbers and zero
String.Format("{0:0.00;minus 0.00;zero}", 123.4567);   // "123.46"
String.Format("{0:0.00;minus 0.00;zero}", -123.4567);  // "minus 123.46"
String.Format("{0:0.00;minus 0.00;zero}", 0.0);        // "zero"

Some funny examples
String.Format("{0:my number is 0.0}", 12.3);   // "my number is 12.3"
String.Format("{0:0aaa.bbb0}", 12.3);
Admin1

Another method, starting with the method:

double i = (10 * 0.69);
Console.Write(ToStringFull(i));       // Output 6.89999999999999946709294817
Console.Write(ToStringFull(-6.9)      // Output -6.90000000000000035527136788
Console.Write(ToStringFull(i - 6.9)); // Output -0.00000000000000088817841970012523233890533

A Drop-In Function...

public static string ToStringFull(double value)
{
    if (value == 0.0) return "0.0";
    if (double.IsNaN(value)) return "NaN";
    if (double.IsNegativeInfinity(value)) return "-Inf";
    if (double.IsPositiveInfinity(value)) return "+Inf";

    long bits = BitConverter.DoubleToInt64Bits(value);
    BigInteger mantissa = (bits & 0xfffffffffffffL) | 0x10000000000000L;
    int exp = (int)((bits >> 52) & 0x7ffL) - 1023;
    string sign = (value < 0) ? "-" : "";

    if (54 > exp)
    {
        double offset = (exp / 3.321928094887362358); //...or =Math.Log10(Math.Abs(value))
        BigInteger temp = mantissa * BigInteger.Pow(10, 26 - (int)offset) >> (52 - exp);
        string numberText = temp.ToString();
        int digitsNeeded = (int)((numberText[0] - '5') / 10.0 - offset);
        if (exp < 0)
            return sign + "0." + new string('0', digitsNeeded) + numberText;
        else
            return sign + numberText.Insert(1 - digitsNeeded, ".");
    }
    return sign + (mantissa >> (52 - exp)).ToString();
}

How it works

To solve this problem I used the BigInteger tools. Large values are simple as they just require left shifting the mantissa by the exponent. For small values we cannot just directly right shift as that would lose the precision bits. We must first give it some extra size by multiplying it by a 10^n and then do the right shifts. After that, we move over the decimal n places to the left. More text/code here.

user_23525070
Related