C# high precision calculations

A

4

5

Consider this code:

double result = Math.Sqrt(4746073226998689451);

For result I get 2178548422 instead of 2178548421.999999854etc... How can I get more precise result?

Alanalana answered 13/11, 2011 at 18:23 Comment(0)

E

6

For the particular problem, computing the square root, you can use Decimal type and Newton's algorithm:

using System;

class Program
{
  public static void Main()
  {
    long x = 4746073226998689451;
    decimal sqrt_x = (decimal)Math.Sqrt(x);
    for (int i = 0; i < 10; ++i)
      sqrt_x = 0.5m * (sqrt_x + x / sqrt_x);
    Console.WriteLine("{0:F16}", sqrt_x);
  }
}

The result is:

2178548421.9999998547197773

Evitaevitable answered 13/11, 2011 at 18:35 Comment(0)

F

7

There is a bunch of high precision maths libraries for .NET mentioned on wikipedia - Arbitrary-percision artithmatic page.

I have seen BigNum recommended here before, though the wikipedia link is broken and I can't find the library elsewhere at the moment.

The other option on the page is the C# binding for MPIR.

Felten answered 13/11, 2011 at 18:27 Comment(1)

@GeorgeDuckett - So it is. Do you have a link to a working site? – Felten 13/11, 2011 at 18:33

E

6

For the particular problem, computing the square root, you can use Decimal type and Newton's algorithm:

using System;

class Program
{
  public static void Main()
  {
    long x = 4746073226998689451;
    decimal sqrt_x = (decimal)Math.Sqrt(x);
    for (int i = 0; i < 10; ++i)
      sqrt_x = 0.5m * (sqrt_x + x / sqrt_x);
    Console.WriteLine("{0:F16}", sqrt_x);
  }
}

The result is:

2178548421.9999998547197773

Evitaevitable answered 13/11, 2011 at 18:35 Comment(0)

D

1

Using digit by digit calculation will give you as many digits as you are looking for.

Deutzia answered 13/11, 2011 at 18:27 Comment(3)

That would be reinventing the wheel. – Margherita 13/11, 2011 at 18:33

@Margherita - Why? I am not suggesting he should code the algorithm himself. If there is an implementation, by all means, use it. – Deutzia 13/11, 2011 at 18:42

That's true, but there are many algorithms for arbitrary calculations, this is just one. I think it would've been more useful to link to an implementation. – Margherita 13/11, 2011 at 18:47

S

1

instead of double , try big integer and also check out this link

Big numbers with fraction support

Sporophyll answered 13/11, 2011 at 18:33 Comment(0)

Recommended topics

Hot tags