BigInteger FormatException

Question

In my program, I get a FormatException with the message The value could not be parsed on the following line:

c = BigInteger.Parse("" + Math.Pow(b, 3));

The exception occurs when b = 99000, but if I manually replace b with 99000, the exception doesn't occur. I have also tested it with b = 99001, and other higher values but I don't get the error.

Answer 1

The underlying cause is that Math.Pow operates on doubles, not integral data types. By default, Double.ToString() returns only 15 digits of precision, and the double data type can only store a maximum of 17 digits of precision internally.

When b=100,000, b ³ = 1,000,000,000,000,000 (16 digits), and the conversion to a string will return the result as 1E+15 . BigInteger.Parse can only parse integral values (optional whitespace, followed by a sign, followed by 1 or more numbers), and will throw a FormatException when the string value "1E+15" is provided as an argument.

A solution is to avoid Math.Pow entirely, and to instead use BigInteger.Pow(BigInteger, int) . This approach avoids conversions to double entirely, and can exponentiate arbitrarily large integers.

With this change, your code might resemble something like this:

BigInteger c;
for (BigInteger b = 1; b < 1000000; b++)
{
   c = BigInteger.Pow(b, 3);
   // c stores b^3 as a BigInteger
}

Answer 2

Weird ... This seems to work for me, and you were right about the string in parse.

BigInteger c;
long b = 9000;
c = BigInteger.Parse("" + Math.Pow(b, 3));

No errors thrown ...

As for codemonkeh comment about the ToString() , you could also do this

BigInteger d = new BigInteger(Math.Pow(b, 3));

Assuming you are not running in a loop that will create heaps of those ...