math question (natural log - ln)

[Tom Clement]

Hi, working on a project which must calculate natural logarithms,
I see calculators do it all over, but my luck with source code evaluation hasn't turned up what I need.

Example:
ln 27 = 3.29583686600432907418573571076758

Must be a relatively easy way to do it, closest I came was the rounded 3:

with the following code


import java.math.*;
public class Test2 {


static double xDouble;

public static void main(String[] args) {

double d = 1.0;


BigInteger target = BigInteger.valueOf(27);

target = doLog(target,d);
System.out.println(target);
}

static public BigInteger doLog (BigInteger target, double divisor)
{

xDouble = logBig(target.abs())/divisor;

return BigInteger.valueOf((long)(xDouble));
}

static public double logBig(BigInteger z) {
// Needed because Math.log(z.doubleValue()) returns infinite for z bigger than MAX DOUBLE
BigInteger b;
int temp = z.bitLength() - 1000;
if (temp > 0) {
// b=z.divide(BigInteger.valueOf(2).pow(temp)); //replaced with below
b=z.shiftRight(temp);
return (Math.log(b.doubleValue()) + temp*Math.log(2));
}
else {return (Math.log(z.doubleValue()));}
}
}


Many thanks......................Tom
[ February 26, 2005: Message edited by: Tom Clement ]

[M Beck]

you're seeing your results rounded to the nearest integer value because you're storing them in a BigInteger variable. have you tried using BigDecimal instead?

[Tom Clement]

No, I haven't yet, I'll try it now, thanks

[Tom Clement]

tried BigDecimal, it's not a straightforward swap, it'll take more time than I
have tonight, I'll definitely try tomorrow. With all those calculators out there, I'm hoping someone's done this already.

[David Harkness]

Why can't you use java.lang.Math.log(double)? Since you're taking the double value of your BigInteger (and eventually BigDecimal), why bother with those classes at all?

[Layne Lund]

As their names imply, BigInteger and BigDecimal are meant to be used for arbitrarily large numbers. Unless, you are using numbers that are over 9*10^19 (the maximum value for a long), then you should probably just use the built-in types like double and long. The Math class from the standard API provides methods to perform just about any mathematical operation you need. These include the basic trigonometric functions and logarithms, to name a few.

HTH

Layne

[Ilja Preuss]

Somehow this discussion seems to be continued at http://www.coderanch.com/t/326805/java/java/natural-logarithms-ln

Closing this thread...