Java: Display float with precision of 30

Question

I'm trying to calculate PI with some algorithms, but the main thing is to display it with precision of 30. I tried to output a string with format etc. But it seems that maximum precision of double is 15. Is there any class or some method to help me with this ?

I've already tried:

public class Challenge6PI {

    public static void main(String[] args){

        System.out.format( "%.30f",
                Math.log(Math.pow(640320, 3) + 744) / Math.sqrt(163));
    }
}

Uhm, what are you doing with your `DecimalFormat`? – fge Jun 15 '13 at 13:03 — fge, Jun 15 '13 at 13:03
@fge Sorry it's remnant what I tried earlier. – ashur Jun 15 '13 at 13:04 — ashur, Jun 15 '13 at 13:04
You can try an alternative library, see my answer – fge Jun 15 '13 at 13:13 — fge, Jun 15 '13 at 13:13

fge · Answer 1 · 2013-06-15T13:23:13.357

3

While the JDK has BigDecimal, unfortunately you have no .sqrt() on it.

Which means your best bet is to use apfloat:

final Apfloat f1 = new Apfloat(640320);
final Apfloat f2 = ApfloatMath.pow(f1, 3L);
final Apfloat f3 = f2.add(new Apfloat(744));
final Apfloat f4 = ApfloatMath.sqrt(new Apfloat(163), 1000L);
final Apfloat ret = ApfloatMath.log(f3.divide(f4));

System.out.println(ApfloatMath.round(ret, 30L, RoundingMode.HALF_DOWN));

You have a shorter way for pi, though ;)

System.out.println(ApfloatMath.pi(30L));

edited Jun 15 '13 at 13:23

answered Jun 15 '13 at 13:13

fge

119,121
33
254
329

Thanks, I will try with `apfloat`, but I've also tried with `BigDecimal` which gave me theoretically nearly exact PI number with precision of 30. But some numbers are not equivalent when I compared it with http://3.141592653589793238462643383279502884197169399375105820974944592.com/index3141.html , what I've got is `3.141592652589793238462643383529`, which is strange because most of numbers are correct and some in the middle differ. – ashur Jun 16 '13 at 12:13

score 1 · Answer 2 · answered Jun 15 '13 at 13:02

1

Take a look at the BigDecimal class! http://docs.oracle.com/javase/6/docs/api/java/math/BigDecimal.html

answered Jun 15 '13 at 13:02

faffaffaff

3,429
16
27

score 1 · Answer 3 · answered Jun 15 '13 at 13:03

1

I guess Float does not supports that much precision. You may use BigDecimal for that.

answered Jun 15 '13 at 13:03

Ankit Zalani

3,068
5
27
47

score 0 · Answer 4 · answered Jun 15 '13 at 13:25

0

double doesn't have that much precision. Have a look at how the double type is implemented. For arbitrary precision have a look at BigDecimal

For example:

BigDecimal bd = new BigDecimal(22.0/7.0);
df.setMaximumFractionDigits(30);
System.out.println(df.format(bd));

answered Jun 15 '13 at 13:25

squib

1

2

Do NOT build a `BigDecimal` from doubles! You will lose precision before you even _build_ it – fge Jun 15 '13 at 13:28
Good point fge, thanks. Luckily 22.0/7.0 is a poor approximation of pi. – squib Jun 15 '13 at 22:01

Java: Display float with precision of 30

4 Answers4