floating point numbers in java

spyghost · 09-06-2003, 10:07 AM

hi,

i know this is a very stupid newbie question, but i would like to ask it... (sorry for my dumbness)

how would i extract the decimal part of a floating point and how would i extract the whole number part from a floating point?

ex: 735.15 - i would like to assign 735 to one variable and 0.15 to another.

thanks

Looking_Lost · 09-06-2003, 11:35 AM

This is a quick put together which doesn't do all the validation it really should but for now..

Code:


public class test
{
 public static void main(String args[])
 {

  double number=735.15;
  int intPart=0; 
  int decimalPoint=0;
  double decimalPart=0;

  StringBuffer stringNum=new StringBuffer(Double.toString(number));

  decimalPoint=stringNum.indexOf(".");

  intPart=Integer.parseInt(stringNum.substring(0,decimalPoint));
  
  decimalPart=Double.parseDouble(stringNum.substring(decimalPoint,stringNum.length()));

  
  System.out.println("Whole Number: " + intPart + "\n" + "Decimal Part: " + decimalPart);


 }
}

kev82 · 09-06-2003, 12:13 PM

i dont know java so this is probably wrong but if you can typecast then i would do this

Code:

float num=12.345;
float fPart;
int iPart;

iPart = (int)num;
fPart = num - (float)iPart;

Looking_Lost · 09-06-2003, 12:34 PM

I was think of that but thought the casting to int might end up rounding up the number in some cases.

kev82 · 09-06-2003, 12:53 PM

as i said i dont know java but in c/c++ casting to an int just takes the integer part it doesnt round.

Looking_Lost · 09-06-2003, 01:07 PM

Yeah, it does, once you reach the limit, although admittedly it is quite a few decimal places, 0.9999999999999999 or whatever amount, I guess it just depends on what they're doing or whatever size of numbers they'll be dealing with, either way, they've got two suggestions now so....

kev82 · 09-06-2003, 04:53 PM

ive just tried myself and have written the following to test if typecasting rounds because i was pretty sure it didnt, anyway i wrote the following program

Code:

import java.util.*;
import java.lang.Math.*;

class rounding
{
    public static void main(String[] args)
    {
	for(int p=0;p<25;p++) {
	    double num = (double)10 - Math.pow(10, -p);
	    int iPart = (int)num;
	    double fPart = num - (double)iPart;

	    System.out.println("num: "+num+"\tip: "+iPart+"\tfp: "+fPart);
	}
    }
}

and got this as my output

Code:

num: 9.0        ip: 9   fp: 0.0
num: 9.9        ip: 9   fp: 0.9000000000000004
num: 9.99       ip: 9   fp: 0.9900000000000002
num: 9.999      ip: 9   fp: 0.9990000000000006
num: 9.9999     ip: 9   fp: 0.9999000000000002
num: 9.99999    ip: 9   fp: 0.9999900000000004
num: 9.999999   ip: 9   fp: 0.9999990000000007
num: 9.9999999  ip: 9   fp: 0.9999999000000006
num: 9.99999999 ip: 9   fp: 0.9999999899999992
num: 9.999999999        ip: 9   fp: 0.9999999989999999
num: 9.9999999999       ip: 9   fp: 0.9999999999
num: 9.99999999999      ip: 9   fp: 0.9999999999900009
num: 9.999999999999     ip: 9   fp: 0.9999999999989999
num: 9.9999999999999    ip: 9   fp: 0.9999999999999005
num: 9.99999999999999   ip: 9   fp: 0.9999999999999893
num: 9.999999999999998  ip: 9   fp: 0.9999999999999982
num: 10.0       ip: 10  fp: 0.0
num: 10.0       ip: 10  fp: 0.0
num: 10.0       ip: 10  fp: 0.0
num: 10.0       ip: 10  fp: 0.0
num: 10.0       ip: 10  fp: 0.0
num: 10.0       ip: 10  fp: 0.0
num: 10.0       ip: 10  fp: 0.0
num: 10.0       ip: 10  fp: 0.0
num: 10.0       ip: 10  fp: 0.0

as you can see iPart is always the integer part of num, it doesnt round, however 9.99999... does eventually round to 10.0, i think this is what led you to your conclusion that typecasting to an int rounds but as you can see it doesnt.

Looking_Lost · 09-06-2003, 05:42 PM

I'm the one that needs the help now

So what's happening is when the type reaches it's precision limit for it's size it's been rounded up because of that not because of casting?

Goes to show appearances can be deceptive and looking again my overblown way would have suffered from the exact thing it was trying to prevent because it parsed back into a number.

Another little bit of insight gleemed from the LQ boards

kev82 · 09-06-2003, 06:21 PM

Quote:

I'm the one that needs the help now

everbody needs help sometimes, took me 3 minutes to write that program then over an hour to compile/run it

Quote:

So what's happening is when the type reaches it's precision limit for it's size it's been rounded up because of that not because of casting?

yep

Quote:

looking again my overblown way would have suffered from the exact thing it was trying to prevent because it parsed back into a number.

your method would suffer like mine but it would be due to the fact that number would be rounded up at the point you set it, which is obviosly before you convert it into a string. however your method is slightly more accurate than mine because i subtract and hence introduce numerical errors, you method has no loss of accuracy at all.

Quote:

Another little bit of insight gleemed from the LQ boards

thats why we're all here.