Hi Folks,

Well here's a program i wrote to check out how the Math.floor() and Math.ceil() works. I kinda grasped the idea of these two methods. However, hear me out...

/*************

This program is written with the spcific intention of redirection to a file instead of output to console


***************/

public class Whatzit
{
    public static void main(String[] args)
    {
        double i = 0.0;
        for(; i < 11.0; i+=0.1)
        {

            double ceil = Math.ceil(i);

            System.out.println("\ni = " + i + "\t\t ceil = " + ceil);

            double floor = Math.floor(i);

            System.out.println("i = " + i + "\t\t floor = " + floor);

            for(int count = 0; count < 51; count++)
            {
                System.out.print("-"); // make some sense by drawing a line
            } //end of for count loop

        } // end for i
    } // end of main()
} // end of prog.

after compiling i execute as follows:

c:\$JAVAHOME$\SOURCE>java -ea Whatzit >> Whatzit

the output now gets redirected to a file called Whatzit which i open using MS Word 2000 in UTF-8 format.

here's a portion of the output


i = 0.0 ceil = 0.0
i = 0.0 floor = 0.0
---------------------------------------------------
i = 0.1 ceil = 1.0
i = 0.1 floor = 0.0
---------------------------------------------------
i = 0.2 ceil = 1.0
i = 0.2 floor = 0.0
---------------------------------------------------
i = 0.30000000000000004 ceil = 1.0
i = 0.30000000000000004 floor = 0.0
---------------------------------------------------
i = 0.4 ceil = 1.0
i = 0.4 floor = 0.0
---------------------------------------------------
i = 0.5 ceil = 1.0
i = 0.5 floor = 0.0
---------------------------------------------------
i = 0.6 ceil = 1.0
i = 0.6 floor = 0.0
---------------------------------------------------
i = 0.7 ceil = 1.0
i = 0.7 floor = 0.0
---------------------------------------------------
i = 0.7999999999999999 ceil = 1.0
i = 0.7999999999999999 floor = 0.0
---------------------------------------------------
i = 0.8999999999999999 ceil = 1.0
i = 0.8999999999999999 floor = 0.0



---SNIP--- (file deliberately truncated by kaizer)

My qustion is this...

even though i've instructed to specifically increment i in steps of 0.1 WHY DOES THE VARIABLE I GET ROUNDED UP TO 0.4 FROM 0.30000000000000004? (See above). Infact, it happens again from 0.7999999999999999 and jumps to 0.8999999999999999!!

Is this an inherant anomaly with the SDK or is there something i'm missing here?

Thanks to y'all

And thanks for reading patiently!!

Have a nice day,
Kaiz.


Kaizer Billimoria
Bangalore, India

__________________
Kaizer Billimoria
Bangalore, India

Yeah, this is really confusing isn't it. The problem here, which exists in languages besides java, is that the results produced when using double and float are not always precise. If you have a look at IEEE specification 754, you will see that it states:

//quote

Why is 0.1 not 0.1?

Binary floating-point numbers consist of signed integers multiplied by powers of two. When fractional, you can also consider them as integers divided by powers of two.The decimal number 0.1, or 1/10, is not an integer over a power of two.

//end of quote

Often this isn't a problem, however if you are dealing with more precise calculations then this behaviour is not acceptable. Let me explain using this example that I've copied from the java.sun.com web site:

double f = 2.9 - 1.1;
System.out.println(f);

It outputs 1.799999999999998 rather than 1.8

To get around this problem you could always use java.math.BigDecimal.


import java.math.*;

class BigDecimalTest
{
    public static void main(String[] args)
    {
        BigDecimal a = new BigDecimal("2.9");
        BigDecimal b = new BigDecimal("1.1");
        BigDecimal c =a.subtract(b);
        System.out.println(c);
    }
}


This outputs 1.8


I hope this helps.

Cheers

Martyn

Hi Martyn,

Thanks for the insight!

Kaizer.

Kaizer Billimoria
Bangalore, India