How to print a number upto 4 decimal places without rounding off in java without using BigDecimal

Question

if input = 31.6227890 then output = 31.6227 but the output is coming as 31.6228

What type are you working with? `double`, or something else? Note that floating point arithmetic is not exact in Java. — Tim Biegeleisen, Jul 19 '18 at 06:42

score 4 · Answer 1 · edited May 20 '19 at 06:50

4

You can use below function:

import java.text.DecimalFormat;
import java.math.RoundingMode;

public static double formatValue(Double number) {
        DecimalFormat df = new DecimalFormat("####0.0000");
        return Double.parseDouble(df.format(number));
    }

Edit : you have to add below code for rounding off.

df.setRoundingMode(RoundingMode.DOWN);

Input = 31.6227890

Output = 31.6227

edited May 20 '19 at 06:50

Anubhav Singh

8,321
4
25
43

answered Jul 19 '18 at 06:44

Md. Sajedul Karim

6,749
3
61
87

Are you sure this *doesn't round* the result? – Peter Lawrey Jul 19 '18 at 06:45
1.) This rounds. 2.) Depending on your locale it will output the string being comma seperated over point seperated (`2,1234` not `2.1234`) which then again can't be parsed back. – Ben Jul 19 '18 at 06:46
thanks but it's return 31.6228 instead of 31.6227 with this approach – HMT Jul 19 '18 at 06:49
@PeterLawrey It rounds the result – HMT Jul 19 '18 at 06:49
can I truncate the value after rounding it upto 5 decimals – HMT Jul 19 '18 at 06:50
Sorry, I edited my answer for rounding off :) Thanks for your attention – Md. Sajedul Karim Jul 19 '18 at 06:53
But I don't want it to round off the number – HMT Jul 19 '18 at 06:56
Input = 31.6227890 OutPUT = 31.6227 – Md. Sajedul Karim Jul 19 '18 at 07:02

Peter Lawrey · Answer 2 · 2018-07-19T07:15:21.393

If you want to round down, you can use truncation like this.

double d = 31.6227890;
d = ((long) (d * 10000)) / 10000.0; // truncates down to 0
System.out.println(d);

prints

31.6227

Note: this might print less than 4 digits. If you always want 4 you need to use formatting.

d = 31.60007890;
d = ((long) (d * 10000)) / 10000.0; // truncates down to 0
System.out.println(d);
System.out.printf("%.4f%n", d);

prints

31.6
31.6000

NOTE: For large values you wouldn't round them as this would result in an overflow.

private static final double WHOLE_NUMBER = 1L << 53;
public static double trunc4(double d) {
    final double factor = 1e4;
    return Math.abs(d) >= WHOLE_NUMBER / factor 
           ? d 
           : ((long) (d * factor)) / factor;
}

The point at which you get an error is for numbers so large they are whole numbers anyway (due to the limits of the precision on double)

NOTE: If you used float the limit at which you would get a precision error is for much smaller values.

double d = 1024.00039999;
float f= (float) (long) (d * 10000) / 10000;
System.out.println(f);

prints

1024.0002

There's the usual fp inaccuracies to deal with it you want more decimal places. This will also not work correctly for large values of d (i.e. we use accuracy and risk overflows). For"normal" values this will be fine though — Voo, Jul 19 '18 at 06:50

score 1 · Answer 3 · answered Jul 19 '18 at 06:57

1

You can round down using a BigDecimal:

new BigDecimal(31.6227890d).setScale(4, RoundingMode.DOWN)

That returns the desired result: 31.6227.

For a double, you can use .doubleValue():

new BigDecimal(31.6227890d).setScale(4, RoundingMode.DOWN).doubleValue()

answered Jul 19 '18 at 06:57

ernest_k

44,416
5
53
99

Note `System.out.println(new BigDecimal(0.0003d).setScale(4, RoundingMode.DOWN));` prints `0.0002` I suspect `BigDecimal.valueOf(...)` would be better. – Peter Lawrey Jul 19 '18 at 07:20

score 0 · Accepted Answer · answered Jul 19 '18 at 06:50

0

float i=(float)((int) (31.6227890 *10000))/10000;

System.out.println(i);

answered Jul 19 '18 at 06:50

subbu

254
2
7

1

As float only has 6 digits of accuracy this won't work for "large" numbers around 100 or more – Peter Lawrey Jul 19 '18 at 07:02
can overflow the int as well – Ofer Skulsky Jul 19 '18 at 07:10
1

Correct, the smallest value this fails for is 1024.0003, see my answer – Peter Lawrey Jul 19 '18 at 07:13

score 0 · Answer 5 · answered Jul 19 '18 at 06:56

0

You can multiply it per 1000, use Math.floor() and divide it per 1000

    double a = 10.345665654;
    double b = Math.floor(a * 10000) / 10000;
    System.out.println(b);

answered Jul 19 '18 at 06:56

Kaostias

321
4
24

How to print a number upto 4 decimal places without rounding off in java without using BigDecimal

5 Answers5