How do you reduce a java double to n significant digits?

Question

I am looking to print small digits (doubles) for the purpose of printing the errors in using the Newton & Secant methods.

One of my errors is 5.433306166802499E-5

I'd like to print 5.4333E-5

I thought of using BigDecimal but I am not familiar with this class.

You just want to format the number for printing. Don't mess around with the value (i.e. don't "reduce to n digits"). — Henry, Oct 20 '16 at 06:08
@ScaryWombat Disagree. He just wants to format a double for output. — Jim Garrison, Oct 20 '16 at 06:09
@JimGarrison Ok, looks like I put too much weight on the title. — Scary Wombat, Oct 20 '16 at 06:12

score 6 · Answer 1 · edited Oct 20 '16 at 06:10

6

I am looking to print small digits (doubles)

System.out.printf("%.4e",  5.433306166802499E-5);

Result: 5.4333e-05

Note: it doesn't reduce the precision of your original value, it just prints it with a lower precision.

edited Oct 20 '16 at 06:10

Jim Garrison

85,615
20
155
190

answered Oct 20 '16 at 06:06

Adrian Colomitchi

3,974
1
14
23

You left out the "e" format specifier; I added it to make your answer actually work. – Jim Garrison Oct 20 '16 at 06:10
@JimGarrison Thanks, I don't know how I lost the `e` – Adrian Colomitchi Oct 20 '16 at 06:11

score 2 · Answer 2 · answered Oct 20 '16 at 06:06

2

double d = 5.433306166802499E-5; BigDecimal dc = new BigDecimal(d); dc = dc .round(new MathContext(3)); // desired significant digits double rounded = dc .doubleValue();

answered Oct 20 '16 at 06:06

Ghost Developer

1,283
1
10
18

Lucurious · Answer 3 · 2016-10-20T06:38:41.700

1

You can indeed use BigDecimal, it would then look as follows:

BigDecimal d = BigDecimal.valueOf(val).setScale(scale, RoundingMode.HALF_UP);
double scaled = d.doubleValue();

Or you could use:

Math.round(val*Math.pow(10, scale))/Math.pow(10, scale);

edited Oct 20 '16 at 06:38

answered Oct 20 '16 at 06:18

Lucurious

174
6

First example as too many commas, and if `val = 5.433306166802499E-5` as given in question, result is `1.0E-4`, which is far from the desired output of `5.4333E-5`. – Andreas Oct 20 '16 at 06:37

Mourits de Beer · Answer 4 · 2022-02-07T07:38:14.430

-1

To answer the general question of reducing a double's significant digits, and not just the case of printing with System.out.printf:

If you're using Java: Double.parseDouble(String.format("%1.4e", 5.433306166802499E-5))

In Scala, you can use the f string interpolator, or the format method:

val a: Double = 5.433306166802499E-5
val reduced1: Double = f"$a%1.4e".toDouble
val reduced2: Double = "%1.4e".format(a).toDouble

Both are equal: 5.4333e-05

edited Feb 07 '22 at 07:38

answered Nov 13 '19 at 10:00

Mourits de Beer

79
1
6

The question asks about java but this answer is Kotlin. – Michael Feb 05 '22 at 00:13
1

My answer was for Scala. Not sure about the f string interpolator, but the `String.format` method should work for Java. – Mourits de Beer Feb 07 '22 at 07:03

How do you reduce a java double to n significant digits?

4 Answers4