rounding double in java

Question

I have a double valued diagonal matrix stored in a text file.

size(file)~410 Mo

I would like to reduce the size by rounding my double values.

If its a good idea, how to do it in java

0.1706524958886193=>0.17

I need to use this file later in matlab

when i try

dlmread(file) i get out of memory error

as a file text (.txt)// i need to use it later in matlab program — nawara, Jun 06 '13 at 17:26
Just use e.g. [`printf`](http://docs.oracle.com/javase/1.5.0/docs/api/java/io/PrintStream.html#printf%28java.lang.String,%20java.lang.Object...%29) to format as required. — Oliver Charlesworth, Jun 06 '13 at 17:27
@nawara is it sufficient to reduce size...or rounding is must? — pinkpanther, Jun 06 '13 at 17:32
@pinkpanther ; my goal is to reduce size because i will use this file later in a matlab program — nawara, Jun 06 '13 at 17:33
have you tried [`Math.round`](http://docs.oracle.com/javase/6/docs/api/java/lang/Math.html#round%28double%29)? — Jacob, Jun 06 '13 at 17:41
@nawara my mistake, i forgot Java doesn't have a form that takes a second parameter. `Math.round(num * 100) / 100.0` will work, though. — Jacob, Jun 06 '13 at 18:08

Joni · Accepted Answer · 2013-06-06T18:03:19.017

5

If you round the values you are throwing away precision. That may change the results you will get.

A better approach is to store only the diagonal items. There is no point in storing the n^2-n off-diagonal zeroes. Use the diag function to convert a vector into a diagonal matrix. http://www.mathworks.es/es/help/matlab/ref/diag.html

Even more efficient: store the numbers in a binary format instead of text.

edited Jun 06 '13 at 18:03

answered Jun 06 '13 at 17:53

Joni

108,737
14
143
193

+1 because I agree that reducing to 2 decimals might alter the results. – Schorsch Jun 06 '13 at 18:22
great idea but ....the same problem with function diag =out of memory – nawara Jun 06 '13 at 19:48
Then the matrix is just too big to process with the memory currently assigned to Matlab. There's a pagewith information on how to give Matlab more memory and tips on how to reduce memory cconsumption here: http://www.mathworks.es/es/help/matlab/matlab_prog/resolving-out-of-memory-errors.html – Joni Jun 06 '13 at 20:29
yes thanks , i asked about this problem [here](http://stackoverflow.com/questions/16971111/out-of-memory-error-when-using-diag-function-in-matlab/16974023?noredirect=1#16974023): – nawara Jun 07 '13 at 08:43

pinkpanther · Answer 2 · 2013-06-06T17:48:21.977

2

Use DecimalFormat class to format the double value to your needs. For example, if you want to keep only 2 digits after decimal point use, "#0.00 and so on. DecimalFormat#format class returns String output which you can use to instantiate Double value. from

import java.text.DecimalFormat;
public class Tester
{
  public static void main(String arg[]) throws Throwable
  {
    double bigDouble=0.1706524958886193;
    DecimalFormat df=new DecimalFormat("#0.00");
    String numberString=df.format(bigDouble);
    double smallDouble=new Double(numberString);
    System.out.println(smallDouble);
  }
}

Hope this helps...

edited Jun 06 '13 at 17:48

answered Jun 06 '13 at 17:41

pinkpanther

4,770
2
38
62

+1 As it is the best solution to round numbers. – eternay Jun 06 '13 at 17:54
+1 and please to add implemented methods in [NumberFormat](http://stackoverflow.com/a/8582656/714968), otherwise you have to use BigDecimal – mKorbel Jun 06 '13 at 17:54
[true is](http://stackoverflow.com/a/7553827/714968) by @EJP – mKorbel Jun 06 '13 at 17:56
@mKorbel please to add? I didn't get you...you want' me to edit my answer or? – pinkpanther Jun 06 '13 at 18:04
not there isn't something from my linked answer, nor to edit something, do you have any issue with my comments or – mKorbel Jun 06 '13 at 18:41
@mKorbel `please to add implemented methods in NumberFormat, otherwise you have to use BigDecimal`... I didn't get that sentence...I don't have an issue :) – pinkpanther Jun 06 '13 at 18:48
@pinkpanther ignore that, doesn't matter, empty answer – mKorbel Jun 06 '13 at 18:51
exception `Exception in thread "main" java.lang.NumberFormatException: For input string: "0,17"` – nawara Jun 06 '13 at 19:48
@nawara use the methods from the NumberFormat class for convertion....note that DecimalFormat is a subclass of number format.. see this question for details http://stackoverflow.com/questions/888088/how-do-i-convert-a-string-to-double-in-java-using-a-specific-locale – pinkpanther Jun 07 '13 at 05:41

score 0 · Answer 3 · answered Jun 06 '13 at 17:43

0

Unless I'm misunderstanding the question you can do it like this:

double value = 0.1706524958886193;
value = (double)((int)(value*100))/100; // returns 0.17

answered Jun 06 '13 at 17:43

Josh Maggard

419
3
9

That would be truncation. The asker seems to want rounding – Jacob Jun 06 '13 at 17:44
Also, in most cases this will not produce an "exact" result... – Oliver Charlesworth Jun 06 '13 at 18:15

rounding double in java

3 Answers3