Pi generating routine doesn't seem to converge.

[floitschG]

From the mailing-list:

Thanks for amazing work on dart, I'd like to report a issue on core
library math.

What steps will reproduce the problem ?

1. run below code by dartVM.

double pi() {
   final double COUNT = Math.pow(10,7);
   final double r = 1.0;
   int hit =0;

   for(int i=0; i<COUNT; i++) {
       double x = Math.random();
       double y = Math.random();

       if( Math.sqrt(Math.pow(x,2) + Math.pow(y,2)) <= r ) {
           hit ++;
       }
   }

   return ( 4.00000 * hit / COUNT) ;
}

main() {
   Stopwath watch = new Stopwatch();

   watch.start();
   print( pi() );
   watch.stop();

   print("time used: ${watch.elapsedInMs() }");
}
2. The performance also need to improved, about 10x slow than run on
V8.

What is the expected output? What do you see instead?

expected value is round 3.14 but get 2.9310668 in fact. If I compile
it to javascript by frogc, get 3.142088.

What version of the product are you using? On what operating system?
dartvm 4800

Please provide any additional information below.

http://www.juanhu.name/dartvm-performance.html

[floitschG]

My guess is that this is a duplicate of issue #499.

[DartBot]

This comment was originally written by danri...@gmail.com

---

The following program produces a plot of (x, y) pairs from Math.random. If you view the output in a postscript interpreter, you will not be surprised that the answer doesn't converge to PI. :-)

class Random {

  Random() {
  }

  void run() {
    print('''
%!PS

matrix currentmatrix /originmat exch def
/umatrix {originmat matrix concatmatrix setmatrix} def
[28.3465 0 0 28.3465 10.5 100.0] umatrix
20 20 scale
    
0 setgray
0 setlinewidth

0 0 newpath moveto
1 0 lineto
1 1 lineto
0 1 lineto
closepath
stroke

0.001 setlinewidth 1 setlinecap

/pt { 1 index 1 index moveto lineto stroke} def
''');
    for (int i = 0; i < 10000; i++) {
      double x = Math.random();
      double y = Math.random();
      print("$x $y pt");
    }
    print("showpage");
  }
}

[whesse]

The prng generates one random bit at a time, so if you generate about 30 random numbers after x, and after y, then you should get a reasonable answer.

---

Added Duplicate label.
Marked as being merged into #499.

[DartBot]

This comment was originally written by danri...@gmail.com

---

I've attached a screen shot of (x, y) pairs where 32 random numbers are discarded between selecting x and y. It's better, but still not really usable.

---

Attachment:
[Screen shot 2012-03-05 at 12.24.34 PM.png](https://storage.googleapis.com/google-code-attachments/dart/issue-1983/comment-4/Screen shot 2012-03-05 at 12.24.34 PM.png) (156.47 KB)

[DartBot]

This comment was originally written by danr...@gmail.com

---

Here is the screnshot for comment #­2.

---

Attachment:
[Screen shot 2012-03-05 at 1.37.30 PM.png](https://storage.googleapis.com/google-code-attachments/dart/issue-1983/comment-5/Screen shot 2012-03-05 at 1.37.30 PM.png) (34.06 KB)