| 1 |
|
/* |
| 2 |
|
* %W% %E% |
| 3 |
|
* |
| 4 |
< |
* Copyright 2004 Sun Microsystems, Inc. All rights reserved. |
| 4 |
> |
* Copyright 2005 Sun Microsystems, Inc. All rights reserved. |
| 5 |
|
* SUN PROPRIETARY/CONFIDENTIAL. Use is subject to license terms. |
| 6 |
|
*/ |
| 7 |
|
|
| 10 |
|
import java.util.concurrent.atomic.AtomicLong; |
| 11 |
|
|
| 12 |
|
/** |
| 13 |
< |
* An instance of this class is used to generate a stream of |
| 14 |
< |
* pseudorandom numbers. The class uses a 48-bit seed, which is |
| 15 |
< |
* modified using a linear congruential formula. (See Donald Knuth, |
| 16 |
< |
* <i>The Art of Computer Programming, Volume 2</i>, Section 3.2.1.) |
| 13 |
> |
* An instance of this class is used to generate a stream of |
| 14 |
> |
* pseudorandom numbers. The class uses a 48-bit seed, which is |
| 15 |
> |
* modified using a linear congruential formula. (See Donald Knuth, |
| 16 |
> |
* <i>The Art of Computer Programming, Volume 2</i>, Section 3.2.1.) |
| 17 |
|
* <p> |
| 18 |
< |
* If two instances of <code>Random</code> are created with the same |
| 19 |
< |
* seed, and the same sequence of method calls is made for each, they |
| 20 |
< |
* will generate and return identical sequences of numbers. In order to |
| 21 |
< |
* guarantee this property, particular algorithms are specified for the |
| 22 |
< |
* class <tt>Random</tt>. Java implementations must use all the algorithms |
| 23 |
< |
* shown here for the class <tt>Random</tt>, for the sake of absolute |
| 24 |
< |
* portability of Java code. However, subclasses of class <tt>Random</tt> |
| 25 |
< |
* are permitted to use other algorithms, so long as they adhere to the |
| 18 |
> |
* If two instances of <code>Random</code> are created with the same |
| 19 |
> |
* seed, and the same sequence of method calls is made for each, they |
| 20 |
> |
* will generate and return identical sequences of numbers. In order to |
| 21 |
> |
* guarantee this property, particular algorithms are specified for the |
| 22 |
> |
* class <tt>Random</tt>. Java implementations must use all the algorithms |
| 23 |
> |
* shown here for the class <tt>Random</tt>, for the sake of absolute |
| 24 |
> |
* portability of Java code. However, subclasses of class <tt>Random</tt> |
| 25 |
> |
* are permitted to use other algorithms, so long as they adhere to the |
| 26 |
|
* general contracts for all the methods. |
| 27 |
|
* <p> |
| 28 |
< |
* The algorithms implemented by class <tt>Random</tt> use a |
| 29 |
< |
* <tt>protected</tt> utility method that on each invocation can supply |
| 28 |
> |
* The algorithms implemented by class <tt>Random</tt> use a |
| 29 |
> |
* <tt>protected</tt> utility method that on each invocation can supply |
| 30 |
|
* up to 32 pseudorandomly generated bits. |
| 31 |
|
* <p> |
| 32 |
< |
* Many applications will find the <code>random</code> method in |
| 32 |
> |
* Many applications will find the <code>random</code> method in |
| 33 |
|
* class <code>Math</code> simpler to use. |
| 34 |
|
* |
| 35 |
|
* @author Frank Yellin |
| 63 |
|
public Random() { this(++seedUniquifier + System.nanoTime()); } |
| 64 |
|
private static volatile long seedUniquifier = 8682522807148012L; |
| 65 |
|
|
| 66 |
< |
/** |
| 67 |
< |
* Creates a new random number generator using a single |
| 66 |
> |
/** |
| 67 |
> |
* Creates a new random number generator using a single |
| 68 |
|
* <code>long</code> seed: |
| 69 |
|
* <blockquote><pre> |
| 70 |
|
* public Random(long seed) { setSeed(seed); }</pre></blockquote> |
| 71 |
< |
* Used by method <tt>next</tt> to hold |
| 71 |
> |
* Used by method <tt>next</tt> to hold |
| 72 |
|
* the state of the pseudorandom number generator. |
| 73 |
|
* |
| 74 |
|
* @param seed the initial seed. |
| 80 |
|
} |
| 81 |
|
|
| 82 |
|
/** |
| 83 |
< |
* Sets the seed of this random number generator using a single |
| 84 |
< |
* <code>long</code> seed. The general contract of <tt>setSeed</tt> |
| 85 |
< |
* is that it alters the state of this random number generator |
| 86 |
< |
* object so as to be in exactly the same state as if it had just |
| 87 |
< |
* been created with the argument <tt>seed</tt> as a seed. The method |
| 88 |
< |
* <tt>setSeed</tt> is implemented by class Random as follows: |
| 89 |
< |
* <blockquote><pre> |
| 90 |
< |
* synchronized public void setSeed(long seed) { |
| 91 |
< |
* this.seed = (seed ^ 0x5DEECE66DL) & ((1L << 48) - 1); |
| 92 |
< |
* haveNextNextGaussian = false; |
| 93 |
< |
* }</pre></blockquote> |
| 94 |
< |
* The implementation of <tt>setSeed</tt> by class <tt>Random</tt> |
| 95 |
< |
* happens to use only 48 bits of the given seed. In general, however, |
| 96 |
< |
* an overriding method may use all 64 bits of the long argument |
| 97 |
< |
* as a seed value. |
| 98 |
< |
* |
| 99 |
< |
* Note: Although the seed value is an AtomicLong, this method |
| 100 |
< |
* must still be synchronized to ensure correct semantics |
| 101 |
< |
* of haveNextNextGaussian. |
| 83 |
> |
* Sets the seed of this random number generator using a single |
| 84 |
> |
* <code>long</code> seed. The general contract of |
| 85 |
> |
* <tt>setSeed</tt> is that it alters the state of this random |
| 86 |
> |
* number generator object so as to be in exactly the same state |
| 87 |
> |
* as if it had just been created with the argument <tt>seed</tt> |
| 88 |
> |
* as a seed. The method <tt>setSeed</tt> is implemented by class |
| 89 |
> |
* Random using a thread-safe update of the seed to <code> (seed * |
| 90 |
> |
* 0x5DEECE66DL + 0xBL) & ((1L << 48) - 1)</code> and clearing the |
| 91 |
> |
* <code>haveNextNextGaussian</code> flag used by {@link |
| 92 |
> |
* #nextGaussian}. The implementation of <tt>setSeed</tt> by class |
| 93 |
> |
* <tt>Random</tt> happens to use only 48 bits of the given |
| 94 |
> |
* seed. In general, however, an overriding method may use all 64 |
| 95 |
> |
* bits of the long argument as a seed value. |
| 96 |
|
* |
| 97 |
|
* @param seed the initial seed. |
| 98 |
|
*/ |
| 104 |
|
|
| 105 |
|
/** |
| 106 |
|
* Generates the next pseudorandom number. Subclass should |
| 107 |
< |
* override this, as this is used by all other methods.<p> |
| 108 |
< |
* The general contract of <tt>next</tt> is that it returns an |
| 109 |
< |
* <tt>int</tt> value and if the argument bits is between <tt>1</tt> |
| 110 |
< |
* and <tt>32</tt> (inclusive), then that many low-order bits of the |
| 111 |
< |
* returned value will be (approximately) independently chosen bit |
| 112 |
< |
* values, each of which is (approximately) equally likely to be |
| 113 |
< |
* <tt>0</tt> or <tt>1</tt>. The method <tt>next</tt> is implemented |
| 114 |
< |
* by class <tt>Random</tt> as follows: |
| 115 |
< |
* <blockquote><pre> |
| 116 |
< |
* synchronized protected int next(int bits) { |
| 117 |
< |
* seed = (seed * 0x5DEECE66DL + 0xBL) & ((1L << 48) - 1); |
| 118 |
< |
* return (int)(seed >>> (48 - bits)); |
| 119 |
< |
* }</pre></blockquote> |
| 120 |
< |
* This is a linear congruential pseudorandom number generator, as |
| 127 |
< |
* defined by D. H. Lehmer and described by Donald E. Knuth in <i>The |
| 128 |
< |
* Art of Computer Programming,</i> Volume 2: <i>Seminumerical |
| 107 |
> |
* override this, as this is used by all other methods.<p> The |
| 108 |
> |
* general contract of <tt>next</tt> is that it returns an |
| 109 |
> |
* <tt>int</tt> value and if the argument bits is between |
| 110 |
> |
* <tt>1</tt> and <tt>32</tt> (inclusive), then that many |
| 111 |
> |
* low-order bits of the returned value will be (approximately) |
| 112 |
> |
* independently chosen bit values, each of which is |
| 113 |
> |
* (approximately) equally likely to be <tt>0</tt> or |
| 114 |
> |
* <tt>1</tt>. The method <tt>next</tt> is implemented by class |
| 115 |
> |
* <tt>Random</tt> using a thread-safe update of the seed to <code> |
| 116 |
> |
* (seed * 0x5DEECE66DL + 0xBL) & ((1L << 48) - 1)</code> and |
| 117 |
> |
* returning <code>(int)(seed >>> (48 - bits))</code>. This is a |
| 118 |
> |
* linear congruential pseudorandom number generator, as defined |
| 119 |
> |
* by D. H. Lehmer and described by Donald E. Knuth in <i>The Art |
| 120 |
> |
* of Computer Programming,</i> Volume 2: <i>Seminumerical |
| 121 |
|
* Algorithms</i>, section 3.2.1. |
| 122 |
|
* |
| 123 |
|
* @param bits random bits |
| 128 |
|
long oldseed, nextseed; |
| 129 |
|
AtomicLong seed = this.seed; |
| 130 |
|
do { |
| 131 |
< |
oldseed = seed.get(); |
| 132 |
< |
nextseed = (oldseed * multiplier + addend) & mask; |
| 131 |
> |
oldseed = seed.get(); |
| 132 |
> |
nextseed = (oldseed * multiplier + addend) & mask; |
| 133 |
|
} while (!seed.compareAndSet(oldseed, nextseed)); |
| 134 |
|
return (int)(nextseed >>> (48 - bits)); |
| 135 |
|
} |
| 138 |
|
private static final int BYTES_PER_INT = 4; |
| 139 |
|
|
| 140 |
|
/** |
| 141 |
< |
* Generates random bytes and places them into a user-supplied |
| 142 |
< |
* byte array. The number of random bytes produced is equal to |
| 141 |
> |
* Generates random bytes and places them into a user-supplied |
| 142 |
> |
* byte array. The number of random bytes produced is equal to |
| 143 |
|
* the length of the byte array. |
| 144 |
< |
* |
| 145 |
< |
* @param bytes the non-null byte array in which to put the |
| 144 |
> |
* |
| 145 |
> |
* @param bytes the non-null byte array in which to put the |
| 146 |
|
* random bytes. |
| 147 |
|
* @since JDK1.1 |
| 148 |
|
*/ |
| 165 |
|
|
| 166 |
|
/** |
| 167 |
|
* Returns the next pseudorandom, uniformly distributed <code>int</code> |
| 168 |
< |
* value from this random number generator's sequence. The general |
| 169 |
< |
* contract of <tt>nextInt</tt> is that one <tt>int</tt> value is |
| 168 |
> |
* value from this random number generator's sequence. The general |
| 169 |
> |
* contract of <tt>nextInt</tt> is that one <tt>int</tt> value is |
| 170 |
|
* pseudorandomly generated and returned. All 2<font size="-1"><sup>32 |
| 171 |
< |
* </sup></font> possible <tt>int</tt> values are produced with |
| 172 |
< |
* (approximately) equal probability. The method <tt>nextInt</tt> is |
| 171 |
> |
* </sup></font> possible <tt>int</tt> values are produced with |
| 172 |
> |
* (approximately) equal probability. The method <tt>nextInt</tt> is |
| 173 |
|
* implemented by class <tt>Random</tt> as follows: |
| 174 |
|
* <blockquote><pre> |
| 175 |
|
* public int nextInt() { return next(32); }</pre></blockquote> |
| 205 |
|
* } |
| 206 |
|
* </pre></blockquote> |
| 207 |
|
* <p> |
| 208 |
< |
* The hedge "approximately" is used in the foregoing description only |
| 208 |
> |
* The hedge "approximately" is used in the foregoing description only |
| 209 |
|
* because the next method is only approximately an unbiased source of |
| 210 |
< |
* independently chosen bits. If it were a perfect source of randomly |
| 211 |
< |
* chosen bits, then the algorithm shown would choose <tt>int</tt> |
| 210 |
> |
* independently chosen bits. If it were a perfect source of randomly |
| 211 |
> |
* chosen bits, then the algorithm shown would choose <tt>int</tt> |
| 212 |
|
* values from the stated range with perfect uniformity. |
| 213 |
|
* <p> |
| 214 |
|
* The algorithm is slightly tricky. It rejects values that would result |
| 252 |
|
|
| 253 |
|
/** |
| 254 |
|
* Returns the next pseudorandom, uniformly distributed <code>long</code> |
| 255 |
< |
* value from this random number generator's sequence. The general |
| 256 |
< |
* contract of <tt>nextLong</tt> is that one long value is pseudorandomly |
| 257 |
< |
* generated and returned. All 2<font size="-1"><sup>64</sup></font> |
| 258 |
< |
* possible <tt>long</tt> values are produced with (approximately) equal |
| 259 |
< |
* probability. The method <tt>nextLong</tt> is implemented by class |
| 255 |
> |
* value from this random number generator's sequence. The general |
| 256 |
> |
* contract of <tt>nextLong</tt> is that one long value is pseudorandomly |
| 257 |
> |
* generated and returned. All 2<font size="-1"><sup>64</sup></font> |
| 258 |
> |
* possible <tt>long</tt> values are produced with (approximately) equal |
| 259 |
> |
* probability. The method <tt>nextLong</tt> is implemented by class |
| 260 |
|
* <tt>Random</tt> as follows: |
| 261 |
|
* <blockquote><pre> |
| 262 |
|
* public long nextLong() { |
| 293 |
|
* Returns the next pseudorandom, uniformly distributed <code>float</code> |
| 294 |
|
* value between <code>0.0</code> and <code>1.0</code> from this random |
| 295 |
|
* number generator's sequence. <p> |
| 296 |
< |
* The general contract of <tt>nextFloat</tt> is that one <tt>float</tt> |
| 297 |
< |
* value, chosen (approximately) uniformly from the range <tt>0.0f</tt> |
| 296 |
> |
* The general contract of <tt>nextFloat</tt> is that one <tt>float</tt> |
| 297 |
> |
* value, chosen (approximately) uniformly from the range <tt>0.0f</tt> |
| 298 |
|
* (inclusive) to <tt>1.0f</tt> (exclusive), is pseudorandomly |
| 299 |
< |
* generated and returned. All 2<font size="-1"><sup>24</sup></font> |
| 300 |
< |
* possible <tt>float</tt> values of the form |
| 301 |
< |
* <i>m x </i>2<font size="-1"><sup>-24</sup></font>, where |
| 299 |
> |
* generated and returned. All 2<font size="-1"><sup>24</sup></font> |
| 300 |
> |
* possible <tt>float</tt> values of the form |
| 301 |
> |
* <i>m x </i>2<font size="-1"><sup>-24</sup></font>, where |
| 302 |
|
* <i>m</i> is a positive integer less than 2<font size="-1"><sup>24</sup> |
| 303 |
< |
* </font>, are produced with (approximately) equal probability. The |
| 304 |
< |
* method <tt>nextFloat</tt> is implemented by class <tt>Random</tt> as |
| 303 |
> |
* </font>, are produced with (approximately) equal probability. The |
| 304 |
> |
* method <tt>nextFloat</tt> is implemented by class <tt>Random</tt> as |
| 305 |
|
* follows: |
| 306 |
|
* <blockquote><pre> |
| 307 |
|
* public float nextFloat() { |
| 308 |
|
* return next(24) / ((float)(1 << 24)); |
| 309 |
|
* }</pre></blockquote> |
| 310 |
< |
* The hedge "approximately" is used in the foregoing description only |
| 311 |
< |
* because the next method is only approximately an unbiased source of |
| 312 |
< |
* independently chosen bits. If it were a perfect source or randomly |
| 313 |
< |
* chosen bits, then the algorithm shown would choose <tt>float</tt> |
| 310 |
> |
* The hedge "approximately" is used in the foregoing description only |
| 311 |
> |
* because the next method is only approximately an unbiased source of |
| 312 |
> |
* independently chosen bits. If it were a perfect source or randomly |
| 313 |
> |
* chosen bits, then the algorithm shown would choose <tt>float</tt> |
| 314 |
|
* values from the stated range with perfect uniformity.<p> |
| 315 |
|
* [In early versions of Java, the result was incorrectly calculated as: |
| 316 |
|
* <blockquote><pre> |
| 317 |
|
* return next(30) / ((float)(1 << 30));</pre></blockquote> |
| 318 |
< |
* This might seem to be equivalent, if not better, but in fact it |
| 319 |
< |
* introduced a slight nonuniformity because of the bias in the rounding |
| 320 |
< |
* of floating-point numbers: it was slightly more likely that the |
| 321 |
< |
* low-order bit of the significand would be 0 than that it would be 1.] |
| 318 |
> |
* This might seem to be equivalent, if not better, but in fact it |
| 319 |
> |
* introduced a slight nonuniformity because of the bias in the rounding |
| 320 |
> |
* of floating-point numbers: it was slightly more likely that the |
| 321 |
> |
* low-order bit of the significand would be 0 than that it would be 1.] |
| 322 |
|
* |
| 323 |
|
* @return the next pseudorandom, uniformly distributed <code>float</code> |
| 324 |
|
* value between <code>0.0</code> and <code>1.0</code> from this |
| 330 |
|
} |
| 331 |
|
|
| 332 |
|
/** |
| 333 |
< |
* Returns the next pseudorandom, uniformly distributed |
| 333 |
> |
* Returns the next pseudorandom, uniformly distributed |
| 334 |
|
* <code>double</code> value between <code>0.0</code> and |
| 335 |
|
* <code>1.0</code> from this random number generator's sequence. <p> |
| 336 |
< |
* The general contract of <tt>nextDouble</tt> is that one |
| 337 |
< |
* <tt>double</tt> value, chosen (approximately) uniformly from the |
| 338 |
< |
* range <tt>0.0d</tt> (inclusive) to <tt>1.0d</tt> (exclusive), is |
| 339 |
< |
* pseudorandomly generated and returned. All |
| 340 |
< |
* 2<font size="-1"><sup>53</sup></font> possible <tt>float</tt> |
| 336 |
> |
* The general contract of <tt>nextDouble</tt> is that one |
| 337 |
> |
* <tt>double</tt> value, chosen (approximately) uniformly from the |
| 338 |
> |
* range <tt>0.0d</tt> (inclusive) to <tt>1.0d</tt> (exclusive), is |
| 339 |
> |
* pseudorandomly generated and returned. All |
| 340 |
> |
* 2<font size="-1"><sup>53</sup></font> possible <tt>float</tt> |
| 341 |
|
* values of the form <i>m x </i>2<font size="-1"><sup>-53</sup> |
| 342 |
< |
* </font>, where <i>m</i> is a positive integer less than |
| 343 |
< |
* 2<font size="-1"><sup>53</sup></font>, are produced with |
| 344 |
< |
* (approximately) equal probability. The method <tt>nextDouble</tt> is |
| 342 |
> |
* </font>, where <i>m</i> is a positive integer less than |
| 343 |
> |
* 2<font size="-1"><sup>53</sup></font>, are produced with |
| 344 |
> |
* (approximately) equal probability. The method <tt>nextDouble</tt> is |
| 345 |
|
* implemented by class <tt>Random</tt> as follows: |
| 346 |
|
* <blockquote><pre> |
| 347 |
|
* public double nextDouble() { |
| 348 |
|
* return (((long)next(26) << 27) + next(27)) |
| 349 |
|
* / (double)(1L << 53); |
| 350 |
|
* }</pre></blockquote><p> |
| 351 |
< |
* The hedge "approximately" is used in the foregoing description only |
| 352 |
< |
* because the <tt>next</tt> method is only approximately an unbiased |
| 353 |
< |
* source of independently chosen bits. If it were a perfect source or |
| 354 |
< |
* randomly chosen bits, then the algorithm shown would choose |
| 355 |
< |
* <tt>double</tt> values from the stated range with perfect uniformity. |
| 351 |
> |
* The hedge "approximately" is used in the foregoing description only |
| 352 |
> |
* because the <tt>next</tt> method is only approximately an unbiased |
| 353 |
> |
* source of independently chosen bits. If it were a perfect source or |
| 354 |
> |
* randomly chosen bits, then the algorithm shown would choose |
| 355 |
> |
* <tt>double</tt> values from the stated range with perfect uniformity. |
| 356 |
|
* <p>[In early versions of Java, the result was incorrectly calculated as: |
| 357 |
|
* <blockquote><pre> |
| 358 |
|
* return (((long)next(27) << 27) + next(27)) |
| 359 |
|
* / (double)(1L << 54);</pre></blockquote> |
| 360 |
< |
* This might seem to be equivalent, if not better, but in fact it |
| 361 |
< |
* introduced a large nonuniformity because of the bias in the rounding |
| 362 |
< |
* of floating-point numbers: it was three times as likely that the |
| 360 |
> |
* This might seem to be equivalent, if not better, but in fact it |
| 361 |
> |
* introduced a large nonuniformity because of the bias in the rounding |
| 362 |
> |
* of floating-point numbers: it was three times as likely that the |
| 363 |
|
* low-order bit of the significand would be 0 than that it would be |
| 364 |
< |
* 1! This nonuniformity probably doesn't matter much in practice, but |
| 365 |
< |
* we strive for perfection.] |
| 364 |
> |
* 1! This nonuniformity probably doesn't matter much in practice, but |
| 365 |
> |
* we strive for perfection.] |
| 366 |
|
* |
| 367 |
< |
* @return the next pseudorandom, uniformly distributed |
| 367 |
> |
* @return the next pseudorandom, uniformly distributed |
| 368 |
|
* <code>double</code> value between <code>0.0</code> and |
| 369 |
|
* <code>1.0</code> from this random number generator's sequence. |
| 370 |
|
*/ |
| 381 |
|
* <code>double</code> value with mean <code>0.0</code> and standard |
| 382 |
|
* deviation <code>1.0</code> from this random number generator's sequence. |
| 383 |
|
* <p> |
| 384 |
< |
* The general contract of <tt>nextGaussian</tt> is that one |
| 385 |
< |
* <tt>double</tt> value, chosen from (approximately) the usual |
| 386 |
< |
* normal distribution with mean <tt>0.0</tt> and standard deviation |
| 387 |
< |
* <tt>1.0</tt>, is pseudorandomly generated and returned. The method |
| 388 |
< |
* <tt>nextGaussian</tt> is implemented by class <tt>Random</tt> as follows: |
| 384 |
> |
* The general contract of <tt>nextGaussian</tt> is that one |
| 385 |
> |
* <tt>double</tt> value, chosen from (approximately) the usual |
| 386 |
> |
* normal distribution with mean <tt>0.0</tt> and standard deviation |
| 387 |
> |
* <tt>1.0</tt>, is pseudorandomly generated and returned. The method |
| 388 |
> |
* <tt>nextGaussian</tt> is implemented by class <tt>Random</tt> as if |
| 389 |
> |
* by a threadsafe version of the following: |
| 390 |
|
* <blockquote><pre> |
| 391 |
< |
* synchronized public double nextGaussian() { |
| 391 |
> |
* public double nextGaussian() { |
| 392 |
|
* if (haveNextNextGaussian) { |
| 393 |
|
* haveNextNextGaussian = false; |
| 394 |
|
* return nextNextGaussian; |
| 395 |
|
* } else { |
| 396 |
|
* double v1, v2, s; |
| 397 |
< |
* do { |
| 397 |
> |
* do { |
| 398 |
|
* v1 = 2 * nextDouble() - 1; // between -1.0 and 1.0 |
| 399 |
|
* v2 = 2 * nextDouble() - 1; // between -1.0 and 1.0 |
| 400 |
|
* s = v1 * v1 + v2 * v2; |
| 401 |
|
* } while (s >= 1 || s == 0); |
| 402 |
< |
* double multiplier = Math.sqrt(-2 * Math.log(s)/s); |
| 402 |
> |
* double multiplier = StrictMath.sqrt(-2 * StrictMath.log(s)/s); |
| 403 |
|
* nextNextGaussian = v2 * multiplier; |
| 404 |
|
* haveNextNextGaussian = true; |
| 405 |
|
* return v1 * multiplier; |
| 406 |
|
* } |
| 407 |
|
* }</pre></blockquote> |
| 408 |
< |
* This uses the <i>polar method</i> of G. E. P. Box, M. E. Muller, and |
| 409 |
< |
* G. Marsaglia, as described by Donald E. Knuth in <i>The Art of |
| 410 |
< |
* Computer Programming</i>, Volume 2: <i>Seminumerical Algorithms</i>, |
| 408 |
> |
* This uses the <i>polar method</i> of G. E. P. Box, M. E. Muller, and |
| 409 |
> |
* G. Marsaglia, as described by Donald E. Knuth in <i>The Art of |
| 410 |
> |
* Computer Programming</i>, Volume 2: <i>Seminumerical Algorithms</i>, |
| 411 |
|
* section 3.4.1, subsection C, algorithm P. Note that it generates two |
| 412 |
< |
* independent values at the cost of only one call to <tt>Math.log</tt> |
| 413 |
< |
* and one call to <tt>Math.sqrt</tt>. |
| 412 |
> |
* independent values at the cost of only one call to <tt>StrictMath.log</tt> |
| 413 |
> |
* and one call to <tt>StrictMath.sqrt</tt>. |
| 414 |
|
* |
| 415 |
|
* @return the next pseudorandom, Gaussian ("normally") distributed |
| 416 |
|
* <code>double</code> value with mean <code>0.0</code> and |
| 424 |
|
return nextNextGaussian; |
| 425 |
|
} else { |
| 426 |
|
double v1, v2, s; |
| 427 |
< |
do { |
| 427 |
> |
do { |
| 428 |
|
v1 = 2 * nextDouble() - 1; // between -1 and 1 |
| 429 |
< |
v2 = 2 * nextDouble() - 1; // between -1 and 1 |
| 429 |
> |
v2 = 2 * nextDouble() - 1; // between -1 and 1 |
| 430 |
|
s = v1 * v1 + v2 * v2; |
| 431 |
|
} while (s >= 1 || s == 0); |
| 432 |
< |
double multiplier = Math.sqrt(-2 * Math.log(s)/s); |
| 432 |
> |
double multiplier = StrictMath.sqrt(-2 * StrictMath.log(s)/s); |
| 433 |
|
nextNextGaussian = v2 * multiplier; |
| 434 |
|
haveNextNextGaussian = true; |
| 435 |
|
return v1 * multiplier; |
| 491 |
|
|
| 492 |
|
} |
| 493 |
|
|
| 494 |
< |
} |
| 494 |
> |
} |
