commons-math/xdocs/userguide/random.xml

174 lines
9.7 KiB
XML

<?xml version="1.0"?>
<?xml-stylesheet type="text/xsl" href="./xdoc.xsl"?>
<!-- $Revision: 1.4 $ $Date: 2003/11/15 18:38:16 $ -->
<document url="random.html">
<properties>
<title>The Commons Math User Guide - Data Generation</title>
<author email="phil@steitz.com">Phil Steitz</author>
</properties>
<body>
<section name="2 Data Generation">
<subsection name="2.1 Overview" href="overview">
<p>
The Commons Math random package includes utilities for
<ul>
<li>generating random numbers</li>
<li>generating random strings</li>
<li>generating cryptographically secure sequences of random numbers or strings</li>
<li>generating random samples and permuations</li>
<li>analyzing distributions of values in an input file and generating values "like"
the values in the file</li>
<li>generating data for grouped frequency distributions or histograms</li>
</ul></p>
</subsection>
<subsection name="2.2 Random numbers" href="deviates">
<p>
The <a href="../apidocs/org/apache/commons/math/random/RandomData.html">
org.apache.commons.math.RandomData</a> interface defines methods for generating
random sequences of numbers. The API contracts of these methods use the following concepts:
<dl>
<dt>Random sequence of numbers from a probability distribution</dt>
<dd>There is no such thing as a single "random number." What can be generated
are <i>sequences</i> of numbers that appear to be random. When using the
built-in JDK function <code>Math.random(),</code> sequences of values generated
follow the <a href="http://www.itl.nist.gov/div898/handbook/eda/section3/eda3662.htm">
Uniform Distribution</a>, which means that the values are evenly spread over the interval
between 0 and 1, with no sub-interval having a greater probability of containing generated
values than any other interval of the same length. The mathematical concept of a <a href="http://www.itl.nist.gov/div898/handbook/eda/section3/eda36.htm">
probability distribution</a> basically amounts to asserting that different ranges in the set
of possible values for of a random variable have different probabilities of containing the value.
Commons Math supports generating random sequences from the following probability distributions. The
javadoc for the <code>nextXxx</code> methods in <code>RandomDataImpl</code> describes the algorithms used
to generate random deviates from each of these distributions.
<ul>
<li><a href="http://www.itl.nist.gov/div898/handbook/eda/section3/eda3662.htm">uniform distribution</a></li>
<li><a href="http://www.itl.nist.gov/div898/handbook/eda/section3/eda3667.htm">exponential distribution</a></li>
<li><a href="http://www.itl.nist.gov/div898/handbook/eda/section3/eda366j.htm">poisson distribution</a></li>
<li><a href="http://www.itl.nist.gov/div898/handbook/eda/section3/eda3661.htm">Gaussian distribution</a></li>
</ul>
</dd>
<dt>Cryptographically secure random sequences</dt>
<dd>It is possible for a sequence of numbers to appear random, but nonetheless to be
predictable based on the algorithm used to generate the sequence. If in addition to
randomness, strong unpredictability is required, it is best to use a
<a href="http://www.wikipedia.org/wiki/Cryptographically_secure_pseudo-random_number_generator">
secure random number generator</a> to generate values (or strings). The nextSecureXxx methods
in the <code>RandomDataImpl</code> implementation of the <code>RandomData</code> interface use the
JDK <code>SecureRandom</code> pseudo-random number generator (PRNG)
to generate cryptographically secure sequences. The <code>setSecureAlgorithm</code> method
allows you to change the underlying PRNG. These methods are <strong>much slower</strong> than
the corresponding "non-secure" versions, so they should only be used when cryptographic security
is required.</dd>
<dt>Seeding pseudo-random number generators</dt>
<dd>By default, the implementation provided in <code>RandomDataImpl</code> uses the JDK-provided
PRNG. Like other PRNGs, the JDK generator generates sequences of random numbers based on an initial
"seed value". For the non-secure methods, starting with the same seed always produces the same
sequence of values. Secure sequences started with the same seeds will diverge. When a new
<code>RandomDataImpl</code> is created, the underlying random number generators are
<strong>not</strong> intialized. The first call to a data generation method, or to a
<code>reSeed()</code> method initializes the appropriate generator. If you do not explicitly
seed the generator, it is by default seeded with the current time in milliseconds. Therefore,
to generate sequences of random data values, you should always instantiate <strong>one</strong>
<code>RandomDataImpl</code> and use it repeatedly instead of creating new instances for
subsequent values in the sequence. For example, the following will generate a random sequence
of 50 long integers between 1 and 1,000,000, using the current time in milliseconds as the seed
for the JDK PRNG:
<pre><code>
RandomDataImpl randomData = new RandomDataImpl();
for (int i = 0; i &lt; 1000; i++) {
value = randomData.nextLong(1, 1000000);
}
</code></pre>
The following will not in general produce a good random sequence, since the PRNG is reseeded
each time through the loop with the current time in milliseconds:
<pre><code>
for (int i = 0; i &lt; 1000; i++) {
RandomDataImpl randomData = new RandomDataImpl();
value = randomData.nextLong(1, 1000000);
}
</code></pre>
The following will produce the same random sequence each time it is executed:
<pre><code>
RandomDataImpl randomData = new RandomDataImpl();
randomData.reSeed(1000);
for (int i = 0; i = 1000; i++) {
value = randomData.nextLong(1, 1000000);
}
</code></pre>
The following will produce a different random sequence each time it is executed.
<pre><code>
RandomDataImpl randomData = new RandomDataImpl();
randomData.reSeedSecure(1000);
for (int i = 0; i &lt; 1000; i++) {
value = randomData.nextSecureLong(1, 1000000);
}
</code></pre>
</dd></dl>
</p>
</subsection>
<subsection name="2.3 Random Strings" href="strings">
<p>
The methods <code>nextHexString</code> and <code>nextSecureHexString</code>
can be used to generate random strings of hexadecimal characters. Both of these
methods produce sequences of strings with good dispersion properties.
The difference between the two methods is that the second is cryptographically secure.
Specifically, the implementation of <code>nextHexString(n)</code> in <code>RandomDataImpl</code>
uses the following simple algorithm to generate a string of <code>n</code> hex digits:
<ol>
<li>n/2+1 binary bytes are generated using the underlying Random</li>
<li>Each binary byte is translated into 2 hex digits</li></ol>
The <code>RandomDataImpl</code> implementation of the "secure" version,
<code>nextSecureHexString</code> generates hex characters in 40-byte "chunks"
using a 3-step process:
<ol>
<li>20 random bytes are generated using the underlying <code>SecureRandom</code>.</li>
<li>SHA-1 hash is applied to yield a 20-byte binary digest.</li>
<li>Each byte of the binary digest is converted to 2 hex digits</li></ol>
Similarly to the secure random number generation methods, <code>nextSecureHexString</code>
is <strong>much slower</strong> than the non-secure version. It should be used only for
applications such as generating unique session or transaction ids where predictability of
subsequent ids based on observation of previous values is a security concern. If all
that is needed is an even distribution of hex characters in the generated strings, the
non-secure method should be used.
</p>
</subsection>
<subsection name="2.4 Random permutations, combinations, sampling" href="combinatorics">
<p>
To select a random sample of objects in a collection, you can use the
<code>nextSample</code> method in the <code>RandomData</code> interface. Specifically,
if <code>c</code> is a collection containing at least <code>k</code> objects, and
<code>ranomData</code> is a <code>RandomDataImpl</code> instance
<code>randomData.nextSample(c, k)</code>
will return an <code>object[]</code> array of length <code>k</code> consisting of
elements randomly selected from the collection. If <code>c</code> contains
duplicate references, there may be duplicate references in the returned array;
otherwise returned elements will be unique -- i.e., the sampling is without
replacement among the object references in the collection. </p>
<p>
If <code>randomData</code> is a <code>RandomDataImpl</code> instance, and
<code>n</code> and <code>k</code> are integers with <code> k &lt;= n</code>,
then <code>randomData.nextPermutation(n, k)</code> returns an <code>int[]</code>
array of length <code>k</code> whose whose entries are selected randomly,
without repetition, from the integers <code>0</code> through <code>n-1</code> (inclusive), i.e.,
<code>randomData.nextPermutation(n, k)</code> returns a random permutation of
<code>n</code> taken <code>k</code> at a time.
</p>
</subsection>
<subsection name='2.5 Generating data "like" an input file' href="empirical">
<p>
This is yet to be written. Any contributions will be gratefully accepted!
</p>
</subsection>
</section>
</body>
</document>