[Varun Mehta]

I've been programming in C and REALBasic for a long time now, and as a developer have finally decided to move to Objective-C and OS X. I'm trying to generate random numbers, but the strangest anomaly keeps occurring. when I rand()%6 the results always alternate between odd and even!

So my output usually goes like this

0 3 2 5 4 1 2 3

What do you use for your srand() if you think it's needed? Could bad seed initialization be the problem? Is there a better way in Cocoa apps to generate random numbers?

[Seb G]

Funny. rand() seems to generate those alternatingly odd and even sequences on Mac OS - works better on Sun. Doesn't seem to depend on the seed either. Try

(rand()/2) % 6 /* works for me, get: 0 0 3 1 2 2 1 3 4 5 5 ... */

or better use random() and srandom(time(NULL)) instead. This calls a "better" random number generator (see also 'man random' in the terminal):

random()%6 + 1 /* gives 1 0 1 2 3 3 0 5 1 0 1 3 5 ... */

[Apocalypse]

This is common to many implementations of random functions. For this reason, you are often instructed to perform integer division by some large number to ignore the lower-order bits of the generated number.

Sorry to be telling you that mod is exactly what not to do.

Hope that helps,
Jeff.

[skipjack]

It sounds like you know what's going on, so I'm not sure what the problem is. The rand function generates a pseudo-random integer and you should get the same sequence. (In fact, P. J. Plauger states that it is desirable to get the same sequence every time for debugging purposes in his "The Standard C Library.") The srand function varies the starting point in that sequence, so you need to call both rand and srand. (If rand is called before srand, the sequence generated is for srand with a seed of 1.) Srand is commonly seeded with the time() function. This is standard C.

I don't see any additional methods in C++. As mentioned previously, the low-order bits are often suspect using rand() %n. Stroustrup says this: "Often int((double(rand())/RAND_MAX)*n) gives acceptable results. However, to seriously use that formula, we must take care of the minuscule probability that the result will be n."

Stroustrup provides a class representation of a random number generator in his "The C++ Programming Language".

[ 05-18-2002: Message edited by: skipjack ]

[markko]

Originally posted by Varun Mehta:
<STRONG>I've been programming in C and REALBasic for a long time now, and as a developer have finally decided to move to Objective-C and OS X. I'm trying to generate random numbers, but the strangest anomaly keeps occurring. when I rand()%6 the results always alternate between odd and even!

So my output usually goes like this

0 3 2 5 4 1 2 3

What do you use for your srand() if you think it's needed? Could bad seed initialization be the problem? Is there a better way in Cocoa apps to generate random numbers?</STRONG>

As far as i know this is just what you can expect from the standard ANSI C implementation. Randomization with srand() , as fa as i know, as before, is always needed. For simple stuffs i just include time.h and use

srand(time(NULL))

as randomization seed.
just my two cents

[lindberg]

Don't use srand/rand. Use srandom()/random() instead. It's not as portable, but it's a much much better random number generator. rand() is awful. If you need portability, find the source code to a generator out there. Mersenne Twister is good, for one, and there are plenty of others. Just about anything is better than rand().

[skipjack]

Originally posted by lindberg:
<STRONG>Don't use srand/rand. Use srandom()/random() instead. It's not as portable, but it's a much much better random number generator. rand() is awful. If you need portability, find the source code to a generator out there. Mersenne Twister is good, for one, and there are plenty of others. Just about anything is better than rand().</STRONG>

Sorry for being such a rookie, but I can't find any reference to srandom()/random() in "The C++ Programming Language" (1997), "The Draft Standard C++ Library" (1995), or "The Standard C Library" (1992). I'm not a programmer and I know my references may be a bit out of date. How do we get these functions?

[Seb G]

They are in the Darwin stdlib, i.e. you
#include &lt;stdlib.h&gt;
as you would for rand() and srand(). They are declared as
long random(void);
and
void(srandom(unsigned seed);
You get the documentation by typing 'man random' in the terminal.

The random()/ srandom() have (almost) the same calling sequence and initialization properties as rand(3)/ srand(3). The difference is that rand produces a much less random sequence -- in fact, the low dozen bits generated by rand go through a cyclic pattern. All the bits generated by random() are usable. For example, `random()&01' will produce a random binary value.

(from the manpage)

[skipjack]

Originally posted by Seb G:
<STRONG>They are in the Darwin stdlib</STRONG>

Thanks! I guess this means I need to update my C language references.

[Amorph]

You won't have any luck with updating your references, because srandom() and random() aren't part of the C or C++ languages.

They are, however, common on UNIXen, and possibly even POSIX standard (I'm not sure about that, though).

[Mskr]

Originally posted by Amorph:
<STRONG>You won't have any luck with updating your references, because srandom() and random() aren't part of the C or C++ languages.

They are, however, common on UNIXen, and possibly even POSIX standard (I'm not sure about that, though).</STRONG>

According to UNIX Systems Programming for SVR4 (great book, btw), pg 485:

<STRONG>Some versions of UNIX, usually those based on BSD, also suppy <font face = "courier">random</font> and <font face = "courier">srandom</font>, with similar semantics.</STRONG>

So, it appears to not be POSIX, but rather a BSD UNIX extension, thus included in Mac OS X's libc. The Linux manpage that I have access to right now says that glibc's version (2.2.5 on my Debian box) complies with BSD 4.3. It also gives a more technical description of the entropy function that is used. The rand() manpage also gives some very good pointers on more technical discussion of random number generation:

<STRONG><font face = "courier">In Numerical Recipes in C: The Art of Scientific Computing
(William H. Press, Brian P. Flannery, Saul A. Teukolsky,
William T. Vetterling; New York: Cambridge University
Press, 1992 (2nd ed., p. 277)), the following comments are
made:
"If you want to generate a random integer between 1
and 10, you should always do it by using high-order
bits, as in

j=1+(int) (10.0*rand()/(RAND_MAX+1.0));

and never by anything resembling

j=1+(rand() % 10);

(which uses lower-order bits)."

Random-number generation is a complex topic. The Numeri�
cal Recipes in C book (see reference above) provides an
excellent discussion of practical random-number generation
issues in Chapter 7 (Random Numbers).

For a more theoretical discussion which also covers many
practical issues in depth, please see Chapter 3 (Random
Numbers) in Donald E. Knuth's The Art of Computer Program�
ming, volume 2 (Seminumerical Algorithms), 2nd ed.; Read�
ing, Massachusetts: Addison-Wesley Publishing Company,
1981.
</font></STRONG>

I know, I know, too much information....