srand() troubles

[Merrill & Michele]

Many of us watched the World Series of Poker this last week and plotted how
we were to take over that world.

My current problem begins with shuffling the deck. For the apps I've
written before, I've always been satisfied with the usual:

#include <time.h>
#include <stdio.h>

int main( 2 things) {
time_t timer;
int tja;
srand(&timer);

tja=rand();
printf("%?", tja);
return(0);
}
The problem is that the time is always around 109 billion. This, in turn,
causes the generated pseudo-random to be around 23,000 with my
implementation. Even though I've been careful to slice up the 32768
outcomes of rand() in a reasoned way, the initial tilting has me wondering.
Two questions.
A) Do reasonable probabilists use rand(), given that you're not fooling
Gian Carlo Rota, but Chris Moneymaker?
B) If yes to A), then I think I'd want take that timer value, mask out the
things that don't change, maybe do a flip and an Xor, and then seed srand().
Does that sound like I'd then get values for the initial rand call with mu
around 16,000 and a bell curve?

[Ben Pfaff]

"Merrill & Michele" <be********@com cast.net> writes:

My current problem begins with shuffling the deck.

http://www.stanford.edu/~blp/writings/clc/shuffle.html
http://www.stanford.edu/~blp/writings/clc/random.html
--
"...Almost makes you wonder why Heisenberg didn't include postinc/dec operators
in the uncertainty principle. Which of course makes the above equivalent to
Schrodinger's pointer..."
--Anthony McDonald

[Keith Thompson]

"Merrill & Michele" <be********@com cast.net> writes:

Many of us watched the World Series of Poker this last week and plotted how
we were to take over that world.

My current problem begins with shuffling the deck. For the apps I've
written before, I've always been satisfied with the usual:

#include <time.h>
#include <stdio.h>

int main( 2 things) {
time_t timer;
int tja;
srand(&timer);

tja=rand();
printf("%?", tja);
return(0);
}
Show us real code. We can't guess how your pseudo-code differs from
the code you actually compiled.
The problem is that the time is always around 109 billion. This, in turn,
causes the generated pseudo-random to be around 23,000 with my
implementation. Even though I've been careful to slice up the 32768
outcomes of rand() in a reasoned way, the initial tilting has me wondering.
Don't assume that there are 32768 possible values; use RAND_MAX.
Two questions.
A) Do reasonable probabilists use rand(), given that you're not fooling
Gian Carlo Rota, but Chris Moneymaker?
B) If yes to A), then I think I'd want take that timer value, mask out the
things that don't change, maybe do a flip and an Xor, and then seed srand().
Does that sound like I'd then get values for the initial rand call with mu
around 16,000 and a bell curve?

The standard guarantees a few things about the behavior of rand() and
srand(), but much of their behavior is implementation-defined. For
example, it's not uncommon for rand() to return alternating odd and
even numbers. Many implementations provide higher quality random
number generators.

There's also some good information in section 13 of the C FAQ,
<http://www.eskimo.com/~scs/C-faq/faq.html>.

--
Keith Thompson (The_Other_Keit h) ks***@mib.org <http://www.ghoti.net/~kst>
San Diego Supercomputer Center <*> <http://users.sdsc.edu/~kst>
We must do something. This is something. Therefore, we must do this.

[Keith Thompson]

"Merrill & Michele" <be********@com cast.net> writes:

Show us real code. We can't guess how your pseudo-code differs from
the code you actually compiled.
Thanks both. The chance of me being able to reproduce code on a console
that does anything except draw scorn from a compiler is vanishingly small.
Furthermore, your output will be different from mine as a matter of
cosmology. I've attached a screen dump that shows the situation without
showing too much bad programming form. (Do people prefer to see code pasted
into the body of a message or as an attachment?)

Attachments are strongly discouraged, binary attachments even more so.
If you want to post a code sample, paste it into the body of your
message. If at all possible, post a small self-contained compilable
program. Take a look at some of the other articles in the newsgroup
to get an idea of how things are done around here.

If you're really unable to create compilable C code, I'm afraid you
may not be ready for us to help you. Learn the basics of the language
first. Kernighan & Ritchie's _The C Programming Language_, 2nd
edition, is widely considered to be one of the best tutorials. Work
through the exercises.

Going back to the pseudo-code you posted earlier, you had:

time_t timer;

srand(&timer);

srand() expects an unsigned int argument; you're passing it the
address of a time_t variable, which makes no sense. If your compiler
doesn't warn you about this, either it's a really bad compiler, or
you're missing a "#include <stdlib.h>" directive, or you've suppressed
or ignored diagnostic messages somehow.
I had no idea that I could have other than 2^15 outcomes. How often is this
the case, given that I'm not ever going to play Texas Hold'em with an
embedded system? MPJ

Your system should have documentation for the standard library
functions, include srand() and rand(). If you can't find that,
consult any decent C textbook or do a web search. And I already
pointed you to the C FAQ.

A quick summary:

srand() takes an unsigned int argument which is used as the seed for a
sequence of pseudo-random numbers. It should normally be called
exactly once in your program, before any calls to rand(). The call
srand(time(NULL ));
is often good enough (but *only* if the appropriate headers are
included).

rand() takes no arguments, and returns an int result. The result is a
pseudo-random integer in the range 0 to RAND_MAX. RAND_MAX is
guaranteed to be at least 32767.

--
Keith Thompson (The_Other_Keit h) ks***@mib.org <http://www.ghoti.net/~kst>
San Diego Supercomputer Center <*> <http://users.sdsc.edu/~kst>
We must do something. This is something. Therefore, we must do this.

[Guillaume]

Many implementations of rand() are extremely poor.
And the Microsoft's one is particularly hideous. It just isn't suitable
for anything else that toying around.

I suggest you read Knuth's Seminumerical Algorithms to begin with.

[Merrill & Michele]

> > Thanks both. The chance of me being able to reproduce code on a console

that does anything except draw scorn from a compiler is vanishingly

small.
Attachments are strongly discouraged, binary attachments even more so.

Thank you again for your thoughtful response. My shortcomings with
srand(time(&tim er)) were amended in my binary attachment, which, along with
the .eml it came with, seems to have disappeared. Anscheinend ist Ashcroft
nicht der Einzige, der mir die Post verweigert:-) Is the no binaries policy
to keep the porno punks out?

What to do? I won't read K&R. I did pay $2^5 for the last book I bought on
comp sci. I would like to announce that my demographic and I would gladly
pay $2^6 for a book entitled: Essential Knuth in C. MPJ

-----
All right
Send lawyers, guns and money
Huh!
--Warren Zevon

[Keith Thompson]

"Merrill & Michele" <be********@com cast.net> writes:

> Thanks both. The chance of me being able to reproduce code on a console
> that does anything except draw scorn from a compiler is vanishingly

small.

Attachments are strongly discouraged, binary attachments even more so.

Thank you again for your thoughtful response. My shortcomings with
srand(time(&tim er)) were amended in my binary attachment, which, along with
the .eml it came with, seems to have disappeared. Anscheinend ist Ashcroft
nicht der Einzige, der mir die Post verweigert:-) Is the no binaries policy
to keep the porno punks out?

Many of us use newsreaders that don't support binary attachments.
There are newsgroups for binaries; many servers don't carry them
because of the huge bandwidth and storage requirements. This is a
discussion forum. If you want the web, you know where to find it.
What to do? I won't read K&R. I did pay $2^5 for the last book I bought on
comp sci. I would like to announce that my demographic and I would gladly
pay $2^6 for a book entitled: Essential Knuth in C. MPJ

You won't read K&R? Why on Earth not? I find it difficult to believe
that anyone who refuses to read K&R is at all serious about learning C.

--
Keith Thompson (The_Other_Keit h) ks***@mib.org <http://www.ghoti.net/~kst>
San Diego Supercomputer Center <*> <http://users.sdsc.edu/~kst>
We must do something. This is something. Therefore, we must do this.

[CBFalconer]

Merrill & Michele wrote:

.... snip ... cosmology. I've attached a screen dump that shows the situation
without showing too much bad programming form. (Do people prefer
to see code pasted into the body of a message or as an attachment?) .... snip ...
Name: ccode1.JPG
ccode1.JPG Type: JPEG Image (image/jpeg)
Encoding: x-uuencode

NEVER attach a binary to a usenet message. In fact, never attach
anything. Cut your code down to the minimum to show the problem,
and remain compilable (not over about 100 lines) and paste that
into the message.

--
"This is a wonderful answer. It's off-topic, it's incorrect,
and it doesn't answer the question." -- Richard Heathfield

"I support the Red Sox and any team that beats the Yankees"

[Merrill & Michele]

> You won't read K&R? Why on Earth not? I find it difficult to believe

that anyone who refuses to read K&R is at all serious about learning C.

I can't imagine you're interested in my biography, but I'll tell you a
little bit about it. I used to be a champion speller. Then I became
bilingual, trilingual, learned basic, pascal, fortran, c and so forth (not
in that order). Now I can't spell. It turns out, when you know the
Mittelhochdeuts ch and go cross-eyed in cyrillic, it's a one-way street to
missing skills one formerly had.

K&R have **plenty** of influence around here. If the absent demigods want
to step up and be leaders, now would be a good time. MPJ

P.S. My apologies in advance if K or R is unwell.

[Ben Pfaff]

"Merrill & Michele" <be********@com cast.net> writes:

I can't imagine you're interested in my biography, but I'll tell you a
little bit about it. I used to be a champion speller. Then I became
bilingual, trilingual, learned basic, pascal, fortran, c and so forth (not
in that order). Now I can't spell. It turns out, when you know the
Mittelhochdeuts ch and go cross-eyed in cyrillic, it's a one-way street to
missing skills one formerly had.

K&R have **plenty** of influence around here. If the absent demigods want
to step up and be leaders, now would be a good time. MPJ

P.S. My apologies in advance if K or R is unwell.

I nominate the above for "Weird comp.lang.c Article of the Month".
--
Ben Pfaff
email: bl*@cs.stanford .edu
web: http://benpfaff.org