Re: Randomizing Large Integers

Replies are listed 'Best First'.
Re^2: Randomizing Large Integers by almut (Canon) on May 19, 2008 at 19:48 UTC
Is 32K a limitation of the standard rand function? The macro (`uconfig.h`, line 3368 in 5.8.8 or 4057 in 5.10.0) looks like only 15 bits are being used: `#define Drand01() ((rand() & 0x7FFF) / (double) ((unsigned long)1 << +15)) ^^^^` [download] Update: though, a somewhat closer look suggests this might be configuration dependent (at least, `Configure` checks for `drand48()`), so maybe I spoke too soon... Update2: just ran a test build on a system which has `drand48()`... et voila, the macro ends up being defined as `#define Drand01() drand48()` [download] So, as often, the answer is: it depends :)	[reply] [d/l] [select]
Re^3: Randomizing Large Integers by mscharrer (Hermit) on May 19, 2008 at 20:29 UTC
15 bit of the input or of the output? I get the following behavior that lets me assume that 32 bit random integers work: `# perl -e 'printf "%010x\n", int rand 216' 00000052e2 # perl -e 'printf "%010x\n", int rand 232' 006f7ae0c5 # perl -e 'printf "%010x\n", int rand 264' 00ffffffff` [download] i.e. `rand 2N` seems to return a `N` bit random integer, up to `N = 32` but not above.	[reply] [d/l] [select]
Re^4: Randomizing Large Integers by BrowserUk (Patriarch) on May 19, 2008 at 21:25 UTC
The 15-bit rand function is a Win32 Perl (MS CRT-based) problem. See Re^3: Rand Wackiness! and thread at A better rand() for Win32 for some further discussion. Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error. "Science is about questioning the status quo. Questioning authority". In the absence of evidence, opinion is indistinguishable from prejudice. "Too many [] have been sedated by an oppressive environment of political correctness and risk aversion."	[reply]
Re^4: Randomizing Large Integers by ikegami (Patriarch) on May 20, 2008 at 16:59 UTC
15 bit of the input or of the output? 15 bits of entropy. For a given argument, rand will only spit out 2¹⁵ different results (spread over the requested range).	[reply]
Re^2: Randomizing Large Integers by hangon (Deacon) on May 19, 2008 at 20:23 UTC
From what I've read, rand is dependent on whats installed on the OS. On this system, when feeding rand with integers > 32768, the distribution of values returned (from multiple calls using the same number) breaks down badly. So this seems to support 15 bits.	[reply]
Re^3: Randomizing Large Integers by syphilis (Archbishop) on May 20, 2008 at 08:01 UTC
Running `perl -V:randbits` will tell you how many random bits your perl's rand() has. Cheers, Rob	[reply] [d/l]
Re^4: Randomizing Large Integers by ysth (Canon) on May 21, 2008 at 04:49 UTC
Or, if you want it programmatically: `use Config; print $Config{'randbits'};` [download] -- Online Fortune Cookie Search	[reply] [d/l]