Bitcoin Forum
November 10, 2024, 06:54:52 AM *
News: Latest Bitcoin Core release: 28.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: [1]
  Print  
Author Topic: How do you calculate probability?  (Read 10325 times)
mizerydearia (OP)
Hero Member
*****
Offline Offline

Activity: 574
Merit: 513



View Profile
July 19, 2010, 03:13:43 AM
Last edit: July 19, 2010, 03:32:42 AM by mizerydearia
 #1

.I am trying to correct some PHP code to calculate probability of generating a block for bitbot.  bd_ and llama have provided some debate regarding how to calculate it and I suggest it be moved to forum for further eyes to analyze their findings and to come to a conclusion on best formula to calculate probability.

For now, here's code I've been using:
Code:
		//var p = target*rate*1000/Math.pow(2,256);
//var txt = "<table border='1'>
// <tr style='font-weight:bold'> <td>Probability</td> <td>Time</td></tr>\n"+"
// <tr> <td>Average</td> <td>"+getHumanTime(1/p)+"</td></tr>\n"+getRow(0.50,p)+getRow(0.95,p)+"</table>";
//document.getElementById('answer').innerHTML = txt;
//function getRow(prob,p) { return "<tr><td>"+100*prob+"%</td><td>"+getHumanTime(-Math.log(1-prob)/p)+"</td></tr>\n";

bcscale(256);
$a = bcsub(bcpow(2,256),1);
$b = bcmul(bcpow(2,32),$difficulty);
$target = bcdiv($a,$b); // target
$pph = bcdiv($target,bcpow(2,256)); // probability per hash
if (isset($_GET["r"]) && $_GET["r"] != "") {
$rate = $_GET["r"];
if (is_numeric($rate)) {
function humantime($secs) {
if ($secs<0) return false;
$m = (int)($secs / 60); $s = $secs % 60; $s = ($s <= 9) ? "0$s" : $s;
$h = (int)($m / 60); $m = $m % 60; $m = ($m <= 9) ? "0$m" : $m;
$d = (int)($h / 24); $h = $h % 24;
return $d."d $h:$m:$s";
}
echo "http://bitcointalk.org/index.php?topic=471.0";exit;
// Where does this go?
// <bd_> To compute the number of hashes needed to reach a success probability of P_targ, you'll want n >= log(P_targ)/log(1-P)
// <bd_> So time needed is (log(P_targ)/log(1-P))/rate


// <bd_> To compute the probability per second, you'll want 1-(1-p)^rate, where p is the probability of a single hash being correct
//$pps = bcdiv(bcmul($target,1000),bcpow(2,256)); // probability per second (according to bd_ this is wrong)
bcscale(16); // Next calculation takes too long for higher bits
$pps = bcsub(1,bcpow(bcsub(1,$pph),$rate)); // probability per second
//echo "probabiliy per second == 1-(1-p)^rate == 1-(1-$pph)^$rate == $pps";exit;
// Comment next line for 256-bits of data
bcscale(16);
$ppr = bcmul($pps,$rate);
//$ppr = $pps;
$etaAvg = humantime(bcdiv(1,$ppr));
$eta25 = humantime(bcdiv(-log(.75),$ppr));
$eta50 = humantime(bcdiv(-log(.5),$ppr));
$eta75 = humantime(bcdiv(-log(.25),$ppr));
$eta95 = humantime(bcdiv(-log(.05),$ppr));
$eta99 = humantime(bcdiv(-log(.01),$ppr));
echo "ProbabilityPerSecond($ppr) Avg($etaAvg) 25%($eta25) 50%($eta50) 75%($eta75) 95%($eta95) 99%($eta99)";
} else echo "This command requires either no argument or a numeric argument representing khash/sec.";
}
else {
bcscale(32);
$pph = bcmul($pph,1);
echo "ProbabilityPerHash($pph)";
}

http://nullvoid.org/bitcoin/bc.php?q=gethashprobability&r=2000 where r is your khash/sec
llama
Member
**
Offline Offline

Activity: 103
Merit: 61


View Profile
July 19, 2010, 03:38:52 AM
 #2

Actually, it seems that BD_ and I have come to an agreement!

To calculate average (expected value of generation time):

1/(pr)

To calculate how long until you have a T chance of producing at least 1 block:

-ln(1-T)/pr


Where:

p = probability of a given hash resulting in success (TARGET / 2^256)
r = rate in hashes / sec
T = target probability, such as 0.95
ln = log base e


I got to mine just by plugging variables and rearranging the poisson distribution.  BD_ can give you a different formula for the second equation, which he heroically derived from the binomial distribution, but which yields the same effective results.

Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!