Title: Statistical distribution of mining bitcoins? Post by: kseistrup on March 06, 2011, 05:32:56 PM Hi there,
Does anyone know which statistical distribution bitcoin mining is following? I've been looking at the output from jgarzik's cpuminer, and I expected the “PROOF OF WORK RESULT: true (yay!!!)” lines to approximate a Poisson distribution (http://en.wikipedia.org/wiki/Poisson_distribution) (completely random, rare event). Possion distributed data has same mean and variance, but the data I'm looking at has a variance of appr. mean˛. Anyone? Cheers, Title: Re: Statistical distribution of mining bitcoins? Post by: Cryptoman on March 06, 2011, 05:55:26 PM Yes, it's Poisson.
Title: Re: Statistical distribution of mining bitcoins? Post by: kseistrup on March 06, 2011, 06:27:43 PM Yes, it's Poisson. I beg to differ. When I analyze the logfile from the cpuminer I get a mean (i.e., average time between delivering PoW) of appr. 1725 seconds with a variance of appr. 600E6 s˛. The landmark of the poisson distribution is that its mean equals its variance. Cheers, Title: Re: Statistical distribution of mining bitcoins? Post by: ArtForz on March 06, 2011, 06:56:32 PM It's poisson.
Remember, if events in a fixed timespan is poisson, time between events is a exponential distribution. Title: Re: Statistical distribution of mining bitcoins? Post by: kseistrup on March 06, 2011, 07:22:39 PM It's poisson. Remember, if events in a fixed timespan is poisson, time between events is a exponential distribution. Yes, you're right. Except: Even when I feed events/timeunit into the calculator, the variance is much higher than expected. But perhaps I don't have enough data points… Thanks for your input. Cheers, |