Scribd
Upload
35 views2 pages

Derivations of The Formula's For Poisson Processes

The document derives the formulas for the Poisson distribution and exponential distribution. [1] The Poisson distribution Pn(t) gives the probability of n events occurring between time 0 and t, where t is fixed. [2] The exponential distribution P(t) gives the probability density of the time for a single event to occur. [3] The derivations show that Pn(t) = e^-λt (λt)n / n! and P(t) = λe^-λt, where λ is the constant rate of events.

Uploaded by

wadsss
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
35 views2 pages

Derivations of The Formula's For Poisson Processes

The document derives the formulas for the Poisson distribution and exponential distribution. [1] The Poisson distribution Pn(t) gives the probability of n events occurring between time 0 and t, where t is fixed. [2] The exponential distribution P(t) gives the probability density of the time for a single event to occur. [3] The derivations show that Pn(t) = e^-λt (λt)n / n! and P(t) = λe^-λt, where λ is the constant rate of events.

Uploaded by

wadsss
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

CPSC 405 Handout

Derivations of the formulas for Poisson processes


A Poisson process is a particular random process in time. Associated with it are two distributions.The rst is the Poisson distribution Pn (t), which is the probability that precisely n events happen rom time 0 to time t. It is a pmf in n, for xed value of time t. The second is the exponential distribution P (t), which is a pdf in t. P (t) is the pdf for an event happening after a particular time t. It is also the pdf for the inter-event time for events. You should understand these derivations, but I wont ask you to reproduce them on an exam. I will test your understanding of this material though. This material is supplementary to Chapter 5 (6 in 2nd ed.) of the book. Let an event E have constant probability to occur per time, i.e., a Poisson process. Lets say the probability for it to occur in a small time interval dt is given by dt. The Poisson distribution Pn (t) is the probability that after the elapse of time t, n events happened. This is a p.m.f. in n, but not a p.d.f. in t. Lets derive Pn (t) = et (t)n /n! In particular, P0 (t) = et . P0 (t) is the chance that no events have taken place at time t, starting from t = 0. If we now subdivide the interval [0 t] in small segments of size dt, the event E should not happen in each of those intervals. The chance for no event in a single interval is 1 t. Therefore the chance it does not happen in any interval is P0 (t) = (1 dt)N with N = t/dt the number of intervals. In the limit dt 0 we have, using a well-known formula for the exponential P0 (t) = lim (1 dt)(t/dt) = et .
dt0

What is now the p.d.f. P (t) for the event to happen at a particular time t? (Note this is a continuous distribution with P (t)dt = 1.) P (t)dt is the probability the event will happen in the innitesimal time interval [t t + dt]. We know the chance for the event to happen in an interval dt, namely dt. For the event to happen in [t t + dt] it should not happen until time t and happen in the next time interval of size dt. The chance for it not to happen until time t is just P0 (t) = et . The product of the two is et dt, from which it follows that P (t) = et . P (t) is called the exponential distribution. The inter-event time is distributed according to P (t). What remains to be derived is the rest of the Poisson distribution, Pk (t) for k > 0. Let us derive a formula for Pn (t + dt). Pn (t + dt) is the chance for n events to happen at time 1

t + dt. This can happen by n events happening until time t and no events in the interval [t t + dt], or by n 1 events to happen until time t and one event in the interval [t t + dt]. This means Pn (t + dt) = Pn (t)(1 dt) + Pn1 (t)dt. Substituting Pn (t + dt) = Pn (t) + dt gives dPn (t) dt = (Pn (t) + Pn1 (t))dt. dt Dividing out dt gives dPn (t) = (Pn1 (t) Pn (t)). dt This is a recursive dierential equation for Pn (t). If can be seen easily by substitution the formula for the Poisson distribution that Pn (t) = et (t)n /n! indeed satises this equation. Strictly speaking we should also prove uniqueness, but well stop here. dPn (t) dt

You might also like