3
$\begingroup$

In short, how would one sample an Exponential Distribution with an exponentially-increasing rate parameter, e.g., $\lambda=e^t$? What distribution does such a random variable follow?

Background: The Exponential Distribution models "time until failure" of, say, lightbulbs. It is parametrized by a constant parameter $\lambda$ called the failure rate that is the average rate of lightbulb burnouts. If $\lambda$ is time-varying, say, proportionally to a power of time, then it turns out that the time until failure is distributed by the more general Weibull distribution. This question involves this relationship, although the time-variance of $\lambda$ is not made explicit.

My question is: What if $\lambda$ varies exponentially in time? Is there a distribution for this case? More simply, how would I sample a random variable that is exponentially distributed with time-varying rate parameter $\lambda = e^t$?

I've searched Google and Wikipedia without success. I'd really appreciate any advice or pointers. Thanks in advance.

$\endgroup$
5
$\begingroup$

If $\lambda(t)=\mathrm e^t$, the density $f$ of the time until failure is such that, for every $t\geqslant0$, $$ f(t)=\mathrm e^{t+1-\mathrm e^t}. $$ More generally, $$ f(t)=\lambda(t)\,\mathrm e^{-\Lambda(t)},\quad\Lambda(t)=\int_0^t\lambda(s)\,\mathrm ds. $$ Note that every distribution with PDF $f$ and CDF $F$ can be written as such, using $$ \lambda(t)=\left\{\begin{array}{ccc}\frac{f(t)}{1-F(t)}&\mathrm{if}&F(t)\lt1,\\0&\mathrm{if}&F(t)=1.\end{array}\right. $$

$\endgroup$
  • $\begingroup$ Thanks for the response! But I am not clear on (1) proofs/references supporting these statements or (2) how they apply to my question. $\endgroup$ – ConvexMartian Mar 24 '14 at 16:04
  • $\begingroup$ "Amusing" comment by the OP, which I had somehow forgotten... $\endgroup$ – Did Dec 7 '17 at 6:21
0
$\begingroup$

You can find proofs/references of what is stated above here.

$\endgroup$
0
$\begingroup$

To sample a value from a non-homogeneous exponential distribution you can follow this steps

S1. Sample $x$ from homogeneous exponential with rate 1

S2. Calculate $\Lambda^{-1} (x)$

where $\Lambda(t)$ is the intensity function (the integral of the rate).

The random variable $\Lambda^{-1} (x)$ has a non-homogeneous distribution with rate $\lambda(t)$.

A reference could be the paper "Generating Nonhomogeneous Poisson Processes" by Raghu Pasupathy.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.