Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Regarding a question, where the lifespan of a machine part is exponentially distributed with $\lambda=1/10$. When asked about the average lifespan of $100$ independent components, why is $\mathbb{E}(X)=10$ and $SD(X)=\frac{10}{\sqrt{100}}$? Can someone explain how one arrives at the standard deviation, and why its not equal to $\mathbb{E}(X)$, or why we don't use the Gamma Distribution (where $\mathbb{E}(X)=\frac{r}{\lambda}$ and $SD(X) =\frac{\sqrt{r}}{\lambda}$ respectively in this case? Am I misunderstanding the problem conceptually?

share|cite|improve this question

So we are considering independent $X_1, \ldots, X_{100}$, where each $X_i$ is exponentially distributed and their mean $X := \frac 1{100}\sum_{i=1}^{100} X_i$. We want to compute its exceptation and its standard derivation. First we note, that by linearity of exceptation: \[ \mathbb E[X] = \frac 1{100} \sum_{i=1}^{100} \mathbb E[X_i] = \frac 1{100} \cdot 100 \cdot 10 = 10 \] and by independence we have addivity of the variance, \[ \sigma^2(X) = \frac 1{100^2} \sum_{i=1}^{100} \sigma^2(X_i) = \frac 1{100^2} \cdot 100 \cdot 10^2 = 1 \] Hence $\mathrm{SD}(X) = \sqrt{\sigma^2(X)} = 1$.

Another way is using the Gamma-distribution, as you suggested. We have $X_i \sim \Gamma(1, 10)$ for each $i$, hence, by independence $\sum_i X_i \sim \Gamma(100, 10)$ and therefore $X = \frac 1{100}\sum_i X_i \sim \Gamma(100, 1/10)$. As a $\Gamma(k,\theta)$ distributed variable has expectation $k\theta$ and standard derivation $\sqrt k\theta$, we have $\mathbb E[X] = 100 \cdot \frac 1{10} = 10$ and $\mathrm{SD}(X) = \sqrt{100} \cdot \frac 1{10} = 1$.

share|cite|improve this answer
Thanks for the clear solution, but regarding the solution using Gamma-distribution, I was under the assumption that $\lambda = \frac{1}{10}$, and that for Gamma, $\mathbb{E}[X] = \frac{r}{\lambda}$, why is $\lambda = 10$ in the case of $X = \Gamma(1,10)$ When you apply linearity of expectations, does the second argument of Gamme multiply as well? – zhuyxn Nov 1 '12 at 10:37
There are different common ways to denote the parameters for a Gamma-distribution. I know $\theta$ and $k$, which should correspond to your $r$, $\lambda$ via $r = k$ and $\lambda = \theta^{-1}$, so in $r, \lambda$-notation $X_i \sim \Gamma(1, 1/10)$, $\sum X_i \sim \Gamma(100, 1/10)$, $X \sim \Gamma(100, 10)$. Therefore $\mathbb E[X] = 100/10 = 10$ and $\mathrm{SD}(X) = \frac{\sqrt r}{\lambda} = 10/10 = 1$. – martini Nov 1 '12 at 10:41
Sorry, I'm a bit a confused (possibly over notation), assuming that $\mathbb{E}[X] = \frac{r}{\lambda}$ and $\lambda = \frac{1}{10}$, wouldn't $\mathbb{E}[X] = \frac{100}{\frac{1}{10}} = 1000$? When you find $X$, why does the $\lambda$ get multiplied by 100? – zhuyxn Nov 1 '12 at 10:53
In your last step in the main solution, when you take $X = \Gamma(100,1/10)$, what does that conceptually mean in terms of dividing the survival time ($\theta$) by 100 from 10 to $\frac{1}{10}$? Why not divide $k$ by 100? – zhuyxn Nov 1 '12 at 11:49
Well... it's known, that if $Y \sim \Gamma(k, \theta)$, then $\alpha Y \sim \Gamma(k, \alpha\theta)$. – martini Nov 1 '12 at 12:37

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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