Exponential Distribution

Intuition

If events arrive according to a Poisson process (independently, at a constant average rate), the exponential distribution describes how long you wait between consecutive events. It is the continuous counterpart of the geometric distribution and shares its memoryless property: the probability of waiting at least $t$ more units is the same regardless of how long you have already waited.

Definition

A random variable $X$ follows an exponential distribution $X\sim \text{Exp}(\beta )$ if it represents the time (or distance) between successive events in a Poisson process with mean inter-arrival time $\beta =1/\lambda$.

$X$ takes non-negative real values $x\ge 0$.

Parameterization note: some texts use the rate $\lambda =1/\beta$ directly, writing $f(x)=\lambda e^{-\lambda x}$. The formulas below use $\beta$ (mean lifetime) as the parameter.

Key Formulas

Probability Density Function (PDF):

$f(x;\beta )=\frac{1}{\beta }{e}^{-x/\beta },x\ge 0$

Cumulative Distribution Function (CDF):

$F(x)=1-e^{-x/\beta }$

Mean:

$\mu =\beta$

Variance:

${\sigma }^2={\beta }^2$

Memoryless Property:

$P(X>s+t\mid X>s)=P(X>t)$

This is the only continuous distribution with this property. It means past survival gives no information about remaining lifetime.

Relationship to Poisson:

If events arrive at rate $\lambda$, then inter-arrival times are $\text{Exp}(1/\lambda )$ and the count in time $t$ is $\text{Pois}(\lambda t)$.

Example

An electronic component has an average life of $\beta =5$ years (exponentially distributed). What is the probability it lasts more than 8 years?

$P(X>8)=1-F(8)=e^{-8/5}=e^{-1.6}\approx 0.2019$

There is roughly a 20% chance the component survives past 8 years. Notice that, by the memoryless property, a component that has already lasted 3 years still has the same 20% chance of lasting 8 more years beyond that point:

$P(X>11\mid X>3)=P(X>8)\approx 0.2019$

We can also compute the median lifetime. Setting $F(m)=0.5$:

$1-e^{-m/\beta }=0.5⟹m=\beta \ln 2=5\ln 2\approx 3.47\text{ years}$

The median is always less than the mean for an exponential distribution, reflecting its right skew.

Hazard rate: the exponential has a constant hazard (failure) rate $h(x)=1/\beta =\lambda$. This means a 5-year-old component is no more likely to fail in the next instant than a brand-new one - a strong assumption that limits the model to components without wear-out.

Why It Matters in CS

• Reliability engineering: Mean Time Between Failures (MTBF) for hardware components is modelled exponentially, enabling maintenance scheduling and redundancy planning.
• Operating systems: process service times in scheduling analysis often assume exponential distributions, yielding tractable queueing models (M/M/1).
• Web server performance: response-time modelling uses exponential assumptions for service duration, driving capacity planning and load balancer configuration.
• Simulation: generating exponential random variates via the inverse-transform method ($X=-\beta \ln U$, where $U\sim \text{Uniform}(0,1)$) is a building block of discrete-event simulation.

Related Notes

• Poisson Distribution - counts events in an interval; the exponential models time between those events
• Normal Distribution - another foundational continuous distribution, used when many independent factors combine additively
• Probability Distributions - overview of the broader distribution taxonomy