Exponential Distribution

Core Concepts

If a random variable $X$ follows an Exponential distribution, we write:

Where:

• $\lambda$ (lambda) is the rate parameter (the exact same rate parameter used in the Poisson distribution!).

Probability Density Function (PDF)

The probability density function creates a rapidly decaying curve:

Cumulative Distribution Function (CDF)

For continuous distributions, the CDF is often more useful than the PDF because it directly gives us the probability that the time between events will be less than or equal to $x$.

The "Memoryless" Property: The "Reset" Button

The Intuition

Imagine you are waiting at a bus stop where buses arrive according to an exponential distribution (a Poisson Process). You have already been waiting for 20 minutes, and no bus has come.

• Human Intuition: "I've been waiting 20 minutes! Surely the bus is more likely to arrive in the next 5 minutes than it was when I first got here."
• The Reality: The probability that the bus arrives in the next 5 minutes is exactly the same as it was when you first arrived. The system has "no memory" of your 20-minute wait.

This property implies there is no wear and tear. If a lightbulb's lifespan is exponential, a 10-year-old bulb is just as "healthy" as a brand new one.

The Mathematical Proof

To prove this, we use the Survival Function, which is $P(X > t)$. For the exponential distribution:

We want to show that the probability of waiting an additional time $t$, given that we have already waited time $s$, is the same as the original probability of waiting time $t$.

Step 1: Set up the Conditional Probability
By the definition of conditional probability:

Step 2: Simplify the Numerator
If we know that $X > s + t$, then $X$ must already be greater than $s$. Therefore, the intersection of these two events is simply $X > s + t$:

Step 3: Plug in the Survival Function

Step 4: Use Exponent Rules

Conclusion:

The "already waited" time $s$ has completely vanished from the equation.

The "Discrete" Cousin

The only other distribution with this property is the Geometric Distribution, which is essentially the "discrete version" of the Exponential (counting trials instead of measuring time).

Key Metrics

• Expected Value (Mean): $E[X] = \frac{1}{\lambda}$
  • (Intuition: If a bus arrives at a rate of $\lambda = 4$ times per hour, you expect to wait $\frac{1}{4}$ of an hour (15 minutes) for the next bus).
• Variance: $\text{Var}(X) = \frac{1}{\lambda^2}$

Real-World Examples

1. Waiting Times: The time you wait until the next customer calls the support desk.
2. Reliability Engineering: The amount of time until a mechanical part fails (assuming a constant failure rate).
3. Radioactive Decay: The time until a radioactive atom decays.

Interactive Visualization

Use the slider below to adjust the rate parameter ($\lambda$). Notice how the density is highly concentrated near zero, meaning shorter wait times are always more probable than longer wait times. As the rate $\lambda$ increases, the curve decays even faster.

Test Your Knowledge