Interactive module guide
Poisson Process Relationships Guide
Free · No sign-up · Calculations stay in your browser
A homogeneous Poisson process with rate λ is one random mechanism that produces three famous distributions at once: the number of events in any window of length t is Poisson(λt), the gap between consecutive events is Exponential(λ), and the waiting time until the k-th event is Gamma(k, 1/λ).
This page plots the first-arrival and k-th-arrival densities together and verifies the exact identity P(N(1) ≥ k) = P(T_k ≤ 1)—"at least k events by time 1" and "the k-th arrival happens by time 1" are the same event, so counting and timing must agree perfectly.
Is this the right module for my question?
Use it when
- Use this view to understand why Poisson, exponential, and gamma distributions appear together in queueing, reliability, and arrival modeling: they are three bookkeeping choices for one process, not three separate assumptions.
- Use the identity panel to convert between count questions and timing questions: capacity planning often states questions about counts, while service-level statements are about waiting times, and the identity is the bridge.
- Use the k slider to see the gamma family emerge from sums: the k-th arrival time is a sum of k independent exponential gaps, and the density visibly shifts and symmetrizes as k grows.
Choose another method when
- Avoid Poisson-process reasoning when the rate varies over time (rush hours, seasonality); counts stay Poisson under a time-varying rate only with a rescaled clock, and gaps stop being exponential altogether.
- Avoid it when events cluster or inhibit each other—failures triggering further failures, or refractory periods after each event—since independence of disjoint windows is the defining assumption.
- Avoid reading the two plotted curves as an approximation pair: unlike the other pages in this tool, nothing here converges to anything; every displayed relationship is exact.
Interactive tool
The calculator loads as you approach this section so the guide remains fast on mobile connections.
Example preview
Counts, gaps, and arrivals linked
- Inputs
- Rate λ=4 per hour; ask for at least k=3 events within the first hour.
- Representative result
- P(N(1)≥3)=P(T₃≤1)≈0.7619 computed both through Poisson counts and the Gamma arrival time — an exact identity.
Illustrative only. Load the interactive tool to enter your own values and review assumptions.
Watch the explanation
2:28 minThe player loads only after you press play. You can also watch on YouTube.
Read the complete transcript
Poisson Processes: Counts, Gaps, and Arrivals. One homogeneous Poisson process gives three exact views of the same arrivals: counts in a window, gaps between events, and the time of the k-th arrival. If the constant rate is lambda per unit time, the count N of t follows Poisson with mean lambda times t. Units are part of the model. Each gap between consecutive events follows an exponential distribution with rate lambda and mean one over lambda. The gap has no aging memory. The k-th arrival time adds k independent gaps, so T sub k follows Gamma with shape k and scale one over lambda, not exponential when k exceeds one. At least k events by time t means exactly that the k-th arrival occurred by t. Therefore P of N of t at least k equals P of T sub k at most t. This equality is an event identity, not an approximation. At k equals one, the first-arrival gamma curve and the exponential gap curve coincide exactly. The arrival-time center moves predictably: its mean is k over lambda, and its standard deviation is square root k over lambda.
For support tickets arriving at four per hour, ask whether at least three tickets land in the first hour. Here lambda is four, k is three, and t is one. The Poisson count calculation gives P of N of one at least three equal to point seven six one eight nine seven, about seventy-six percent. The Gamma timing calculation gives P of T three at most one equal to the same point seven six one eight nine seven. Counting and timing agree exactly. One ticket gap has mean point two five hours and median about point one seven three hours. The shorter median reflects the exponential distribution's right skew. The third ticket arrives after point seven five hours on average, with standard deviation about point four three three hours. That is a sum of three random gaps. Do not use the homogeneous model for rush-hour rate changes, clustering, inhibition, or self-exciting arrivals. The page shows theory; diagnostics must test real data. Choose counts, gaps, or arrival times to match the question, then keep the rate and time units aligned. Explore Poisson-process relationships free at Distri Scope dot com.
How to read the result
The expected k-th arrival time is k/λ with standard deviation √k/λ; as k grows the arrival-time density becomes more symmetric—an early preview of gamma-to-normal behavior.
The memoryless property lives in the exponential curve: its shape restarting from any time is the same, which is why gaps carry no aging.
Reading the identity at several k values traces the whole count distribution: P(N(1) ≥ k) as k steps through 1, 2, 3, … is exactly the Poisson survival function at λ.
Inputs and parameter meaning
The rate slider sets λ, the expected number of events per unit time, from 0.5 to 10; the main slider sets the arrival index k from 1 to 20.
At k = 1 the two curves coincide exactly, because the first arrival time is the first gap; that overlap is a built-in correctness check, not a display bug.
The identity panel always evaluates at t = 1; rescaling time rescales λ correspondingly, so no generality is lost.
Limit, parameter, and conditions
Interactive parameter: Process rate λ and the arrival index k.
- Events occur independently at a constant rate λ with no simultaneous arrivals.
- Counts in disjoint windows are independent Poisson variables with mean λt.
- The identity between counts and arrival times is exact, not an approximation.
What the method is doing
The k-th arrival density is Gamma with shape k and scale 1/λ, evaluated exactly; the first-arrival curve is Exponential(λ), its k = 1 special case.
The identity P(N(1) ≥ k) = P(T_k ≤ 1) is computed on both sides—a Poisson upper tail and a gamma CDF—and displayed to six decimals so their exact agreement is visible.
No approximation-quality machinery applies here; the usual overlap metric is replaced by the identity check, since the relationship is exact for every parameter choice.
Worked example: support tickets at four per hour
Tickets arrive as a Poisson process at λ = 4 per hour. The manager wants the chance that at least k = 3 tickets arrive within the first hour, the typical gap between tickets, and the expected time of the third ticket. Set the rate slider to 4 and the arrival slider to 3.
- 1Counts view: P(N(1) ≥ 3) = 1 − e⁻⁴(1 + 4 + 8) ≈ 0.761897 from the Poisson(4) distribution.
- 2Timing view: P(T₃ ≤ 1) from the Gamma(3, 1/4) CDF is 0.761897 as well—the identity panel shows both to six decimals.
- 3Gaps: each inter-arrival gap is Exponential(4), with mean 0.25 hours and median ln 2/4 ≈ 0.173 hours—typical gaps are shorter than the mean because the distribution is right-skewed.
- 4Third arrival: E[T₃] = 3/4 hours with standard deviation √3/4 ≈ 0.43 hours, read directly from the gamma parameters.
Interpretation
There is about a 76% chance the third ticket lands inside the first hour, and that number is the same whether you count events or time arrivals—the identity is exact, not approximate. The worked numbers also show the practical division of labor: Poisson answers staffing questions about counts, the exponential describes idle gaps, and the gamma prices the waiting time to a quota.
Common mistakes
- Do not average the memoryless property away: after waiting ten minutes without a ticket, the remaining wait is still Exponential(λ); the process owes nothing.
- Do not confuse the gap distribution with the arrival-time distribution: gaps are exponential, but the time to the k-th event for k > 1 is gamma, visibly non-exponential.
- Do not apply unit-time formulas to other windows without rescaling: counts in a window of length t are Poisson(λt), not Poisson(λ).
Limits and independent validation
The page models the homogeneous process only; time-varying rates, clustering, and self-exciting processes need different machinery.
The identity is evaluated at t = 1 for clarity rather than at arbitrary t; the mathematics is identical after rescaling.
The display is theoretical—whether real arrival data behave like a Poisson process is a fitting and diagnostics question, not something this page can certify.
Before using the result
- Check the k = 1 coincidence: the two curves must lie exactly on top of each other, since the first arrival is the first gap.
- Verify one identity by hand at small k: for k = 1, both sides reduce to 1 − e^{−λ}, a one-line computation.
- Compare E[T_k] = k/λ against the chart’s visible center as k slides; the drift of the density should track the formula linearly.
Related modules and resources
Explore Other Convergence Visualizations
See every option in the Distribution Convergence Explorer or review the DistriScope methodology. Educational information; last reviewed 2026-08-09. Verify consequential calculations independently.