Probabilistic foundations of Poisson Process & Waiting Times.
The Poisson process is the fundamental mathematical model for counting random events occurring continuously and independently over time. Rooted in the Poisson distribution for event tallies and the continuous Exponential distribution for inter-arrival waiting times, this process underpins queuing theory, telecommunications, radioactive decay, and reliability modeling. In this laboratory, we simulate stochastic arrival streams, verify the exponentiality of waiting times, evaluate the unit Fano factor, and empirically confirm the foundational equidispersion gradient m = 1.000. In mathematical statistics, empirical data science, and modern computational engineering, understanding the asymptotic sampling properties and theoretical behavior of Poisson Process, Rate Parameter Estimation, and Exponential Inter-Arrival Times provides crucial guarantees. Rigorous experimental validation connects foundational probability laws with observable empirical phenomena, ensuring that statistical inference remains valid under finite sample sizes and real-world data collection constraints.
Viva voce questions & analytical theory
What mathematical postulates define a homogeneous Poisson point process?
A homogeneous Poisson process with rate λ > 0 is a continuous-time counting process {N(t), t ≥ 0} satisfying three core axioms: (1) N(0) = 0; (2) Independent increments: the numbers of arrivals in non-overlapping intervals are mutually independent random variables; (3) Stationary increments: the distribution of arrivals in [s, s + t] depends only on the interval length t; and (4) Infinitesimal probabilities: as h → 0, P(N(h) = 1) = λh + o(h) and P(N(h) ≥ 2) = o(h). Together, these axioms uniquely imply that N(t) ~ Poisson(λt).
How are Poisson counts and Exponential inter-arrival times mathematically connected?
The relationship between Poisson counts and Exponential waiting times represents a fundamental duality in probability theory. Let T_1 be the time of the first arrival. The event {T_1 > t} occurs if and only if zero arrivals occur in [0, t]: P(T_1 > t) = P(N(t) = 0) = (λt)^0 e^(-λt) / 0! = e^(-λt). Thus, the cumulative distribution function is F(t) = 1 - e^(-λt), which is the CDF of an Exponential(λ) random variable with density f(t) = λ e^(-λt). By the strong Markov property, all successive inter-arrival intervals T_k = t_k - t_{k-1} are independent and identically distributed Exponential(λ).
What is the memoryless property of the Exponential distribution, and why is it unique?
A positive continuous random variable T is memoryless if for all s, t ≥ 0, P(T > s + t | T > s) = P(T > t). In terms of reliability, the probability that an item survives an additional time t given that it has already survived time s is independent of its past age s. The only continuous probability distributions possessing the memoryless property are the Exponential distributions. In discrete time, the geometric distribution is the unique memoryless distribution.
What is the Fano factor, and how is it used to detect overdispersion or underdispersion?
The Fano factor (index of dispersion) is defined as D = Var(N) / E[N]. For an ideal Poisson process, mean equals variance, so D = 1.0. A process with D > 1.0 exhibits overdispersion, typical of clustered point processes (e.g., negative binomial arrivals or compound Poisson processes). A process with D < 1.0 exhibits underdispersion, reflecting regular or inhibitory arrival spacing (e.g., renewal processes with Erlang inter-arrival times).
How does thinning and superposition operate on Poisson processes?
By the Thinning Theorem, if arrivals from a Poisson(λ) process are independently retained with probability p and deleted with probability 1 - p, the retained arrivals form a Poisson process with rate pλ. Conversely, by the Superposition Theorem, the sum of independent Poisson processes with rates λ_1, λ_2, ..., λ_k yields a unified Poisson process with aggregate rate λ = ∑_{i=1}^k λ_i.
What is Campbell's Theorem in the context of shot noise and compound Poisson processes?
Campbell's theorem calculates the expectation and variance of linear functionals of Poisson point processes. For a signal S(t) = ∑ h(t - t_i) generated by Poisson impulses at times t_i with rate λ, Campbell's formulas state that E[S(t)] = λ ∫_{-∞}^∞ h(u) du and Var(S(t)) = λ ∫_{-∞}^∞ h^2(u) du, providing the mathematical foundation for electronic shot noise analysis.
How does the distribution of arrival times behave conditional on the total arrival count N(t) = n?
Conditional on observing exactly N(t) = n arrivals in the interval [0, t], the arrival epochs t_1, t_2, ..., t_n have the joint distribution of the order statistics of n independent random variables distributed uniformly on [0, t]. This property provides an efficient method for simulating Poisson arrival streams.
Experimental precautions and statistical checks
- Ensure the time increment dt is chosen sufficiently small when simulating continuous arrivals to avoid bin collision artifacts.
- Verify that arrival timestamps are recorded with double-precision floating-point resolution to prevent false event coincidence.
- Account for edge effects when calculating binned arrival counts at the boundaries of the observation window.
- Verify that the sample size (number of time bins) is large enough (N ≥ 100) to ensure sample variance stability.
Theoretical assumptions & limitations
- Real-world physical processes often violate the strict independence assumption due to dead-time effects or refractory periods.
- Non-stationary arrival rates create apparent overdispersion when evaluated with homogeneous Poisson formulas.