Poisson branching process 2026-10-07
A Galton-Watson process with a Poisson distribution of offspring of mean has this probability generating function. For its branching survival probability satisfies , and . Indeed lies between and . These exact inequalities belong to the Poisson branching process, not to every finite binomial branching process.
Let be a nonnegative integer-valued random variable with finite expected value , and independently label each of its items with either of two equally likely labels. If the two resulting counts are independent random variables, then has a Poisson distribution. Indeed, its probability generating function obeys . With , iteration gives . Finite expected value gives , hence and . Conversely, Poisson thinning produces independent Poisson distributions for the two counts.
Poissonization Created 2026-10-20 Updated 2026-11-05
Poissonization replaces a fixed number of independent observations by an independent Poisson count. It often turns dependent category counts into independent ones. Let labels have probabilities and draw independent of those labels. The category counts then satisfy
Indeed their joint Probability generating function is . This proves both the marginal laws and independence. Conditioning their sum back to recovers the Multinomial distribution for observations.
A rate- Poisson process gives a dynamic version. Independently marking its arrivals by labels splits it into independent Poisson processes of rates . The first arrival of each label is an independent exponential random variable, which simplifies coupon-collection calculations. Returning to a fixed-sample problem requires conditioning or a separate comparison argument; merely changing a clock does not preserve every probability. For an integrable stopping count determined by the labels and independent clock intervals, Wald's equation gives expected elapsed time equal to the expected count divided by .
When each parent has a Poisson number of independently displaced offspring, the conditional offspring process has the displayed directing measure. For disjoint target sets, multinomial distribution splitting of a parent's Poisson count gives joint probability generating function . Multiply over independent parents to obtain independent Poisson counts conditional on the parents. Compactly supported displacement ensures local finiteness for a locally finite parent process.
Poisson thinning Created 2026-09-24 Updated 2026-10-14
Poisson thinning splits a Poisson process by independently marking each arrival as retained or rejected. Let the original Homogeneous Poisson process have rate , and let be a deterministic measurable function of time. Attach independent uniform distributions on , independent of the entire original Poisson process. At arrival time , retain the arrival if . Given all arrival times, these decisions are independent and have the specified retention probabilities. Define the two counting processes using an indicator function:
The Poisson thinning theorem gives independent retained and rejected Inhomogeneous Poisson processes. Their rates are and , respectively, in the sense of the intensity function of a point process. In particular, for ,
The independence is of the two whole processes: every finite collection of retained counts is independent of every finite collection of rejected counts. It is not merely zero covariance for one interval. For constant , the streams are Homogeneous Poisson processes of rates and ; the endpoints include the identically zero stream.[ref]
For an exact toy calculation, take per minute and retain with on and on . In the first two minutes, the retained and rejected expected values are and . In the next two minutes they are and . Over all four minutes, are independent Poisson distributions with mean each, and their sum has mean . Thus
These probabilities and expected values describe the model, not empirical measurements. In the diagram, the original nine synthetic arrivals split into five retained and four rejected arrivals; those realized counts need not equal their expected values.
Figure 1.
Independent marks split a synthetic Poisson arrival path, with exact expected counts for two time-dependent retention probabilities
. Details.
The mechanism explains why mutually exclusive labels can yield independent counts: randomness of the original total offsets the dependence from allocating a fixed total. For constant and an interval with original mean , the conditional retained count has a binomial distribution given . The joint probability generating function is
This factorization proves the two independent Poisson distributions; disjoint original intervals give independent pairs. For varying , the Independent marking theorem for Poisson point processes gives the same conclusion with the two integrated intensities. The joint red and green Poisson count formula gives the corresponding direct calculation with probability mass functions.
Independence here is unconditional. After conditioning on , the identity gives covariance for constant . Thus the retained and rejected counts are dependent for and , even though their unconditional covariance is zero. Poisson thinning can model independent routing of arrivals, or independent detection of arrivals with a known efficiency schedule; using only a retention fraction without checking the marking mechanism can give a wrong count law.
Probability generating function Created 2026-09-24 Updated 2026-10-18
For a nonnegative integer-valued random variable , write . Its probability generating function packages the entire probability mass function into one power series:
This domain is guaranteed, even for complex , because . At the constant monomial is one, including when : the convention here gives . Also . The power series may converge beyond this domain, but that needs an additional argument. Encoding a probability distribution this way turns convolution of independent random variables into multiplication.[ref]
Probability generating function of an independent random sum Created 2026-10-18 Updated 2026-10-25
Let be a nonnegative integer-valued random variable, and let be nonnegative integer-valued independent and identically distributed random variables, with common probability generating function . Require the whole sequence to be independent of , so are jointly independent. Set the random sum
with the empty sum equal to zero. Given , the probability generating function of an independent sum yields the conditional expectation . By the law of total expectation,
The composition is valid because . Merely making the independent of each other does not preserve their conditional laws when depends on them. This is the probability generating function version of transform composition for an independent random sum.
For a Poisson distribution count with mean , substitution into gives a compound Poisson distribution with
If the jumps have finite second moment, conditioning also gives mean and variance . The extra spread comes from randomness in the count as well as randomness in the jumps.[ref]
For an original example, choose and jumps taking values one and two with equal probability. Keep the independence assumptions above. Then
Here , , and . The law of total expectation and law of total variance give
Thus the aggregate is not itself a Poisson distribution, whose mean and variance would coincide.
To recover the probability mass function without repeated differentiation, use and compare coefficients:
For example, . As an independent check, expanding gives
Figure 1.
A random number of one-or-two jumps gives a compound Poisson count
. The upper diagram separates the Poisson distribution count, the independent and identically distributed jumps, and their random sum. Below are calculated coefficients of for , with the omitted tail probability reported explicitly. They are exact model probabilities evaluated numerically, not observed data; the displayed masses have not been renormalized. Details.
The plotted probability masses stop at 20, although the support of a probability distribution is unbounded. The omitted tail probability is
This model has bounded jumps and an unbounded count. Its entire probability generating function does give a finite moment-generating function at every real argument, but the general random sum composition only guarantees the stated domain.
For nonnegative integer-valued independent random variables , factor the expected value of a product:
The Cauchy product explains what this means for individual probability masses:
Every way of splitting the total contributes once. This is convolution of independent random variables; multiplying probability generating functions performs the bookkeeping automatically.
For an original example, take independent and , using the Bernoulli distribution and binomial distribution. Then
so
Direct enumeration of the six pairs gives the same probability masses. For example, total one collects and , with contributions . The coefficients sum to one. The expected value is ; since , the variance is , agreeing with .
Figure 1.
Multiplying probability generating functions accounts for every pair with the same total
. The six cells give the exact joint probability mass function for the chosen independent Bernoulli distribution and binomial distribution. Colors group equal totals; the stacked bars collect their contributions to the coefficients of the product probability generating function. These are calculated model probabilities. Details.
Uncorrelated random variables are insufficient for the product identity. For a concrete counterexample, let be uniform on and set , . Then , but , whereas . Zero covariance controls one mixed moment, not the full joint probability mass function.
Quadratic Galton-Watson extinction bound Created 2026-09-24 Updated 2026-11-05
Let a Galton-Watson process have offspring count with finite variance and expected value . Its branching extinction probability satisfies
To prove the bound, write for the offspring probability generating function and for the branching survival probability. Nonnegative coefficients give on . The integral remainder in Taylor's theorem therefore gives
The Galton-Watson extinction fixed point says . Cancelling and dividing by gives , with . Thus two offspring moments already give a quantitative survival guarantee, without solving the full fixed-point equation.
For and , the expected value is and , so the bound gives . The offspring probability generating function is , whose smallest fixed point is exactly ; this example attains the bound.
Random-sum transform identity Created 2026-10-14 Updated 2026-11-05
For independent and identically distributed random variables independent of a nonnegative integer count , set , including when . Conditioning on the count gives , hence the displayed composition of the moment-generating function with the probability generating function. For positive summands the identity holds for all ; positive requires convergence of the composed series. This permits Laplace transforms of nonnegative random variables to identify the law even when positive exponential moments do not exist.
Redistribution of polynomial factors for dice Created 2026-10-14 Updated 2026-11-05
Encode a fair die with face labels by . Its probability generating function is . Two independent dice with polynomials have the same sum probability distribution as another pair if and all four dice have faces. Redistributing polynomial factors can construct unexpected examples, but each resulting polynomial must have nonnegative integer coefficients, no constant term if labels must be positive, and value at . For example, reallocating from one copy of to the other gives face labels and .
Renewal equation Created 2026-10-06 Updated 2026-11-05
A renewal equation decomposes a process at its first renewal. For returns of a Markov chain to a fixed state, is the first-return probability and the probability of being at that state at time . Their probability generating functions satisfy .
Segregating-site likelihood under the neutral coalescent Created 2026-10-14 Updated 2026-11-05
For , the probability generating function of the segregating-site count conditional on mutation parameter is
This follows by integrating the Poisson generating function over the independent exponential epoch lengths. With , its density is . Expanding the binomial factor and integrating gives
For use , including at . A posterior density divided by a positive prior density is proportional to this likelihood, allowing posterior simulation output to estimate its maximizer.
Shifted-exponential Poisson count recursion Created 2026-10-07 Updated 2026-11-05
For the shifted-exponential Poisson mixture, the displayed relation holds for , initialized by and . It follows by differentiating the probability generating function and comparing coefficients. Nonnegativity follows from the independent Poisson-geometric convolution of independent random variables representation.
Size-biased geometric distribution Created 2026-10-31 Updated 2026-11-05
For the positive-support geometric distribution with success probability , size bias gives . The sum of two independent copies has positive-integer decompositions at value , so its shifted probability is exactly this expression. Its probability generating function is , also obtained by the size-biased probability generating function identity.
Size-biased probability generating function Created 2026-10-31 Updated 2026-11-05
For a positive integer-valued random variable with finite mean , the size-biased distribution has probabilities . They sum to one by the definition of expected value. Multiplying the differentiated probability generating function by gives the displayed identity for , with the endpoint interpreted by its limit.
Thermal light 2026-10-18
For a single mode in thermal equilibrium, its quantum harmonic oscillator energy ladder gives a geometric distribution
Its probability generating function is . Differentiating twice gives , hence . At the chosen mean , , , , , and . Rare high occupations increase the pair count even though the mean matches the other preparations. The geometric weights also arise from a Gaussian mixture of coherent states.[ref]
Figure 1.
Three original one-mode photon-number models with the same mean two and different pair correlations
. The bars are calculated probabilities, not measured data. Thermal occupations above fourteen are omitted without renormalization. The displayed second-order correlations use the complete distributions. Details.
Single-mode selection is essential to this factor of two. If independent equally populated thermal modes are collected together, let with . Adding their variances gives , hence . Averaging over many unresolved modes can make thermal pair counts approach the independent-count reference.
For an almost surely finite total progeny of a branching process, let be the offspring probability generating function and that of the total, including one ancestor. Conditional on children, the total is one plus the sum of independent descendant totals. Independence and the law of total expectation give for , including the empty-product case . If the offspring expected value is , differentiation for gives . The denominator is positive because . Letting and using the monotone convergence theorem on the derivative series proves , establishing finiteness rather than assuming it. For Poisson branching process offspring, .
Start a branching process with one ancestor and offspring distribution for . Its probability generating function is and its mean offspring count is , making it a critical branching process. Splitting the family at the ancestor gives the total progeny recursion . The probability generating function of an independent random sum then gives , whose root satisfying is . Its limit at equals , consistent with finite total progeny almost surely; its derivative diverges there, so the expectation of is nevertheless infinite.
Transform composition for an independent random sum Created 2026-10-14 Updated 2026-10-31
For independent identically distributed summands, independent also of the count , conditioning on gives . The law of total expectation gives the composition of the count's probability generating function with the summand's moment-generating function. This identity is valid as an extended nonnegative expectation; finiteness near zero requires additional exponential-integrability assumptions.