Geometric return count before absorption Created 2026-10-14 Updated 2026-11-05
Let be a state outside an absorbing set of a Markov chain. Write for its first hitting time of and . Assume each excursion ends: , and put . Then the number of visits to before absorption, including the initial visit, has the positive-support geometric distribution
After a return, the Strong Markov property restarts the same experiment. Thus and .
The excursion-ending assumption matters: on an infinite state space, an excursion may escape forever without either absorption or return. In that case the geometric parameter for the number of visits is , which need not equal the absorption-before-return probability.
For a continuous-time Markov chain, use the embedded jump chain and assume independent holding times at with exponential distribution of rate , independent of the jump destinations. The total occupation time before absorption is a geometric sum of exponential variables, with exponential distribution of rate and expected value .
Geometric sum of shape-two gamma variables Created 2026-10-07 Updated 2026-11-05
For and , let the summands have gamma distribution with shape two and rate , and let the independent positive-support geometric distribution count have parameter . Put and . The aggregate moment-generating function factors as , so it is the sum of two independent exponential distributions of rates . Its probability density function is for , nonnegative and integrating to one. This contrasts with the geometric sum of exponential variables, which is itself exponential.
For example, and give , , and probability density function . Its mean is , agreeing with the summand mean times the expected count . At the endpoint , the count is exactly one and : the displayed difference quotient must be replaced by its limit , the original shape-two gamma distribution density.
For the exponential distribution of mean , . Substitution into the geometric-sum moment-generating function gives
This is the moment-generating function of an exponential distribution with rate . By the uniqueness theorem for moment-generating functions, the geometric sum of exponential variables therefore has probability density function
Its expected value is , agreeing with in the random sum of independent claims formula.
The moment-generating function of a mixture distribution is the mixture of its component transforms, so
For the aggregate, positivity of every claim implies exactly when . Choose
If , take to have the zero-truncated claim-count distribution, namely the conditional law of given . Then
Choose this count independently of a fresh independent claim-size sequence and put . Its probability distribution is that of conditional on being positive. Thus the hurdle decomposition of a positive random sum gives
and an independent Bernoulli random variable of success probability realizes . This establishes the distributional representation, including that is itself a positive random sum of independent claims. If , the aggregate is identically zero; set and choose any positive , for example one claim. Conditioning the count on positivity is then unnecessary and would be undefined.
For the specified geometric distribution on the nonnegative integers,
An exponential distribution of expected value has transform . Substitution yields
Hence is exponential with rate and expected value . This is the geometric sum of exponential variables with a rescaling of the claim mean. Identification can also use the uniqueness theorem for Laplace transforms of nonnegative random variables by taking .
The resulting distribution function is
Its jump of size at zero is important: the aggregate law is not a purely continuous exponential distribution.
Conditioning on and using the moment-generating function of a unit exponential variable,
where the geometric series converges precisely when . This is the moment-generating function of an exponential variable of rate , so the geometric sum of exponential variables satisfies
Zero-based geometric sum of exponential variables Created 2026-10-14 Updated 2026-11-05
If independent summands have exponential distribution of mean and the independent count has a zero-based geometric distribution of parameter , their random sum of independent claims has an atom at zero and an exponential positive component of rate . Its Laplace transform of a nonnegative random variable is . Thus for . Unlike the geometric sum of exponential variables with a positive count, this law is not purely exponential.