A Continuous-time Markov chain on a finite or countable state space is specified by transition rates for , with finite total exit rate . At a state with , it waits an exponential time of rate and then moves to with probability . A state with is absorbing: it has infinite holding time rather than a zero-rate exponential waiting law.
Its transition-rate matrix is , with off-diagonal entries and diagonal entries , so each row sums to zero. The rates describe first-order transition probabilities: for . On a finite state space, the transition semigroup is and explosion is impossible. On a countably infinite state space, finite individual exit rates do not alone exclude infinitely many jumps in finite time; nonexplosion must be checked separately.
For two states with rates and , the mean holding times are and . Detailed balance gives Stationary distribution , andThis starts at one and relaxes to the stationary mass of state zero. Changing holding rates can preserve an embedded jump chain while changing the proportions of elapsed time spent in its states.
A continuous-time Markov chain is nonexplosive if it makes only finitely many jumps on every bounded time interval. A uniformly bounded total jump intensity permits domination by a Poisson process. More generally, if a population process stays in a finite random set of states on each finite interval and every individual state has finite exit rate, localizing to fixed finite sets gives the same conclusion. In an open migration process, external Poisson immigration bounds the total population while internal moves preserve it and exits reduce it, giving this localization.
A Markov jump process is a nonexplosive, piecewise-constant continuous-time Markov process. Transition rates give exponential holding time of rate and next-state probabilities proportional to these rates. Being Markov relative to a specified filtration requires the conditional future law given that filtration to depend only on the present state.
// Target: probability-and-statistics.bigb
A Markov chain with transition kernel , sampled at an independent Poisson process of rate , is a Pseudo-Poisson process. Its infinitesimal generator is and its transition semigroup is . Conditioning on proves the latter formula, and differentiation at zero gives the former. Its resolvent operator is for , by integrating the nonnegative scalar Poisson weights. The series converge in operator norm on bounded Borel measurable functions.
This finite-state Markov jump process models two opposing populations, each suffering losses at a rate proportional to the surviving opposing population. The displayed rates apply when and ; both coordinate axes are absorbing. Its embedded jump chain removes one individual at each step, with probabilities and . Before the first population is exhausted, the corresponding deterministic equations , conserve ; finite stochastic populations satisfy a near-invariant instead.
In the jump chain of the symmetric stochastic attrition process, the two possible increments of are and . Their mean is the displayed quantity, bounded by one in absolute value. Subtracting these conditional means produces a martingale. If both initial populations are at most , it has at most nonzero increments, each bounded by . Thus Azuma-Hoeffding inequality controls changes of order with exponentially small probability of order .
Start the symmetric stochastic attrition process at , with , and let be the final first-population size divided by . The bound holds for all sufficiently large , depending on . At absorption, one coordinate is zero. If the first survives, the terminal squared difference is ; if it is extinguished, that difference is nonpositive. In either case the indicated deviation forces an error of at least from the limiting squared invariant. The initial rounding error and total conditional drift are only . Apply Azuma-Hoeffding inequality to the centered squared-population near-invariant for stochastic attrition, using its increments bounded by , to obtain the displayed exponent.
Let be a rate-one exponential random variable independent of a uniform random variable on . In its natural filtration, this Markov jump process jumps once from zero to its mark and then stays there. Its destination rate kernel is . For a Borel mark set , the martingale follows directly from the memoryless residual lifetime. Information about the future mark or lifetime must not be inserted into the initial filtration.
A kernel specifying the jump rate from state for each displacement . With the displacement convention the infinitesimal generator is . The destination convention instead records a rate measure on the new state, translating each displacement by . For finite positive total rate, that total is the exponential holding rate and the normalized measure is the next-jump distribution.
With the displacement convention for a Lévy kernel of a Markov jump process, suppose its total rate is bounded, its first moment is zero and its second moment is bounded by . The bounded rate prevents explosion. The compensated displacement integral is a square-integrable martingale with . The displayed process is the sum of and the compensated sum of squared jumps. The first term is square integrable since ; the second is integrable because its expected total jump-square sum is at most . Consequently the square compensation is a true martingale, without a fourth-moment assumption.
For a Markov jump process with off-diagonal rate kernel , the jump measure records the jump time and the destination state. Relative to a filtration for which the process has the Markov jump property, its predictable compensator is the displayed rate measure, up to explosion. The corresponding generator is . An enlarged filtration revealing future jumps need not preserve this compensator, so the filtration condition is essential.
The predictable random measure matching the expected integrals of the jump measure against every nonnegative predictable integrand. For displacement marks of a Markov jump process, . Multiplying the integrand by an event known before an interval proves that the compensated integral has conditionally centered increments; expected absolute integrability makes it a true martingale. With the compensator absolutely continuous in time, squared-integrand integrability gives the square-integrable compensation isometry.
A predictable mark-dependent integrand gives a true martingale whenever on every finite horizon. Local versions follow when this expected integrability holds after localization. The compensator identity applied to predictable indicators from the past proves that every increment has conditional expectation zero. Choosing gives the generator martingale under the same local integrability conditions.
With scale , such a process has finitely many possible jump vectors occurring at rates depending on the current scaled state. Its drift is , while the predictable covariance rate of its martingale fluctuations is .
// Target: probability-and-statistics.bigb
In a population of playing rock paper scissors, let losers copy winners. With pair-selection events of rate , the proportion process has jumps , , at rates , , . Its fluid limit obeys , with indices interpreted cyclically. The product of numbers is conserved in this ordinary differential equation, whereas the finite chain eventually reaches a monochromatic absorbing state.
For the cyclic imitation chain, its infinitesimal generator satisfies . Therefore . While all three types remain, the product of numbers is at least for , so this identity proves that one type is lost almost surely. The remaining pair has a fixed winner, which eventually occupies the whole population. The conserved product in the deterministic fluid limit therefore describes fixed finite horizons, not the chain's infinite-time behavior.
With jumps at rates , compensate each jump count by its integrated rate. The drift is and the predictable quadratic variation matrix of the martingale term is . On a compact region with bounded rates this bracket is of order . A Lipschitz drift and the Gronwall inequality transfer martingale concentration to the fluid limit.
For jump rates when , assume and . The positive-rate states form the unique Closed communicating class; every zero-rate state is an open singleton. The stationary law is , with Detailed balance . When at least two rates are positive, the embedded jump-chain law is proportional to . With only one positive rate, that state is absorbing and the genuine sequence of jump times stops there; an extended discrete chain can use an absorbing self-loop.
Consider states for integer levels and strictly positive rates . From the chain moves to at rate and to at rate . From , , it moves to at rate or to at rate . At the reflecting boundary there is only the switch to ; no negative level is allowed. All rates are bounded, so the chain is nonexplosive.
Let be the probability of hitting , with . First-step equations arewith the first equation also valid at . Put . If , then andSubstitution verifies both equations and the boundary. Eliminating the variables gives a second-order recurrence with modes . The displayed decaying solution is the minimal nonnegative harmonic solution and hence the Hitting probability.
When , the growing mode is excluded by boundedness; at equality the second mode is linear and is likewise excluded. The only bounded solution with the boundary value is one, proving almost-sure hitting. Since the specified reflected chain is irreducible, these cases classify it as recurrent or transient respectively. Specifying the boundary rule is essential for that recurrence statement; making absorbing would be a different chain.
A continuous-time Markov chain with finite holding rates is nonexplosive when its jump times have no finite accumulation almost surely. It then remains in its specified state space at every finite time, rather than being lost to a cemetery state after infinitely many jumps. Bounded rates are sufficient for nonexplosion, but finite rates at individual states alone are not.
For a nonexplosive continuous-time Markov chain on the nonnegative integers, suppose zero is absorbing and each positive state has positive rate to zero. On every finite positive set the conditional chance of absorption at the next jump has a positive minimum. By the Strong Markov property, the chance of surviving departures from that set is bounded by . Without absorption only finitely many such departures occur, and finite positive holding rates prohibit staying forever at a positive state. Thus eventually the path stays outside each finite set: it escapes to infinity. This conclusion needs no upward skip-free assumption.
An Equilibrium distribution of a continuous-time Markov chain is a probability row vector invariant under its transition semigroup:It satisfies when the generator sums are defined. An invariant law is a property of the entire evolution, so formal generator equations must be interpreted with the process's nonexplosion and domain conditions.
For a finite irreducible chain, the equilibrium distribution is unique and every transition row converges to it. For a general finite chain, the limiting row from state is a mixture of the stationary distributions of its closed irreducible classes:The chain eventually enters one of these classes and then mixes within it. If every closed class is a single absorbing state, this reduces to absorption probabilities at those states. Merely having some absorbing states does not exclude other closed classes.
Holding times explain the link to an embedded jump chain. For a finite irreducible chain with strictly positive exit rates and jump-chain Stationary distribution , the time-stationary distribution isVisit frequency is multiplied by mean time per visit. In the countable setting, these time weights need a finite positive normalizer and a nonexplosive process before they define an equilibrium probability. For an alternating two-state jump chain, and rates give : slower holding rates increase time occupancy.
For a finite-state continuous-time Markov chain, a transition intensity matrix has nonnegative off-diagonal rates and . Every row sums to zero. In the time-homogeneous case its transition probability matrix is the matrix exponential . An absorbing state has a zero row.
For a finite irreducible Markov chain in continuous time, time averages converge almost surely to averages under its unique stationary distribution, from any initial state. Applying this to an activation indicator yields the corresponding throughput.
A continuous-time Markov chain is nonexplosive when it makes only finitely many jumps in every finite time interval, almost surely. Uniformly bounded total jump rates suffice. Another useful criterion bounds the states visited over each finite interval by a random finite set on which rates are bounded.
Global balance equates total stationary incoming and outgoing transition intensities at each state. Detailed balance for a continuous-time Markov chain is sufficient but stronger: it equates each pair separately.
A jump process with event types , increments , and rates can be represented as , with independent unit-rate Poisson processes. This represents the Markov jump-process generator through time-changed counts.
Place one distinct label at each vertex of a finite connected weighted graph. At rate , interchange the labels at and . This continuous-time Markov chain has a uniform distribution on a finite set as its stationary distribution. The position of one label is the random walk with the same rates.
On a finite connected weighted graph, the interchange process and the single-label random walk with the same symmetric rates have equal spectral gaps. This is a substantial theorem, not just the elementary inclusion of the single-label spectrum in the full spectrum. A proof is given in Proof of Aldous' spectral gap conjecture, Theorem 1.1.
For an equilibrium chain with strictly positive stationary probabilities, reversed jump rates are . Under detailed balance they equal the original rates, making the stationary path law reversible. Holding times and jump directions reverse together; this is more than symmetry of one-time distributions.
A partition-valued Markov chain in which every unordered pair of current blocks merges at rate one. On initial singleton blocks, the holding time with blocks is exponential with rate . Restrictions to subsets of labels have the same law.
In Kingman's coalescent, the epoch with ancestral lineages lasts , independently across epochs. If each lineage carries an independent Poisson process of mutations with rate , its epoch mutation count satisfies . The total mutation count is . Under the infinite sites mutation model, this also counts segregating sites; without that assumption it still counts mutation events, which need not correspond one-to-one to observed polymorphisms.
The independent mutation counts in coalescent epochs have the geometric mutation count in a coalescent epoch laws. Summing their expectations and variances gives the displayed formulas, where and . The extra quadratic term records randomness of the ancestral tree, beyond the conditional Poisson distribution variance.
Integrating the Poisson distribution probability generating function over the exponential distribution of the epoch length givesThus has a geometric distribution on , with expectation and variance . The support convention matters: the count includes zero mutations.
The epoch lengths in Kingman's coalescent are independent random variables. Conditional on these lengths, mutation counts on disjoint tree pieces are independent Poisson random variables. Their conditional laws depend only on the corresponding epoch length, so integrating the product of conditional laws against the independent epoch laws leaves independent unconditional epoch mutation counts. This justifies adding their variances, rather than assuming independence from conditional independence alone.
For , in coalescent units, independent waiting times satisfy , and total branch length is . Therefore . The variables are independent exponentials with rates , so has the same distribution as the maximum of independent unit exponentials, with density . The last identity follows either from its Laplace transform or from the independent spacings of exponential order statistics.
The consistent partition process on whose finite restrictions are Kingman's coalescent. It starts with infinitely many singleton blocks.
For , the number of blocks is finite almost surely. On a finite restriction, the expected time to reach blocks is at most . Markov inequality, followed by increasing the restriction, gives . Let , then use rational times and monotonicity to obtain the assertion simultaneously for every .
The entry count counts jumps of a continuous-time Markov chain into state by time , excluding initial occupation. It is a counting process. For a finite-state chain with homogeneous transition intensities, its mean has derivative conditional on initial state , so its expected value is the sum of incoming rates times the expected occupation times of a continuous-time Markov chain. This density is an expected count rate, not necessarily a normalized probability density function.
The occupation time records the duration spent in a state of a continuous-time Markov chain. Conditional on , its expected value is , where is the transition probability.
For a continuous-time Markov chain with absorbing endpoints and rates , the jump chain is a simple symmetric random walk. On leaving , absorption before returning has probability , by gambler's ruin on each side. The geometric return count before absorption and independent rate- holding times give . The rates at other states affect elapsed excursion times but not their return probabilities.
For a finite-state homogeneous continuous-time Markov chain with transition intensity matrix , integrate the indicator of occupying state over . Interchanging the bounded time integral and expected value gives the displayed integral of transition probabilities. It counts all visits, not only the first holding time. Summing over states gives . A zero eigenvalue of does not cause a problem for this integral, although it prevents casually writing it with an inverse matrix .
The holding time of a continuous-time Markov chain is the time spent in its current state before the next jump. For it has an exponential distribution of total outgoing rate ; at an absorbing state with it is infinite.
The holding rate at state of a continuous-time Markov chain with generator matrix isFor a conservative chain with finite , its holding time has exponential distribution: and . Conditional on a jump, the next state is with probability . Thus the rate determines when departure occurs, while normalized outgoing rates determine where it goes.
If the outgoing rates are and for two distinct other states, the total rate is five, the mean holding time is , and departure chooses them with probabilities and . A zero rate gives an absorbing state with infinite holding time. Statewise finite rates alone do not rule out explosion of a continuous-time Markov chain.
A state's holding time in a time-homogeneous continuous-time Markov chain is exponentially distributed with rate . Its mean is ; the probability that its next jump is to state is . An absorbing state has infinite holding time.
For an exponential distribution with positive rate , the mean is . A confidence interval with for the rate transforms to one for the mean by reversing and inverting its endpoints. The monotonic transformation preserves the coverage event; it does not give a prediction interval for one realized holding time. A zero lower rate endpoint corresponds to an infinite upper mean endpoint.
Use the master-equation convention that is the transition rate from to . In a stationary law , the mean directed transition count per unit time isThe departure-state occupation multiplies the departure-to-destination rate. The reversed rate index would describe a different transition. A physical event transporting signed charge in the forward direction has mean charge current .[ref]
Stochastic thermodynamics, fluctuation theorems, and molecular machines by Udo Seifert. arXiv:1205.4176v1, Section 6.1.1, equations (110)–(111): rate convention and probability current
For a closed finite chain whose complete transition graph is a tree, every stationary edge current is zero. Removing one edge separates the states into two sets; stationarity makes total probability flow across that cut vanish, and the removed edge is its only crossing. Both directed counts can nevertheless be positive. On a three-state path with symmetric adjacent rates , the stationary law is uniform and the two edge traffic rates are and in each direction.
Different physical transition mechanisms between the same states should be kept as separate edges of a multigraph when their currents matter. Summing their rates produces an aggregated edge. Opposing mechanism currents can cancel in that aggregate, so zero aggregate current does not force every mechanism's current to vanish.
For a finite Continuous-time Markov chain, let be the rate from to , and let be its current state probability. Define forward traffic and current . The total entropy production rate, in dimensionless entropy units with natural logarithms, isThe factor avoids counting an unordered pair twice. The expression is the same as after pairing opposite directions.[ref]
Stochastic thermodynamics, fluctuation theorems, and molecular machines by Udo Seifert. arXiv:1205.4176v1, Section 6.2.3, equation (128): ensemble entropy production
Nonnegativity is elementary: for , the signs of and agree. A pair with both flows zero contributes zero. If exactly one flow is positive, its contribution is , the corresponding one-sided limiting convention. The rate vanishes exactly when every pair has equal opposing traffic at that instant, which is Detailed balance for that Probability distribution.
For a stationary three-state cycle with clockwise rate and anticlockwise rate , the stationary law is uniform. Each edge has current , so . It is zero at equal rates and positive otherwise, even though the state probabilities no longer change. Stationarity balances total flow at each state; it need not eliminate circulating currents.
A Continuous-time random walk waits a random time at each site before jumping. Constant outgoing rates produce an exponential residence time, while the relative transition rates determine the destination of the next jump.
Particles perform independent continuous-time random walks until they meet, after which they share one path. A finite initial particle set remains finite and its cardinality cannot increase. They arise as the common ancestors of sites in a voter model. On the square lattice every pair eventually coalesces, because the difference walk is recurrent Markov chain.
Let be an adapted Càdlàg process on a finite state space, and let be a conservative transition-rate matrix. It solves the martingale problem for when, for every real function on the state space,is a martingale for the specified filtration. On a finite state space all such test functions and their generator values are bounded.
A Continuous-time Markov chain with generator has this property by Dynkin's formula. Conversely, fix a terminal time and put for . The backward equation gives . Applying the martingale identities to its finitely many time-dependent coefficients shows is a martingale. ThusTesting state indicators recovers the entire conditional transition law, proving the Markov property and identifying its semigroup. The càdlàg and filtration assumptions make this a path-process characterization, rather than a statement only about separately fixed-time versions.
A Q-matrix has nonnegative off-diagonal entries and row sums zero. Its diagonal entry is the negative total rate out of the state.
For a time-homogeneous continuous-time Markov chain, and the matrices satisfy , the Chapman-Kolmogorov equation. On a finite state space with Q-matrix , the matrix exponential gives .
The distinction from a discrete-time transition matrix is the elapsed-time parameter: describes where the process is after time units, allowing any number of intervening jumps. Each is a stochastic matrix, with . A Q-matrix instead records instantaneous rates and has row sums zero. Sampling at a fixed spacing yields a Discrete-time Markov chain with transition matrix . See Hao Wu's MIT lecture 21, transition semigroups and lecture 22, infinitesimal rates.
Sample a finite-state continuous-time Markov chain with generator at the independent arrival times of a rate- Poisson process. The sampled states form a Discrete-time Markov chain with transition matrixIts invariant distributions are exactly those of the continuous-time chain.
For a finite-state continuous-time Markov chain with Q-matrix , the backward and forward equations are and , respectively.
A Markov reward model assigns a reward or cost rate to each state of a Continuous-time Markov chain and accumulates it with elapsed time:For a finite chain with generator and reward column vector , its expected accumulated reward isIntegrating state indicators proves the expectation formula; differentiating the matrix exponential gives the differential equation. Indicator rewards count time in a selected state. If rates represent power consumption, the accumulated reward is energy.
For a two-state chain at rate and at rate , let only state one earn unit reward. Starting at zero,It is the expected duration in state one, obtained by integrating its occupancy probability. With exponential discount , the infinite-horizon expected reward instead is , the resolvent applied to . Bounded rewards and positive discount make that integral finite even when the undiscounted cumulative reward grows indefinitely.
For transition rates , the generator acts on a function byFor a test function in the generator domain, under the integrability conditions for Dynkin's formula and differentiation,Choosing polynomial functions gives differential equations for the moments.
For a positive function in the domain of a Markov jump-process generator, the generator martingale for and semimartingale integration by parts give the displayed local martingale. Suitable finite-horizon deterministic bounds on and make it a true martingale. For a single marked exponential jump process, take . Then the exponent's compensating process is , where for and ; it is continuous adapted and hence predictable.
For jumps of rates , the forward operator acts on a Probability mass function byIt is adjoint to the Markov jump-process generator .
Fix a finite embedded jump path whose holding rates are positive and finite. Conditional on those successive states, the complete holding times are independent exponential random variables with the corresponding rates. To observe exactly these jumps by a fixed time , the final holding time is censored rather than completed.
Write for this conditional holding-time probability. With ,The completed holding times supply their density factors; the last portion supplies only an exponential survival factor. The fixed observation time imposes the simplex constraint.
Multiplying by the terminal rate completes the symmetric product of all rates. Reversing the time portions and the state sequence preserves the simplex integral, with unit absolute Jacobian. HenceThe right-hand holding kernel can be defined by this formula even if the reverse embedded path is not feasible; its conditional path interpretation requires that feasibility. For actual unconditioned path probabilities, the holding kernel must also be multiplied by the product of jump-chain transition probabilities. Those products need not agree under reversal, so the holding identity alone is not a reversibility theorem for the chain.
For , the fixed-time skeleton has matrix . If is the holding rate, then for ,Thus diverges exactly when does, proving recurrence equivalence for an irreducible chain.
The jump chain records the successive states visited by a Continuous-time Markov chain while discarding its holding times.
On , let the chain jump from every to with probability and to with probability , and jump from zero to one. It is recurrent exactly when ; at it is null recurrent.
For strictly positive finite holding rates , a measure satisfies exactly when is invariant for the jump chain. Thus when normalization is possible.
Even when the jump chain has invariant probability , the continuous-time invariant measure may have infinite mass if holding rates become too small, producing infinite mean return time.
A recurrent jump chain with infinite invariant measure can yield a positive-recurrent continuous-time chain when .
If the jump chain visits a state of finite positive holding rate infinitely often, the sum of its independent positive holding times at that state diverges almost surely, preventing explosion.
For a continuous-time Markov chain with strictly positive finite holding rates whose jump times tend to infinity, the chain and its jump chain have the same directed paths of positive-probability jumps. They are therefore irreducible simultaneously, and a state is recurrent for one exactly when it is recurrent for the other: the event of returning after departure depends only on the sequence of visited states.
Explosion is determined by whether the successive holding times of the continuous-time Markov chain have a finite sum.
If every holding rate of a continuous-time Markov chain is at most , its holding times can be coupled as , where the are independent rate-one exponential random variables. The strong law of large numbers gives almost surely, so the jump times tend to infinity and the chain cannot explode.
Conditional on a jump path , if , then the sum of exponential holding times has finite expectation and is finite almost surely.
Suppose a transient nearest-neighbour jump chain on has uniformly bounded expected visits to each state and the holding rate at grows geometrically. Thenso the continuous-time chain explodes almost surely.
For a finite-state chain with transition semigroup , a probability row vector is invariant when for every . Equivalently,
A countable nonexplosive continuous-time Markov chain which is an irreducible Markov chain and has an invariant distribution is positive recurrent; its transition probabilities tend to from every initial Markov chain state.
To see why continuous time removes periodicity, fix . Every finite sequence of Markov chain states connected by positive transition intensities has positive probability of finishing before , followed by no further transition until . Consequently the transition matrix has positive entries between every pair of communicating Markov chain states, including its diagonal. It is an irreducible Markov chain and an aperiodic Markov chain with stationary distribution . The countable-state Markov chain convergence theorem gives convergence at times , also in total variation distance. For , , applying contracts total variation distance and preserves , giving the full continuous-time limit.
An invariant distribution of a continuous-time Markov chain whose coordinates are independent Poisson distributions. This property needs verification from the reaction propensity functions; it does not follow merely from having an equilibrium for the deterministic reaction-rate equations.
For example, consider the cyclic chemical reaction network with stoichiometric vectors , and and respective reaction propensity functions , , , where all three constants are positive. Put and . For , the three incoming terms in the chemical master equation are respectively , and . They exactly cancel the outgoing terms. Boundary terms vanish by the zero negative-index convention, and normalization follows from the two Poisson distributions. Establishing a limiting probability distribution additionally requires nonexplosion, an irreducible Markov chain and positive recurrence.
Every Closed communicating class of a Finite Markov chain has its own invariant Probability distribution. All invariant distributions are convex combinations of those class distributions. In particular, a finite irreducible chain has a unique invariant distribution, while a reducible chain may have many.
For an explosive chain, a normalized solution of the formal balance equations need not be an invariant probability for the minimal transition semigroup. Explosion sends Probability mass to a cemetery state, and the generator equations on the original states do not include that boundary loss.
A concrete example uses the reflected jump chain on which moves from zero to one, and from moves up with probability and down with probability . Give state holding rate . The embedded chain has invariant weights , for . Dividing by the holding rates and normalizing yieldsThese weights even satisfy formal Detailed balance: , and for both opposing edge flows equal . Thus holds at each state.
Nevertheless the process explodes. Starting from zero, the embedded walk has expected visit count two at zero and three at every positive level. The zero-state return probability is ; at a positive state it is , giving those geometric visit means. Therefore the expected total elapsed holding time isThe lifetime is finite almost surely. The same bounded expected-visit argument works from every initial state, so the killed minimal semigroup loses all its mass as time tends to infinity. The displayed is consequently a formal generator-balance solution, not a stationary probability of that killed evolution. A conservative continuation after explosion would require an extra boundary rule.
For integer batch size and rates , a Batch-birth linear-death process has transitionsWriting and taking for , its Master equation is
With finite initial second moment, for and , the Markov jump-process generator applied to and givesFor , the limiting moments are
A birth-death process moves only between neighbouring nonnegative integer states, with state-dependent birth and death rates.
A pure birth process on jumps only from to , at a strictly positive rate . Its successive holding times are independent exponential variables of rates .
For a pure birth process with fixed initial state and finite statewise rates, the jump-count compensator gives , so is a local martingale up to explosion. If , a rate- Poisson candidate clock with accepted jumps of probability prevents explosion and dominates the jump count. Its finite exponential moments dominate every stopped on each finite horizon, upgrading the local martingale to a true martingale. A random initial state requires ; uniformly bounded rates do not supply this missing initial integrability.
The Logistic cell-division process is a density-limited pure birth model on , where is a positive integer carrying capacity. With cells, each divides at rate , so the total rate for is . Zero and are absorbing; a positive initial population can only increase until it reaches .
Let and assume in probability, for example . On every fixed finite interval , converges uniformly in probability toAn explicit martingale argument verifies this instance of the Kurtz fluid limit theorem. Compensating the jumps gives , with predictable quadratic variation at most . Doob's L2 maximal inequality gives . Since is Lipschitz on , Gronwall's inequality bounds the trajectory error by , which tends to zero in probability.
The finite-population mean need not itself solve the logistic equation: . The fluid limit follows because the fluctuations vanish on the specified finite horizons.
// Target: analysis.bigb
A pure birth process is nonexplosive exactly whenIts explosion time is the sum of independent exponential holding times. If the displayed series converges, that sum has finite expectation. If it diverges, the Laplace transform of the sum isforcing the sum to be infinite almost surely.
The pure birth process with rates , where , combines linear reproduction with immigration. Starting from zero, its mean satisfiesand hence
Add independent rate- resets to zero to a linear birth process with immigration. The resulting irreducible continuous-time Markov chain is nonexplosive by domination of births and finiteness of Poisson reset counts. When , an excursion from one reaches with probability , so its jump chain is recurrent with infinite expected jump count. The elapsed return-cycle time instead has meanThus the continuous-time chain is positive recurrent while its jump chain is null recurrent. Its stationary probabilities are and for .
The Yule process is the pure birth process with rate in population state . Starting from one individual, it is nonexplosive because the independent holding times have divergent total almost surely.
For , given , the birth times in a Yule process have the law of the order statistics of independent variables with densityThis follows by dividing the joint holding-time density by
.
.
For individual birth and death rates and one initial individual, the finite-time extinction probability satisfiesWhen ,
For birth rates and death rates , a reversible measure satisfiesThus , and it is an invariant distribution when the resulting measure is summable and the chain is nonexplosive.
For per-capita birth and death rates and , the rates are and . If and , thenThe expected value sees only the net growth rate , whereas the variance also sees the total event rate .
For a nonnegative integer-valued random variable, the displayed inversion is valid when . Indeed the gap between consecutive Jordan-Bonferroni exact-occurrence inequalities is , which tends to zero under this hypothesis. No exchange of an infinite signed sum with an expectation is needed.
If nonnegative integer-valued random variables have all fixed factorial moments converging to , then their convergence in distribution is to the Poisson distribution with mean . The first-moment bound gives uniform tightness, and bounds on higher moments give uniform integrability of each fixed lower moment. Every subsequential limit therefore has the indicated factorial moments. For such a limit and , the nonnegative binomial expansion and the monotone convergence theorem give . This identifies its moment-generating function as that of the Poisson distribution, uniquely determining the limit.
Poisson moment closure approximates a nonnegative integer-valued population law at each time by a Poisson distribution with its current mean . Its factorial moments then become ; in particular and .
For example, consider birth rates and death rates , with and . Assuming the relevant moments are finite, the exact first-moment equation isPoisson closure gives the Riccati equation . The exact identity shows what is discarded: the variance correction . Nonlinear rates do not generally preserve a Poisson distribution.
For the Batch-birth linear-death process, a Poisson initial law does remain Poisson. independent survival thins the initial Poisson population, and surviving immigrants form an independent Poisson population; their sum is Poisson. That gives an exact linear example, rather than making the nonlinear closure exact.[ref]
Lectures on the Poisson Process by Günter Last and others. Chapter 5: marking, thinning and superposition
A birth-death process with unit upward and downward jumps and rates and has infinitesimal generatorIf varies slowly over one jump, Taylor expansion through second order gives . Thus its diffusion approximation has drift coefficient and, in the conventiondiffusion coefficient . Equivalently it solves , where is Brownian motion. The sum measures the variance of the short-time increment: births and deaths both add noise even when their mean drifts cancel.
For a population density with jump size and birth/death rates , the drift is and the diffusion coefficient is . Fluctuations therefore shrink as . A useful approximation requires many reaction events while the rates change little; it can fail near an absorbing state or when only a few individuals remain. A continuous diffusion can even visit negative values unless its boundary treatment is specified. The approximation replaces the discrete jump model rather than asserting equality of their laws.[ref]
Stochastic Simulation of Chemical Kinetics by Daniel T. Gillespie. Section 4, printed pp. 44–46, equations (18)–(20), leap conditions and system-size scaling
Articles were limited to the first 100 out of 123 total.
Articles by others on the same topic
There are currently no matching articles.