Class Dqsys_bernoulli1
Port of matlab/src/api/dqsys/dqsys_bernoulli1.m. Time advances in slots. In the slot starting at t with n jobs present the job in service departs with probability p(n) and an arrival occurs with probability b(n), independently; both are recorded at the end of the slot with the departure resolved first (Daduna's LA rule and D/A rule). The queue length at slot boundaries is a discrete birth-death chain with
pi(n) = [prod_{m=0}^{n-1} b(m) / prod_{m=0}^{n} c(m)]
* [prod_{m=1}^{n-1} q(m) / prod_{m=1}^{n} p(m)] / H
c = 1-b and q = 1-p, which is theorem 2.3, and corollary 2.8 once b(n) = 0
above the capacity. For constant b and p it collapses to the Geo/Geo/1 law of
Dqsys_geogeo1 under the LAS_DA convention.
The law seen by an arriving customer, with himself not counted, is theorem
2.11 and is returned in arrivalPmf. It is not the time-stationary
law: discrete time has no PASTA analogue, and the two differ even when the
arrival stream is a state independent Bernoulli process. In that state
independent case pi_1 is exactly the EAS-convention queue length law of
Dqsys_geogeo1, geometric with ratio
r = b(1-p)/(p(1-b)).
-
Method Summary
Modifier and TypeMethodDescriptionstatic Bernoulli1Resultdqsys_bernoulli1(double[] b, double[] p, int L) Finite buffer, state dependent probabilities.static Bernoulli1Resultdqsys_bernoulli1(double b, double p) Unbounded buffer with constant arrival and service probabilities.
-
Method Details
-
dqsys_bernoulli1
Unbounded buffer with constant arrival and service probabilities. -
dqsys_bernoulli1
Finite buffer, state dependent probabilities. An arrival in a slot that finds L jobs present is lost, which is the loss system of corollary 2.8.- Parameters:
b- offered arrival probabilities,b[n]for n = 0..L, or a single-entry array for a state independent streamp- service probabilities,p[n-1] = p(n)for n = 1..L, or a single-entry array for a state independent serverL- buffer capacity in jobs
-