Package jline.api.lossn

Class Lossn_mci

java.lang.Object

jline.api.lossn.Lossn_mci

public final class Lossn_mci
extends Object

Monte Carlo importance-sampling summation for loss networks, after Ross and Wang, "Monte Carlo Summation Applied to Product-Form Loss Networks", Probability in the Engineering and Informational Sciences, 6 (1992), 323-348. Estimates the product-form normalization constant g(C) and per-class blocking probabilities with confidence intervals. Unlike the Erlang fixed point, no link-independence assumption is made and the estimator is consistent.

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static final class	Lossn_mci.Options	Options of lossn_mci(Matrix, Matrix, Matrix, Options), mirroring the fields of the MATLAB options struct.

Method Summary

Modifier and Type	Method	Description
static Ret.lossnMCI	lossn_mci(Matrix nuVec, Matrix Amat, Matrix cVec, int samples, Matrix gammaVec, long seed, double alpha)	Monte Carlo importance-sampling summation for loss networks.
static Ret.lossnMCI	lossn_mci(Matrix nuVec, Matrix Amat, Matrix cVec, int samples, Matrix gammaVec, long seed, double alpha, Matrix revenueVec)	Monte Carlo importance-sampling summation for loss networks, with the revenue sensitivity when a revenue vector is given.
static Ret.lossnMCI	lossn_mci(Matrix nuVec, Matrix Amat, Matrix cVec, Lossn_mci.Options options)	Monte Carlo importance-sampling summation for loss networks.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details

lossn_mci

public static Ret.lossnMCI lossn_mci(Matrix nuVec,
 Matrix Amat,
 Matrix cVec,
 Lossn_mci.Options options)

Monte Carlo importance-sampling summation for loss networks.

Besides the class blocking and its confidence intervals, the result always carries linkLossPoint (JxR), the blocking of route r on link j alone, E(j,r) = 1 - g(C - A(j,r) e_j)/g(C), from the same samples. With options.revenue set (revenue w_k per route-k call and unit time) it also carries the revenue rate W, the revenue sensitivity dW/dnu with delta-method confidence intervals (Ross 1995, Sec. 6.3: dW/dnu_l = Cov(X_l, r(X))/nu_l with r(n) = sum_k w_k n_k), and the route and link implied costs W(C) - W(C - A e_k) and W(C) - W(C - A(j,k) e_j).

Parameters:

nuVec - Offered load per class (1xR).

Amat - Circuit requirement of link j for class r (JxR).

cVec - Link capacity (Jx1).

options - Sampling options; null for the defaults.

lossn_mci

public static Ret.lossnMCI lossn_mci(Matrix nuVec,
 Matrix Amat,
 Matrix cVec,
 int samples,
 Matrix gammaVec,
 long seed,
 double alpha)

Monte Carlo importance-sampling summation for loss networks.

Parameters:

nuVec - Offered load per class (1xR).

Amat - Circuit requirement of link j for class r (JxR).

cVec - Link capacity (Jx1).

samples - Number of Monte Carlo samples.

gammaVec - Importance-sampling parameters (1xR), or null for the Section 3.4 heuristic.

seed - RNG seed; use a negative value for a nondeterministic seed.

alpha - Confidence-interval significance level (e.g. 0.05).

lossn_mci

public static Ret.lossnMCI lossn_mci(Matrix nuVec,
 Matrix Amat,
 Matrix cVec,
 int samples,
 Matrix gammaVec,
 long seed,
 double alpha,
 Matrix revenueVec)

Monte Carlo importance-sampling summation for loss networks, with the revenue sensitivity when a revenue vector is given.

Parameters:

nuVec - Offered load per class (1xR).

Amat - Circuit requirement of link j for class r (JxR).

cVec - Link capacity (Jx1).

samples - Number of Monte Carlo samples.

gammaVec - Importance-sampling parameters (1xR), or null for the Section 3.4 heuristic.

seed - RNG seed; use a negative value for a nondeterministic seed.

alpha - Confidence-interval significance level (e.g. 0.05).

revenueVec - Revenue per call and unit time (1xR), or null.