Package jline.api.pfqn.mva
Class Pfqn_chow
java.lang.Object
jline.api.pfqn.mva.Pfqn_chow
Chow Second Approximation (SA) approximate MVA.
W.-M. Chow, "Approximations for large scale closed queueing networks", Perform. Eval. 3(1), 1983. The arrival-instant queue length is written exactly as
A_k^(c)(N) = Q_k(N - 1_c) = Q_k(N) (1 + theta_ck), theta_ck = [Q_k(N - 1_c) - Q_k(N)] / Q_k(N),
and the theta-terms are estimated ONCE, off the Bard LCP solution, before
the fixed point is run. Two estimators are given: the BACKWARD one uses
Qhat(N - 1_c), the FORWARD one Qhat(N + 1_c). Chow reports the forward form
to be the more accurate of the two, so it is the default here. Setting every
theta to zero recovers Pfqn_lcp.
-
Field Summary
Fields -
Method Summary
Modifier and TypeMethodDescriptionstatic Ret.pfqnAMVAstatic Ret.pfqnAMVAstatic Ret.pfqnAMVAstatic Ret.pfqnAMVAstatic Ret.pfqnAMVApfqn_chow(Matrix L, Matrix N, Matrix Z, double tol, int maxiter, Matrix QN0, SchedStrategy[] type, String variant)
-
Field Details
-
FORWARD
Estimator of the theta-terms.- See Also:
-
BACKWARD
Estimator of the theta-terms.- See Also:
-
-
Method Details
-
pfqn_chow
-
pfqn_chow
-
pfqn_chow
public static Ret.pfqnAMVA pfqn_chow(Matrix L, Matrix N, Matrix Z, double tol, int maxiter, Matrix QN0) -
pfqn_chow
public static Ret.pfqnAMVA pfqn_chow(Matrix L, Matrix N, Matrix Z, double tol, int maxiter, Matrix QN0, SchedStrategy[] type) -
pfqn_chow
public static Ret.pfqnAMVA pfqn_chow(Matrix L, Matrix N, Matrix Z, double tol, int maxiter, Matrix QN0, SchedStrategy[] type, String variant)
-