Package jline.api.sim
This package turns a simulation sample path into an interval estimate. Its
subject is the statistics of the output process, not the queueing model that
produced it, so nothing here takes a NetworkStruct: the input is a
sequence of observations such as successive waiting times exported from a
simulation run.
The steady-state mean is already covered elsewhere: the LDES engine forms batch means and reports confidence-interval half-widths for every average metric, with MSER-5 warmup detection. What this package adds is the steady-state quantile, which is the quantity a tail service-level agreement is written against and for which no interval was previously available anywhere in LINE.
The two entry points are Sim_fquest for a single
sample path and Sim_firquest for independent
replications. Both rest on the standardized time series of the batched
quantile process built by Sim_sts_quantile_areas, and on
two hypothesis tests, Sim_vonneumann for randomness and
Sim_shapirowilk for normality, which are general-purpose
and usable on their own.
- Since:
- LINE 3.1.0
-
ClassesClassDescriptionOutcome of one of the output-analysis hypothesis tests.Point estimate and confidence interval for a steady-state quantile.Procedure constants of
Sim_fquestandSim_firquest.Fixed-sample-size quantile interval from independent replications, FIRQUEST.Fixed-sample-size confidence interval for a steady-state quantile, FQUEST.Shapiro-Wilk test for univariate normality, Royston's AS R94 algorithm.Standardized time series areas of the batched quantile process.Von Neumann ratio test for randomness of a sequence.Normal and Student t quantiles used by the output-analysis routines.Run-length planning for steady-state simulation.Standardized time series statistics of the batched quantile process.