Package jline.solvers.mam.handlers

Class Solver_mam_sync

java.lang.Object
jline.solvers.mam.handlers.Solver_mam_sync
public final class Solver_mam_sync extends Object
The 'sync.faster' and 'sync.slower' methods of SolverMAM: the synchronous approximation of an M/G/c/K station with dual bounds, Qsys_mgck_sync, on the shape applicable(jline.lang.NetworkStruct, jline.solvers.SolverOptions) accepts. 'sync.faster' returns lower bounds on the mean queue length, response time and loss probability, 'sync.slower' upper ones; options.config.sync_phases selects the phase expansion.

Port of MATLAB solver_mam_sync.m and mam_sync_applicable.m.

  • Method Details

    • applicable

      public static Solver_mam_sync.Applicability applicable(NetworkStruct sn, SolverOptions options)
      Can the 'sync.faster' / 'sync.slower' methods answer this model?

      THE RULE IS A SHAPE: the synchronous approximation is a model of one M/G/c/K station, so the network must be a single open class with one Source (Poisson) and one FCFS Queue with a finite capacity and the drop rule, every completion leaving the network. The service law must be one the embedded chain can take exactly: deterministic, or phase type. The synchronous phase expansion (options.config.sync_phases >= 1) needs it deterministic.

      ONE PREDICATE, THREE CALLERS: SolverMAM.supportsModelMethod, the analyzer arm and solver_mam_sync(jline.lang.NetworkStruct, jline.solvers.SolverOptions). It must be given the struct BEFORE SnNonmarkovToPh, which retags the declared law.

    • solver_mam_sync

      public static MAMResult solver_mam_sync(NetworkStruct sn, SolverOptions options)
      Solves the model by the synchronous approximation named by options.method.
      Parameters:
      sn - the model struct, as declared (before SnNonmarkovToPh)
      options - solver options; options.method is "sync.faster" or "sync.slower"
      Returns:
      QN, UN, RN, TN, CN, XN of the M/G/c/K station