Class FluidCacheTran

java.lang.Object
jline.solvers.fluid.analyzers.FluidCacheTran

public final class FluidCacheTran extends Object
Transient refined-mean-field cache trajectory for an (open) integrated cache-queueing network. Java analog of the MATLAB solver_fld_cacheqn_tran.m.

For the open-cache topology used by the random-environment mean-field cache aggregation (Source -> Cache -> Sink), each read class arrives at the cache at its known source rate, so no arrival-rate fixed point is needed: the per-cache isolated inputs (gamma, m, lambda_cache) are built directly from the stage struct, and the RMF drift is integrated over the window from an optional per-cache initial occupancy (used to carry the mean occupancy across environment switches).

  • Method Details

    • tran

      public static CacheTranResult tran(NetworkStruct sn, double time, int nPoints, double[][] xinitPerCache)
      Integrate the transient RMF cache trajectory for all cache nodes.
      Parameters:
      sn - stage NetworkStruct (must contain at least one Cache).
      time - end time of the transient window (start is 0).
      nPoints - number of uniform grid points (>= 2).
      xinitPerCache - optional per-cache initial occupancy; entry may be null for the default initial state, and the array itself may be null.
      Returns:
      CacheTranResult.
    • tran

      public static CacheTranResult tran(NetworkStruct sn, double time, int nPoints, double[][] xinitPerCache, Matrix holdD0, Matrix holdD1)
      As tran(NetworkStruct, double, int, double[][]), and with a holding-time MAP {holdD0, holdD1} also the per-cache average of the transient against that holding time, int x dF(t/Lam) / int dF(t/Lam) over the window, the drift's per-request time mapped to real time by the cache's total request rate Lam. time is then REAL time: the drift is integrated over Lam*time and the returned grid is in real time. The average is integrated with the drift (CacheSojourn), so it does not depend on the nPoints grid.