Class NCResult

java.lang.Object
jline.solvers.SolverResult
jline.solvers.nc.NCResult

public class NCResult extends SolverResult
  • Field Details

    • prob

      public jline.solvers.nc.NCResult.Prob prob
    • lG

      public double lG
    • STeff

      public Matrix STeff
    • it

      public int it
    • pij

      public Matrix pij
    • hitProb

      public Matrix hitProb
    • missProb

      public Matrix missProb
    • spnpf

      public Spn_pf.SpnPfResult spnpf
      Product-form certificate of a Petri net, set by Solver_nc_spn_analyzer.
    • space

      public Matrix space
      Perfect-sampling output of the cftp and cftp.approx methods, null under every other method. spaceAggr holds the distinct sampled states (one row each, stations as columns) and pi their empirical probabilities; space is the same matrix as spaceAggr, since the sampler draws aggregate states. cftpSamples holds the drawn states, one per row (samples x stations), and cftpHorizon the per-sample coalescence horizon (cftp) or number of mixing steps (cftp.approx). Mirrors MATLAB result.cftp.samples / result.cftp.horizon and Python cftpSamples / cftpHorizon on SolverNC.
    • spaceAggr

      public Matrix spaceAggr
    • pi

      public Matrix pi
    • cftpSamples

      public Matrix cftpSamples
    • cftpHorizon

      public Matrix cftpHorizon
    • cacheItemProb

      public Map<Integer,Matrix> cacheItemProb
      Per-item occupancy of each cache node, (nitems x lists+1) keyed by node index; column 0 = miss.
  • Constructor Details

    • NCResult

      public NCResult()
  • Method Details

    • logNormConstAggr

      public double logNormConstAggr()
      Log normalizing constant of the analyzed model, or NaN when none was computed. The analyzers park it in the package-private prob block and leave the flat lG at its zero default, so a caller outside this package that reads the field alone gets 0 for every solve.
      Returns:
      log g(N), or NaN