Package jline.solvers

Class CacheResultRecord

java.lang.Object
jline.solvers.CacheResultRecord
public class CacheResultRecord extends Object
One cache node's average results, as a portable RECORD rather than a write on the node object.

A Cache does not report in Q/U/R/T: it carries its result on the node, which is why a solver that hands its work to another model (SolverENV under mapEnvApprox, say) writes the split onto a model the caller never sees. This record is the hand-back: keyed by node NAME, never by position, so it survives a model whose node order differs.

EVERY FIELD IS OPTIONAL AND ABSENT (null) MEANS NOT COMPUTED, never zero. A recombination that cannot reach a quantity must leave it null so the caller reports it missing rather than reporting a fabricated zero.

See Also:

  • Field Details

    • name

      public String name
      Name of the cache node these results belong to.

    • node

      public int node
      Node index in the model that produced the record, for diagnostics only.

    • hitprob

      public Matrix hitprob
      Actual hit probability per read class [1 x classes], or null.

    • missprob

      public Matrix missprob
      Actual miss probability per read class [1 x classes], or null.

    • delayedhitprob

      public Matrix delayedhitprob
      Actual delayed-hit probability per read class [1 x classes], or null.

    • hitproblist

      public Matrix hitproblist
      Actual hit probability by cache list [classes x lists], or null.

  • Constructor Details

    • CacheResultRecord

      public CacheResultRecord()

    • CacheResultRecord

      public CacheResultRecord(String name, int node)