Package jline.solvers.mva

Class MVAResult

java.lang.Object

jline.solvers.SolverResult

jline.solvers.mva.MVAResult

public class MVAResult
extends SolverResult

Result container for Mean Value Analysis (MVA) solver computations.

MVAResult extends the base SolverResult to include MVA-specific performance metrics and intermediate computation results. This includes normalizing constants, iteration counts, and cache-specific probabilities.

Since:

1.0

See Also:

• SolverMVA
• SolverResult

Field Summary

Fields

Modifier and Type	Field	Description
Map<Integer,Matrix>	cacheItemProb	Per-cache per-item occupancy [nitems x (lists+1)], keyed by node index (used by the cache+queueing analyzer to populate getAvgItemTable)
Boolean	converged	Whether the fixed point met its tolerance, or null when the handler does not report one.
Matrix	hitProb	Cache hit probabilities [items x classes] (used by cache analyzers)
int	iter	Number of iterations performed by the MVA algorithm (used by runAnalyzer method)
double	logNormConstAggr	Logarithm of the aggregate normalizing constant
Matrix	missProb	Cache miss probabilities [items x classes] (used by cache analyzers)

Fields inherited from class jline.solvers.SolverResult

AN, ANfcr, CN, CNt, DropRateJoin, DropRateNfcr, MemOccNfcr, method, pi_t, QN, QNfcr, QNt, RN, RNfcr, RNt, runtime, solver, SS, SysTardN, t, TardN, timedOut, TN, TNfcr, TNt, UN, UNfcr, UNt, WeightNfcr, WN, WNfcr, XN, XNt

Constructor Summary

Constructors

Constructor	Description
MVAResult()

Method Summary

Methods inherited from class jline.solvers.SolverResult

deepCopy, reset

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Details

logNormConstAggr

public double logNormConstAggr

Logarithm of the aggregate normalizing constant

iter

public int iter

Number of iterations performed by the MVA algorithm (used by runAnalyzer method)

converged

public Boolean converged

Whether the fixed point met its tolerance, or null when the handler does not report one. This is NOT iter < iter_max: in Solver_amvald totiter aggregates the NESTED inner sweeps, so it routinely reaches the cap on a solve whose outer residual is exactly 0. Only the residual decides convergence there.

Left null by the single-loop handlers (the pfqn_linearizer family, pfqn_bs, pfqn_aql), for which the count IS a sound signal because it is not an aggregate: those exit on norm(Q-Qlast)<tol || iter > maxiter, so reaching maxiter does mean the tolerance was never met. Callers should prefer this flag and fall back to the count only when it is null.

hitProb

public Matrix hitProb

Cache hit probabilities [items x classes] (used by cache analyzers)

missProb

public Matrix missProb

Cache miss probabilities [items x classes] (used by cache analyzers)

cacheItemProb

public Map<Integer,Matrix> cacheItemProb

Per-cache per-item occupancy [nitems x (lists+1)], keyed by node index (used by the cache+queueing analyzer to populate getAvgItemTable)

Constructor Details

MVAResult

public MVAResult()