Package jline.api.lqn
Class Lqn_export_nlp.Program
java.lang.Object
jline.api.lqn.Lqn_export_nlp.Program
- Enclosing class:
- Lqn_export_nlp
The program, built in memory. Field and method names follow the python
script's; every array is in the program's own 0-based positions.
-
Field Summary
FieldsModifier and TypeFieldDescriptionfinal double[]final int[]final int[]final double[]final String[]final int[][]final int[][]final double[][]final intfinal double[][]final double[]final String[]Variable blocks in layout order, with their sizes.final int[]final int[][]final double[]final int[]final int[]final int[]final double[]final String[]final int[][]final int[][]final int[][]final String[]final int[][]final int[]final int[]final intfinal boolean[]final int[][][]final int[]final String[]final int[]GH_OF[h], the position of host h among QD_HOSTS, -1 when it carries none.final int[]final boolean[]final double[]final boolean[]final String[]final double[]final double[]final double[]final Lqn_export_nlp.Rate[]final boolean[]final String[]final intPosition in the python layout of the model's element shifts, for the analyzer.final intfinal intfinal intfinal intfinal intfinal intfinal intfinal intfinal intfinal intfinal intfinal int[]final int[]final int[]final int[]final double[]final double[][]final int[][]final int[][]final int[][]final doublefinal boolean[]final double[]final int[]final boolean[]final boolean[]final String[]final double[]final double[]final double[]final Lqn_export_nlp.Rate[]final double[]final String[]final int -
Method Summary
Modifier and TypeMethodDescriptiondouble[]Variable block NAME of the iterate X.intSize of variable block NAME.intAbsolute column of variable BLOCK at index I.double[]initialPoint(int warm) A point satisfying every linear relation, after WARM fixed-point sweeps of the QD-AMVA laws, python'sinitial_point.double[][]jacobian(double[] x) lmFallback(double[] x0, int maxiter) Projected Levenberg-Marquardt onmerit(double[])over x >= 0, the fallback the MATLAB, JAR and C++ ports share in place of python's trust-constr.doublemerit(double[] x) The merit the fallback descends: residuals, equality and inequality violations squared.doubleobjective(double[] x) double[]polish(double[] x0) double[]polish(double[] x0, int iters) Newton on the square system [residuals; AEQ x - BEQ] = 0, each step the minimum-norm least-squares solution, python'spolish.doubleqbarHost(double[] Q, int h, int a) List<double[]>qbarHostTerms(int h, int a, double g) g times the arrival-instant estimate at host H for activity A, as (column, coefficient) pairs.doubleqbarTask(double[] Qt, int t, int c) List<double[]>qbarTaskTerms(int t, int c, double g) doubleqdPopHost(double[] Q, int h) doubleqdPopTask(double[] Qt, int t) double[]residuals(double[] x) intsent(int e) Column of the entry's thread-seconds: SA when the model forks, else S.solve(int warm, int maxiter) Newton from an escalating warm start (WARM, then 5, 20 and 100 sweeps), python'ssolve; when no rung reaches a root, the projected Levenberg-Marquardt fallback from the best rung's start, then a polish.booleansolved(double[] x) A root of the square system, feasible to the bounds and the inequalities.varNames()
-
Field Details
-
HOSTS
-
HOST_MULT
public final double[] HOST_MULT -
HOST_MAXMULT
public final double[] HOST_MAXMULT -
HOST_INF
public final boolean[] HOST_INF -
TASKS
-
TASK_HOST
public final int[] TASK_HOST -
TASK_MULT
public final double[] TASK_MULT -
TASK_INF
public final boolean[] TASK_INF -
TASK_ISREF
public final boolean[] TASK_ISREF -
TASK_THINK
public final double[] TASK_THINK -
ENTRIES
-
ENTRY_TASK
public final int[] ENTRY_TASK -
ENTRY_ORDER
public final int[] ENTRY_ORDER -
ACTS
-
ACT_TASK
public final int[] ACT_TASK -
ACT_ENTRY
public final int[] ACT_ENTRY -
ACT_DEMAND
public final double[] ACT_DEMAND -
ACT_VISITS
public final double[] ACT_VISITS -
CALLS
-
CALL_ACT
public final int[] CALL_ACT -
CALL_ENTRY
public final int[] CALL_ENTRY -
CALL_MEAN
public final double[] CALL_MEAN -
CALL_CALLER_TASK
public final int[] CALL_CALLER_TASK -
FORK_ENTRY
public final int[] FORK_ENTRY -
FORK_BRANCHES
public final int[][][] FORK_BRANCHES -
FORK_LABEL
-
SOFTMIN_ALPHA
public final double SOFTMIN_ALPHA -
HOST_LLD
-
TASK_LLD
-
HOST_RATE_CAPPED
public final boolean[] HOST_RATE_CAPPED -
HOST_RATE
-
TASK_RATE
-
NH
public final int NH -
NT
public final int NT -
NE
public final int NE -
NA
public final int NA -
NC
public final int NC -
NFORK
public final int NFORK -
NR
public final int NR -
NGH
public final int NGH -
NGT
public final int NGT -
NVAR
public final int NVAR -
NRES
public final int NRES -
ACTS_OF_HOST
public final int[][] ACTS_OF_HOST -
ACTS_OF_ENTRY
public final int[][] ACTS_OF_ENTRY -
CALLS_OF_ACT
public final int[][] CALLS_OF_ACT -
CALLS_INTO_ENTRY
public final int[][] CALLS_INTO_ENTRY -
CALLS_INTO_TASK
public final int[][] CALLS_INTO_TASK -
ENTRIES_OF_TASK
public final int[][] ENTRIES_OF_TASK -
FORK_ACTS
public final boolean[] FORK_ACTS -
BLOCKS_OF_ENTRY
public final int[][] BLOCKS_OF_ENTRY -
SEQ_ACTS_OF_ENTRY
public final int[][] SEQ_ACTS_OF_ENTRY -
REF_TASKS
public final int[] REF_TASKS -
REF_ENTRY
public final int[] REF_ENTRY -
TASK_DELAY
public final boolean[] TASK_DELAY -
HOST_DELAY
public final boolean[] HOST_DELAY -
TASK_POP
public final double[] TASK_POP -
SIBLING_ACTS
public final int[][] SIBLING_ACTS -
SIBLING_CALLS
public final int[][] SIBLING_CALLS -
QD_HOSTS
public final int[] QD_HOSTS -
QD_TASKS
public final int[] QD_TASKS -
GH_OF
public final int[] GH_OFGH_OF[h], the position of host h among QD_HOSTS, -1 when it carries none. -
GT_OF
public final int[] GT_OF -
HOST_NTOT
public final double[] HOST_NTOT -
TASK_NTOT
public final double[] TASK_NTOT -
HOST_DELTA
public final double[] HOST_DELTA -
TASK_DELTA
public final double[] TASK_DELTA -
BLOCK_NAMES
Variable blocks in layout order, with their sizes. -
BLOCK_SIZES
public final int[] BLOCK_SIZES -
OFFSET
-
AEQ
public final double[][] AEQ -
BEQ
public final double[] BEQ -
AUB
public final double[][] AUB -
BUB
public final double[] BUB -
RES_LIN
public final double[][] RES_LIN -
RES_CONST
public final double[] RES_CONST -
hshift
public final int hshiftPosition in the python layout of the model's element shifts, for the analyzer. -
tshift
public final int tshift -
eshift
public final int eshift -
ashift
public final int ashift
-
-
Method Details
-
col
Absolute column of variable BLOCK at index I. -
sent
public int sent(int e) Column of the entry's thread-seconds: SA when the model forks, else S. -
blockSize
Size of variable block NAME. -
block
Variable block NAME of the iterate X. -
varNames
-
qbarHostTerms
g times the arrival-instant estimate at host H for activity A, as (column, coefficient) pairs. -
qbarTaskTerms
-
residuals
public double[] residuals(double[] x) -
jacobian
public double[][] jacobian(double[] x) -
objective
public double objective(double[] x) -
qdPopHost
public double qdPopHost(double[] Q, int h) -
qdPopTask
public double qdPopTask(double[] Qt, int t) -
qbarHost
public double qbarHost(double[] Q, int h, int a) -
qbarTask
public double qbarTask(double[] Qt, int t, int c) -
initialPoint
public double[] initialPoint(int warm) A point satisfying every linear relation, after WARM fixed-point sweeps of the QD-AMVA laws, python'sinitial_point. -
polish
public double[] polish(double[] x0, int iters) Newton on the square system [residuals; AEQ x - BEQ] = 0, each step the minimum-norm least-squares solution, python'spolish. -
polish
public double[] polish(double[] x0) -
solved
public boolean solved(double[] x) A root of the square system, feasible to the bounds and the inequalities. -
solve
Newton from an escalating warm start (WARM, then 5, 20 and 100 sweeps), python'ssolve; when no rung reaches a root, the projected Levenberg-Marquardt fallback from the best rung's start, then a polish. -
merit
public double merit(double[] x) The merit the fallback descends: residuals, equality and inequality violations squared. -
lmFallback
Projected Levenberg-Marquardt onmerit(double[])over x >= 0, the fallback the MATLAB, JAR and C++ ports share in place of python's trust-constr.
-