Package com.polytechnik.kgo
Class KGOSolutionVectorXVectorF
java.lang.Object
com.polytechnik.kgo.KGOSolutionVectorXVectorF
- All Implemented Interfaces:
VectorXVectorFPointEvaluatable
Knowledge Generalizing Operator solution to a vector X to vector F problem.
The X- and F- spaces are assumed to be regularized: XX and FF are not degenerated and diagonal.
-
Field Summary
FieldsModifier and TypeFieldDescription(package private) final KGOOptimizationResult.Approximation
private final double[]
Contributing subspace.(package private) static final EVSolver
Standard EV solver.private final VectorXVectorFPointEvaluatableCommon
final double
A coverage corresponding to found mapping.private final double[]
final double
A coverage corresponding to "Direct Projection" type of expansion: least squares value is calculated first, then a f-localized function is built at this value.final double
Maximal possible coverage among all projective-type transforms.private final double[]
Old "projective" version.final int
final int
(package private) final RegularizedVectorXVectorF
private final double[]
(package private) final SKFromSampleCalculationMethod
(package private) static final RegularizationMethod
private final AdjustedStateToUnitaryWithSVD
For unit tests.private final double[]
-
Constructor Summary
ConstructorsConstructorDescriptionKGOSolutionVectorXVectorF
(DataReadObservationVectorXVectorF ro, RegularizationMethod regularization_methodX, RegularizationMethod regularization_methodF, KGOOptimizationResult.Approximation approximation, SKFromSampleCalculationMethod SKtype, double eps) -
Method Summary
Modifier and TypeMethodDescription(package private) static double
christoffel
(RegularizedVectorX reg, double[] xorig) static KGOSolutionVectorXVectorF
createFromSample
(DataReadObservationVectorXVectorF ro, KGOOptimizationResult.Approximation approximation, SKFromSampleCalculationMethod SKtype) evaluateAt
(double[] xorig) An evaluation of Vector X to Vector F model.(package private) double[]
forigFromF
(double[] F) (package private) DirectProjectionSolutionVectorXVectorF
Returns DirectProjection (old least squares version) for unit test.(package private) double
getDirectProjectionProbability
(VectorXVectorFPointAccess e, double[] forig) (package private) double[]
getFofMaximalProbability_CalculateInSVDBasis
(double[] xorig) For unit test to matchVectorXVectorFPointAccess.getFProbabilityAtX(double[])
.(package private) final ToDoubleFunction
<double[]> getFProbabilityAtX_CalculateInSVDBasis
(double[] xorig) For unit test to matchVectorXVectorFPointAccess.getFProbabilityAtX(double[])
.(package private) double
getFtotalForTransform
(DataReadObservationVectorXVectorF ro, double[] u_jk) private double[]
Optimized version for speed, without re-calculation and extra function calls.private double[]
getXF_KFF_FX
(double[] KFF) (package private) static double[]
leastSquaresO
(double[] FX, RegularizedVectorX reg, double[] xorig) static void
Trivial test run once.private void
matchMatrices
(DataReadObservationVectorXVectorF ro, double eps) private void
matchSampleValues
(DataReadObservationVectorXVectorF ro, double eps) private void
selfTest
(DataReadObservationVectorXVectorF ro, double eps) private void
Contributing subspace old projective approach.Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
Methods inherited from interface com.polytechnik.utils.VectorXVectorFPointEvaluatable
evaluateAtSequence
-
Field Details
-
EV
Standard EV solver. -
reg
-
nX
public final int nX -
nF
public final int nF -
fF
private final double[] fF -
xX
private final double[] xX -
S_FXFX
private final double[] S_FXFX -
approximation
-
SKtype
-
CX
private final double[] CXContributing subspace. -
FX
private final double[] FXOld "projective" version. -
FTOT
public final double FTOTMaximal possible coverage among all projective-type transforms. -
FLS
public final double FLSA coverage corresponding to "Direct Projection" type of expansion: least squares value is calculated first, then a f-localized function is built at this value. The value may be greater thanFTOT
. -
FCOVU
public final double FCOVUA coverage corresponding to found mapping. -
evaluator
-
test_svd
For unit tests. -
STD_REGULARIZATION_METHOD
-
-
Constructor Details
-
KGOSolutionVectorXVectorF
KGOSolutionVectorXVectorF(DataReadObservationVectorXVectorF ro, RegularizationMethod regularization_methodX, RegularizationMethod regularization_methodF, KGOOptimizationResult.Approximation approximation, SKFromSampleCalculationMethod SKtype, double eps) - Parameters:
ro
- Read Observations (sample).regularization_methodX
- regularizatuion method ofx
.regularization_methodF
- regularizatuion method off
.eps
- Run a selftest with thiseps
if positive.
-
-
Method Details
-
forigFromF
double[] forigFromF(double[] F) -
getFtotalForTransform
-
evaluateAt
public VectorXVectorFPointEvaluatableCommon.VectorXVectorFPointAccessWithDimensionAdjusted evaluateAt(double[] xorig) Description copied from interface:VectorXVectorFPointEvaluatable
An evaluation of Vector X to Vector F model.- Specified by:
evaluateAt
in interfaceVectorXVectorFPointEvaluatable
-
getXF_KFF_FX
private double[] getXF_KFF_FX(double[] KFF) -
getFXFX_F_Christoffel_SPEED_Optimization
Optimized version for speed, without re-calculation and extra function calls. -
testProjectiveContributingBasisAndFTOTCoverage
private void testProjectiveContributingBasisAndFTOTCoverage(DataReadObservationVectorXVectorF ro, double eps) Contributing subspace old projective approach. -
matchMatrices
-
matchSampleValues
-
selfTest
-
getDirectProjection_FXbasis
DirectProjectionSolutionVectorXVectorF getDirectProjection_FXbasis()Returns DirectProjection (old least squares version) for unit test. -
getDirectProjectionProbability
-
getFProbabilityAtX_CalculateInSVDBasis
For unit test to matchVectorXVectorFPointAccess.getFProbabilityAtX(double[])
. -
getFofMaximalProbability_CalculateInSVDBasis
double[] getFofMaximalProbability_CalculateInSVDBasis(double[] xorig) For unit test to matchVectorXVectorFPointAccess.getFProbabilityAtX(double[])
. -
leastSquaresO
-
christoffel
-
createFromSample
public static KGOSolutionVectorXVectorF createFromSample(DataReadObservationVectorXVectorF ro, KGOOptimizationResult.Approximation approximation, SKFromSampleCalculationMethod SKtype) -
main
Trivial test run once.
-