Examples with MultivariateElasticModel dr.evomodel.continuous.MultivariateElasticModel used on opensource projects

Example 6 with MultivariateElasticModel

use of dr.evomodel.continuous.MultivariateElasticModel in project beast-mcmc by beast-dev.

the class ContinuousDataLikelihoodDelegateTest method testLikelihoodDiagonalOURelaxedFactor.

public void testLikelihoodDiagonalOURelaxedFactor() {
    System.out.println("\nTest Likelihood using diagonal Relaxed OU and factor:");
    List<BranchRateModel> optimalTraitsModels = new ArrayList<BranchRateModel>();
    ArbitraryBranchRates.BranchRateTransform transform = make(false, false, false);
    optimalTraitsModels.add(new ArbitraryBranchRates(treeModel, new Parameter.Default("rate.1", new double[] { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9 }), transform, false));
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.2", new double[] { -1.5 })));
    DiagonalMatrix strengthOfSelectionMatrixParam = new DiagonalMatrix(new Parameter.Default(new double[] { 1.5, 20.0 }));
    DiffusionProcessDelegate diffusionProcessDelegate = new OUDiffusionModelDelegate(treeModel, diffusionModelFactor, optimalTraitsModels, new MultivariateElasticModel(strengthOfSelectionMatrixParam));
    // CDL
    ContinuousDataLikelihoodDelegate likelihoodDelegateFactors = new ContinuousDataLikelihoodDelegate(treeModel, diffusionProcessDelegate, dataModelFactor, rootPriorFactor, rateTransformation, rateModel, false);
    dataModelFactor.setLikelihoodDelegate(likelihoodDelegateFactors);
    // Likelihood Computation
    TreeDataLikelihood dataLikelihoodFactors = new TreeDataLikelihood(likelihoodDelegateFactors, treeModel, rateModel);
    testLikelihood("likelihoodDiagonalOURelaxedFactor", dataModelFactor, dataLikelihoodFactors);
    // Conditional simulations
    MathUtils.setSeed(17890826);
    double[] expectedTraits = new double[] { 1.2546097113922914, -1.1761389606670978, 1.305611773283861, -1.0644815941127401, 1.4571577864569687, -1.1477885449972944, 1.749551506462585, -0.9890375857170963, 1.0763987351136657, -1.0671848958534547, 1.5276137550128892, -0.9822950795368887 };
    testConditionalSimulations(dataLikelihoodFactors, likelihoodDelegateFactors, diffusionModelFactor, dataModelFactor, rootPriorFactor, expectedTraits);
}

Example 7 with MultivariateElasticModel

use of dr.evomodel.continuous.MultivariateElasticModel in project beast-mcmc by beast-dev.

the class DiffusionGradientTest method testGradientDiagonalOUWithMissing.

public void testGradientDiagonalOUWithMissing() {
    // Diffusion
    List<BranchRateModel> optimalTraitsModels = new ArrayList<BranchRateModel>();
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.1", new double[] { 1.0 })));
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.2", new double[] { 2.0 })));
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.3", new double[] { -2.0 })));
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.4", new double[] { 1.0 })));
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.5", new double[] { 2.0 })));
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.6", new double[] { -2.0 })));
    DiagonalMatrix strengthOfSelectionMatrixParam = new DiagonalMatrix(new Parameter.Default(new double[] { 0.0, 0.000005, 1.0, 5.0, 10.0, 50.0 }));
    // Wrt Precision
    DiffusionProcessDelegate diffusionProcessDelegate = new OUDiffusionModelDelegate(treeModel, diffusionModel, optimalTraitsModels, new MultivariateElasticModel(strengthOfSelectionMatrixParam));
    System.out.println("\nTest Diagonal OU gradient precision.");
    testGradient(diffusionModel, diffusionProcessDelegate, dataModel, precisionMatrix, strengthOfSelectionMatrixParam);
    System.out.println("\nTest Diagonal OU gradient precision with missing.");
    testGradient(diffusionModel, diffusionProcessDelegate, dataModelMissing, precisionMatrix, strengthOfSelectionMatrixParam);
    // Wrt Variance
    DiffusionProcessDelegate diffusionProcessDelegateVariance = new OUDiffusionModelDelegate(treeModel, diffusionModelVar, optimalTraitsModels, new MultivariateElasticModel(strengthOfSelectionMatrixParam));
    System.out.println("\nTest Diagonal OU gradient variance.");
    testGradient(diffusionModelVar, diffusionProcessDelegateVariance, dataModel, precisionMatrixInv, strengthOfSelectionMatrixParam);
    System.out.println("\nTest Diagonal OU gradient variance with missing.");
    testGradient(diffusionModelVar, diffusionProcessDelegateVariance, dataModelMissing, precisionMatrixInv, strengthOfSelectionMatrixParam);
    // Factor Model
    List<BranchRateModel> optimalTraitsModelsFactor = new ArrayList<BranchRateModel>();
    optimalTraitsModelsFactor.add(new StrictClockBranchRates(new Parameter.Default("rate.1", new double[] { 0.0 })));
    optimalTraitsModelsFactor.add(new StrictClockBranchRates(new Parameter.Default("rate.2", new double[] { 2.0 })));
    DiagonalMatrix strengthOfSelectionMatrixParamFactor = new DiagonalMatrix(new Parameter.Default(new double[] { 0.0, 50.0 }));
    DiffusionProcessDelegate diffusionProcessDelegateFactor = new OUDiffusionModelDelegate(treeModel, diffusionModelFactor, optimalTraitsModelsFactor, new MultivariateElasticModel(strengthOfSelectionMatrixParamFactor));
    System.out.println("\nTest gradient precision.");
    testGradient(diffusionModelFactor, diffusionProcessDelegateFactor, dataModelFactor, rootPriorFactor, rootMeanFactor, precisionMatrixFactor, false, strengthOfSelectionMatrixParamFactor, null, null);
    // Repeated Measures Model
    System.out.println("\nTest gradient precision repeated measures.");
    testGradient(diffusionModel, diffusionProcessDelegate, dataModelRepeatedMeasures, rootPrior, meanRoot, precisionMatrix, false, strengthOfSelectionMatrixParam, null, samplingPrecision);
    testGradient(diffusionModel, diffusionProcessDelegate, dataModelRepeatedMeasuresInv, rootPrior, meanRoot, precisionMatrix, false, strengthOfSelectionMatrixParam, null, samplingPrecisionInv);
}

Example 8 with MultivariateElasticModel

use of dr.evomodel.continuous.MultivariateElasticModel in project beast-mcmc by beast-dev.

the class DiffusionGradientTest method testGradientOUWithMissing.

public void testGradientOUWithMissing() {
    // Diffusion
    List<BranchRateModel> optimalTraitsModels = new ArrayList<BranchRateModel>();
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.1", new double[] { 1.0 })));
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.2", new double[] { 2.0 })));
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.3", new double[] { -2.0 })));
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.4", new double[] { 1.0 })));
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.5", new double[] { 2.0 })));
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.6", new double[] { -2.0 })));
    Parameter[] strengthOfSelectionParameters = new Parameter[6];
    strengthOfSelectionParameters[0] = new Parameter.Default(new double[] { 10, 1.0, 0.0, 0.0, 0.0, 2.0 });
    strengthOfSelectionParameters[1] = new Parameter.Default(new double[] { 1.0, 20, 0.0, 0.0, 0.0, 0.0 });
    strengthOfSelectionParameters[2] = new Parameter.Default(new double[] { 0.0, 0.0, 30, 0.0, 0.0, 0.0 });
    strengthOfSelectionParameters[3] = new Parameter.Default(new double[] { 0.0, 0.0, 0.0, 40, 3.0, 0.0 });
    strengthOfSelectionParameters[4] = new Parameter.Default(new double[] { 0.0, 0.0, 0.0, 3.0, 50, 0.0 });
    strengthOfSelectionParameters[5] = new Parameter.Default(new double[] { 2.0, 0.0, 0.0, 0.0, 0.0, 60 });
    MatrixParameter strengthOfSelectionMatrixParam = new MatrixParameter("strengthOfSelectionMatrix", strengthOfSelectionParameters);
    // Wrt Precision
    DiffusionProcessDelegate diffusionProcessDelegate = new OUDiffusionModelDelegate(treeModel, diffusionModel, optimalTraitsModels, new MultivariateElasticModel(strengthOfSelectionMatrixParam));
    System.out.println("\nTest OU gradient precision.");
    testGradient(diffusionModel, diffusionProcessDelegate, dataModel, precisionMatrix);
    System.out.println("\nTest OU gradient precision with missing.");
    testGradient(diffusionModel, diffusionProcessDelegate, dataModelMissing, precisionMatrix);
    // Wrt Variance
    DiffusionProcessDelegate diffusionProcessDelegateVariance = new OUDiffusionModelDelegate(treeModel, diffusionModelVar, optimalTraitsModels, new MultivariateElasticModel(strengthOfSelectionMatrixParam));
    System.out.println("\nTest OU gradient variance.");
    testGradient(diffusionModelVar, diffusionProcessDelegateVariance, dataModel, precisionMatrixInv);
    System.out.println("\nTest OU gradient variance with missing.");
    testGradient(diffusionModelVar, diffusionProcessDelegateVariance, dataModelMissing, precisionMatrixInv);
    // Factor Model
    List<BranchRateModel> optimalTraitsModelsFactor = new ArrayList<BranchRateModel>();
    optimalTraitsModelsFactor.add(new StrictClockBranchRates(new Parameter.Default("rate.1", new double[] { 0.0 })));
    optimalTraitsModelsFactor.add(new StrictClockBranchRates(new Parameter.Default("rate.2", new double[] { 2.0 })));
    Parameter[] strengthOfSelectionParametersFactor = new Parameter[2];
    strengthOfSelectionParametersFactor[0] = new Parameter.Default(new double[] { 10, 1.0 });
    strengthOfSelectionParametersFactor[1] = new Parameter.Default(new double[] { 1.0, 20 });
    MatrixParameter strengthOfSelectionMatrixParamFactor = new MatrixParameter("strengthOfSelectionMatrix", strengthOfSelectionParametersFactor);
    DiffusionProcessDelegate diffusionProcessDelegateFactor = new OUDiffusionModelDelegate(treeModel, diffusionModelFactor, optimalTraitsModelsFactor, new MultivariateElasticModel(strengthOfSelectionMatrixParamFactor));
    System.out.println("\nTest gradient precision.");
    testGradient(diffusionModelFactor, diffusionProcessDelegateFactor, dataModelFactor, rootPriorFactor, rootMeanFactor, precisionMatrixFactor, false);
    // Repeated Measures Model
    System.out.println("\nTest gradient precision repeated measures.");
    testGradient(diffusionModel, diffusionProcessDelegate, dataModelRepeatedMeasures, rootPrior, meanRoot, precisionMatrix, false, null, null, samplingPrecision);
    testGradient(diffusionModel, diffusionProcessDelegate, dataModelRepeatedMeasuresInv, rootPrior, meanRoot, precisionMatrix, false, null, null, samplingPrecisionInv);
    // ************//
    // Single opt
    List<BranchRateModel> optimalTraitsModelsSingle = new ArrayList<BranchRateModel>();
    CompoundParameter optParamSingle = new CompoundParameter("opt");
    for (int i = 0; i < dimTrait; i++) {
        Parameter rate = new Parameter.Default("opt." + (i + 1), new double[] { 2.2 * i + 3.1 });
        optParamSingle.addParameter(rate);
        optimalTraitsModelsSingle.add(new StrictClockBranchRates(rate));
    }
    // Wrt Precision
    DiffusionProcessDelegate diffusionProcessDelegateSingle = new OUDiffusionModelDelegate(treeModel, diffusionModel, optimalTraitsModelsSingle, new MultivariateElasticModel(strengthOfSelectionMatrixParam));
    System.out.println("\nTest OU single opt gradient precision.");
    testGradient(diffusionModel, diffusionProcessDelegateSingle, dataModel, precisionMatrix, optParamSingle);
    System.out.println("\nTest OU single opt gradient precision with missing.");
    testGradient(diffusionModel, diffusionProcessDelegateSingle, dataModelMissing, precisionMatrix, optParamSingle);
    // Wrt Variance
    DiffusionProcessDelegate diffusionProcessDelegateVarianceSingle = new OUDiffusionModelDelegate(treeModel, diffusionModelVar, optimalTraitsModelsSingle, new MultivariateElasticModel(strengthOfSelectionMatrixParam));
    System.out.println("\nTest OU single opt gradient variance.");
    testGradient(diffusionModelVar, diffusionProcessDelegateVarianceSingle, dataModel, precisionMatrixInv, optParamSingle);
    System.out.println("\nTest OU single opt gradient variance with missing.");
    testGradient(diffusionModelVar, diffusionProcessDelegateVarianceSingle, dataModelMissing, precisionMatrixInv, optParamSingle);
    // Repeated Measures Model
    System.out.println("\nTest gradient precision repeated measures.");
    testGradient(diffusionModel, diffusionProcessDelegateSingle, dataModelRepeatedMeasures, rootPrior, meanRoot, precisionMatrix, false, null, optParamSingle, samplingPrecision);
    testGradient(diffusionModel, diffusionProcessDelegateSingle, dataModelRepeatedMeasuresInv, rootPrior, meanRoot, precisionMatrix, false, null, optParamSingle, samplingPrecisionInv);
    // **************//
    // Same mean
    // Diffusion
    List<BranchRateModel> optimalTraitsModelsSame = new ArrayList<BranchRateModel>();
    for (int i = 0; i < dimTrait; i++) {
        optimalTraitsModelsSame.add(new StrictClockBranchRates(meanRoot.getParameter(i)));
    }
    // Wrt Precision
    DiffusionProcessDelegate diffusionProcessDelegateSame = new OUDiffusionModelDelegate(treeModel, diffusionModel, optimalTraitsModelsSame, new MultivariateElasticModel(strengthOfSelectionMatrixParam));
    System.out.println("\nTest OU Same opt gradient precision.");
    testGradient(diffusionModel, diffusionProcessDelegateSame, dataModel, precisionMatrix, meanRoot);
    System.out.println("\nTest OU Same opt gradient precision with missing.");
    testGradient(diffusionModel, diffusionProcessDelegateSame, dataModelMissing, precisionMatrix, meanRoot);
    // Wrt Variance
    DiffusionProcessDelegate diffusionProcessDelegateVarianceSame = new OUDiffusionModelDelegate(treeModel, diffusionModelVar, optimalTraitsModelsSame, new MultivariateElasticModel(strengthOfSelectionMatrixParam));
    System.out.println("\nTest OU Same opt gradient variance.");
    testGradient(diffusionModelVar, diffusionProcessDelegateVarianceSame, dataModel, precisionMatrixInv, meanRoot);
    System.out.println("\nTest OU Same opt gradient variance with missing.");
    testGradient(diffusionModelVar, diffusionProcessDelegateVarianceSame, dataModelMissing, precisionMatrixInv, meanRoot);
    // Repeated Measures Model
    System.out.println("\nTest gradient precision repeated measures.");
    testGradient(diffusionModel, diffusionProcessDelegateSame, dataModelRepeatedMeasures, rootPrior, meanRoot, precisionMatrix, false, null, meanRoot, samplingPrecision);
    testGradient(diffusionModel, diffusionProcessDelegateSame, dataModelRepeatedMeasuresInv, rootPrior, meanRoot, precisionMatrix, false, null, meanRoot, samplingPrecisionInv);
}

Example 9 with MultivariateElasticModel

use of dr.evomodel.continuous.MultivariateElasticModel in project beast-mcmc by beast-dev.

the class ContinuousDataLikelihoodDelegateTest method testLikelihoodFullOUFactor.

public void testLikelihoodFullOUFactor() {
    System.out.println("\nTest Likelihood using full OU and factor:");
    // Diffusion
    List<BranchRateModel> optimalTraitsModels = new ArrayList<BranchRateModel>();
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.1", new double[] { 1.0 })));
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.1", new double[] { 2.0 })));
    Parameter[] strengthOfSelectionParameters = new Parameter[2];
    strengthOfSelectionParameters[0] = new Parameter.Default(new double[] { 0.5, 0.05 });
    strengthOfSelectionParameters[1] = new Parameter.Default(new double[] { 0.05, 25.5 });
    MatrixParameter strengthOfSelectionMatrixParam = new MatrixParameter("strengthOfSelectionMatrix", strengthOfSelectionParameters);
    DiffusionProcessDelegate diffusionProcessDelegate = new OUDiffusionModelDelegate(treeModel, diffusionModelFactor, optimalTraitsModels, new MultivariateElasticModel(strengthOfSelectionMatrixParam));
    // CDL
    ContinuousDataLikelihoodDelegate likelihoodDelegateFactors = new ContinuousDataLikelihoodDelegate(treeModel, diffusionProcessDelegate, dataModelFactor, rootPriorFactor, rateTransformation, rateModel, false);
    dataModelFactor.setLikelihoodDelegate(likelihoodDelegateFactors);
    // Likelihood Computation
    TreeDataLikelihood dataLikelihoodFactors = new TreeDataLikelihood(likelihoodDelegateFactors, treeModel, rateModel);
    testLikelihood("likelihoodFullOUFactor", dataModelFactor, dataLikelihoodFactors);
    // Conditional simulations
    MathUtils.setSeed(17890826);
    double[] expectedTraits = new double[] { 0.4889263054598222, 1.866143125522109, 0.41845209107775877, 1.978457443711536, 0.5589398189015322, 1.8942177991552116, 0.9699471556784252, 2.0423474270630155, 0.3288819110219145, 1.9759942582707206, 0.8081782260054755, 2.038299849681893 };
    testConditionalSimulations(dataLikelihoodFactors, likelihoodDelegateFactors, diffusionModelFactor, dataModelFactor, rootPriorFactor, expectedTraits);
}

Example 10 with MultivariateElasticModel

use of dr.evomodel.continuous.MultivariateElasticModel in project beast-mcmc by beast-dev.

the class ContinuousDataLikelihoodDelegateTest method testLikelihoodDiagonalOUBMInd.

public void testLikelihoodDiagonalOUBMInd() {
    System.out.println("\nTest Likelihood using Diagonal OU / BM:");
    // Diffusion
    List<BranchRateModel> optimalTraitsModels = new ArrayList<BranchRateModel>();
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.1", new double[] { 1.0 })));
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.2", new double[] { -3.0 })));
    optimalTraitsModels.add(new StrictClockBranchRates(new Parameter.Default("rate.3", new double[] { -2.0 })));
    DiagonalMatrix strengthOfSelectionMatrixParamOUBM = new DiagonalMatrix(new Parameter.Default(new double[] { 0.0, 0.0, 50.0 }));
    DiagonalMatrix strengthOfSelectionMatrixParamOU = new DiagonalMatrix(new Parameter.Default(new double[] { 10.0, 20.0, 50.0 }));
    DiagonalMatrix diffusionPrecisionMatrixParameter = new DiagonalMatrix(new Parameter.Default(new double[] { 1.0, 2.0, 3.0 }));
    MultivariateDiffusionModel diffusionModel = new MultivariateDiffusionModel(diffusionPrecisionMatrixParameter);
    DiffusionProcessDelegate diffusionProcessDelegateOUBM = new OUDiffusionModelDelegate(treeModel, diffusionModel, optimalTraitsModels, new MultivariateElasticModel(strengthOfSelectionMatrixParamOUBM));
    DiffusionProcessDelegate diffusionProcessDelegateOU = new OUDiffusionModelDelegate(treeModel, diffusionModel, optimalTraitsModels, new MultivariateElasticModel(strengthOfSelectionMatrixParamOU));
    DiffusionProcessDelegate diffusionProcessDelegateBM = new HomogeneousDiffusionModelDelegate(treeModel, diffusionModel);
    // CDL
    ContinuousDataLikelihoodDelegate likelihoodDelegateOUBM = new ContinuousDataLikelihoodDelegate(treeModel, diffusionProcessDelegateOUBM, dataModel, rootPriorInf, rateTransformation, rateModel, false);
    ContinuousDataLikelihoodDelegate likelihoodDelegateOU = new ContinuousDataLikelihoodDelegate(treeModel, diffusionProcessDelegateOU, dataModel, rootPriorInf, rateTransformation, rateModel, false);
    ContinuousDataLikelihoodDelegate likelihoodDelegateBM = new ContinuousDataLikelihoodDelegate(treeModel, diffusionProcessDelegateBM, dataModel, rootPriorInf, rateTransformation, rateModel, false);
    // Likelihood Computation
    TreeDataLikelihood dataLikelihoodOUBM = new TreeDataLikelihood(likelihoodDelegateOUBM, treeModel, rateModel);
    TreeDataLikelihood dataLikelihoodOU = new TreeDataLikelihood(likelihoodDelegateOU, treeModel, rateModel);
    TreeDataLikelihood dataLikelihoodBM = new TreeDataLikelihood(likelihoodDelegateBM, treeModel, rateModel);
    // Conditional simulations
    MathUtils.setSeed(17890826);
    double[] traitsOUBM = getConditionalSimulations(dataLikelihoodOUBM, likelihoodDelegateOUBM, diffusionModel, dataModel, rootPriorInf, treeModel, rateTransformation);
    System.err.println(new Vector(traitsOUBM));
    MathUtils.setSeed(17890826);
    double[] traitsOU = getConditionalSimulations(dataLikelihoodOU, likelihoodDelegateOU, diffusionModel, dataModel, rootPriorInf, treeModel, rateTransformation);
    System.err.println(new Vector(traitsOU));
    MathUtils.setSeed(17890826);
    double[] traitsBM = getConditionalSimulations(dataLikelihoodBM, likelihoodDelegateBM, diffusionModel, dataModel, rootPriorInf, treeModel, rateTransformation);
    System.err.println(new Vector(traitsBM));
    // Check that missing dimensions with the same process have the same values
    assertEquals(format.format(traitsBM[3]), format.format(traitsOUBM[3]));
    assertEquals(format.format(traitsBM[4]), format.format(traitsOUBM[4]));
    assertEquals(format.format(traitsBM[7]), format.format(traitsOUBM[7]));
    assertEquals(format.format(traitsOU[5]), format.format(traitsOUBM[5]));
}