Search in sources :

Example 1 with SingleFreeCircularErfGaussian2DFunction

use of gdsc.smlm.function.gaussian.erf.SingleFreeCircularErfGaussian2DFunction in project GDSC-SMLM by aherbert.

the class LSQLVMGradientProcedureTest method gradientProcedureComputesSameOutputWithBias.

@Test
public void gradientProcedureComputesSameOutputWithBias() {
    ErfGaussian2DFunction func = new SingleFreeCircularErfGaussian2DFunction(blockWidth, blockWidth);
    int nparams = func.getNumberOfGradients();
    int iter = 100;
    rdg = new RandomDataGenerator(new Well19937c(30051977));
    ArrayList<double[]> paramsList = new ArrayList<double[]>(iter);
    ArrayList<double[]> yList = new ArrayList<double[]>(iter);
    ArrayList<double[]> alphaList = new ArrayList<double[]>(iter);
    ArrayList<double[]> betaList = new ArrayList<double[]>(iter);
    ArrayList<double[]> xList = new ArrayList<double[]>(iter);
    // Manipulate the background
    double defaultBackground = Background;
    try {
        Background = 1e-2;
        createData(1, iter, paramsList, yList, true);
        EJMLLinearSolver solver = new EJMLLinearSolver(1e-5, 1e-6);
        for (int i = 0; i < paramsList.size(); i++) {
            double[] y = yList.get(i);
            double[] a = paramsList.get(i);
            BaseLSQLVMGradientProcedure p = LSQLVMGradientProcedureFactory.create(y, func);
            p.gradient(a);
            double[] beta = p.beta;
            alphaList.add(p.getAlphaLinear());
            betaList.add(beta.clone());
            for (int j = 0; j < nparams; j++) {
                if (Math.abs(beta[j]) < 1e-6)
                    System.out.printf("[%d] Tiny beta %s %g\n", i, func.getName(j), beta[j]);
            }
            // Solve
            if (!solver.solve(p.getAlphaMatrix(), beta))
                throw new AssertionError();
            xList.add(beta);
        //System.out.println(Arrays.toString(beta));
        }
        //for (int b = 1; b < 1000; b *= 2)
        for (double b : new double[] { -500, -100, -10, -1, -0.1, 0, 0.1, 1, 10, 100, 500 }) {
            Statistics[] rel = new Statistics[nparams];
            Statistics[] abs = new Statistics[nparams];
            for (int i = 0; i < nparams; i++) {
                rel[i] = new Statistics();
                abs[i] = new Statistics();
            }
            for (int i = 0; i < paramsList.size(); i++) {
                double[] y = add(yList.get(i), b);
                double[] a = paramsList.get(i).clone();
                a[0] += b;
                BaseLSQLVMGradientProcedure p = LSQLVMGradientProcedureFactory.create(y, func);
                p.gradient(a);
                double[] beta = p.beta;
                double[] alpha2 = alphaList.get(i);
                double[] beta2 = betaList.get(i);
                double[] x2 = xList.get(i);
                Assert.assertArrayEquals("Beta", beta2, beta, 1e-10);
                Assert.assertArrayEquals("Alpha", alpha2, p.getAlphaLinear(), 1e-10);
                // Solve
                solver.solve(p.getAlphaMatrix(), beta);
                Assert.assertArrayEquals("X", x2, beta, 1e-10);
                for (int j = 0; j < nparams; j++) {
                    rel[j].add(DoubleEquality.relativeError(x2[j], beta[j]));
                    abs[j].add(Math.abs(x2[j] - beta[j]));
                }
            }
            for (int i = 0; i < nparams; i++) System.out.printf("Bias = %.2f : %s : Rel %g +/- %g: Abs %g +/- %g\n", b, func.getName(i), rel[i].getMean(), rel[i].getStandardDeviation(), abs[i].getMean(), abs[i].getStandardDeviation());
        }
    } finally {
        Background = defaultBackground;
    }
}
Also used : ErfGaussian2DFunction(gdsc.smlm.function.gaussian.erf.ErfGaussian2DFunction) SingleFreeCircularErfGaussian2DFunction(gdsc.smlm.function.gaussian.erf.SingleFreeCircularErfGaussian2DFunction) RandomDataGenerator(org.apache.commons.math3.random.RandomDataGenerator) EJMLLinearSolver(gdsc.smlm.fitting.linear.EJMLLinearSolver) ArrayList(java.util.ArrayList) Well19937c(org.apache.commons.math3.random.Well19937c) Statistics(gdsc.core.utils.Statistics) SingleFreeCircularErfGaussian2DFunction(gdsc.smlm.function.gaussian.erf.SingleFreeCircularErfGaussian2DFunction) Test(org.junit.Test)

Example 2 with SingleFreeCircularErfGaussian2DFunction

use of gdsc.smlm.function.gaussian.erf.SingleFreeCircularErfGaussian2DFunction in project GDSC-SMLM by aherbert.

the class FastMLEGradient2ProcedureTest method gradientCalculatorComputesGradient.

@Test
public void gradientCalculatorComputesGradient() {
    gradientCalculatorComputesGradient(new SingleFreeCircularErfGaussian2DFunction(blockWidth, blockWidth));
    // Use a reasonable z-depth function from the Smith, et al (2010) paper (page 377)
    double gamma = 0.389;
    double d = 0.531;
    double Ax = -0.0708;
    double Bx = -0.073;
    double Ay = 0.164;
    double By = 0.0417;
    HoltzerAstimatismZModel zModel = HoltzerAstimatismZModel.create(gamma, d, Ax, Bx, Ay, By);
    gradientCalculatorComputesGradient(new SingleAstigmatismErfGaussian2DFunction(blockWidth, blockWidth, zModel));
}
Also used : SingleAstigmatismErfGaussian2DFunction(gdsc.smlm.function.gaussian.erf.SingleAstigmatismErfGaussian2DFunction) SingleFreeCircularErfGaussian2DFunction(gdsc.smlm.function.gaussian.erf.SingleFreeCircularErfGaussian2DFunction) HoltzerAstimatismZModel(gdsc.smlm.function.gaussian.HoltzerAstimatismZModel) Test(org.junit.Test)

Example 3 with SingleFreeCircularErfGaussian2DFunction

use of gdsc.smlm.function.gaussian.erf.SingleFreeCircularErfGaussian2DFunction in project GDSC-SMLM by aherbert.

the class FastMLEJacobianGradient2ProcedureTest method gradientCalculatorComputesGradient.

@Test
public void gradientCalculatorComputesGradient() {
    gradientCalculatorComputesGradient(1, new SingleFreeCircularErfGaussian2DFunction(blockWidth, blockWidth));
    gradientCalculatorComputesGradient(2, new MultiFreeCircularErfGaussian2DFunction(2, blockWidth, blockWidth));
    // Use a reasonable z-depth function from the Smith, et al (2010) paper (page 377)
    double gamma = 0.389;
    double d = 0.531;
    double Ax = -0.0708;
    double Bx = -0.073;
    double Ay = 0.164;
    double By = 0.0417;
    HoltzerAstimatismZModel zModel = HoltzerAstimatismZModel.create(gamma, d, Ax, Bx, Ay, By);
    gradientCalculatorComputesGradient(1, new SingleAstigmatismErfGaussian2DFunction(blockWidth, blockWidth, zModel));
}
Also used : SingleAstigmatismErfGaussian2DFunction(gdsc.smlm.function.gaussian.erf.SingleAstigmatismErfGaussian2DFunction) SingleFreeCircularErfGaussian2DFunction(gdsc.smlm.function.gaussian.erf.SingleFreeCircularErfGaussian2DFunction) MultiFreeCircularErfGaussian2DFunction(gdsc.smlm.function.gaussian.erf.MultiFreeCircularErfGaussian2DFunction) HoltzerAstimatismZModel(gdsc.smlm.function.gaussian.HoltzerAstimatismZModel) Test(org.junit.Test)

Aggregations

SingleFreeCircularErfGaussian2DFunction (gdsc.smlm.function.gaussian.erf.SingleFreeCircularErfGaussian2DFunction)3 Test (org.junit.Test)3 HoltzerAstimatismZModel (gdsc.smlm.function.gaussian.HoltzerAstimatismZModel)2 SingleAstigmatismErfGaussian2DFunction (gdsc.smlm.function.gaussian.erf.SingleAstigmatismErfGaussian2DFunction)2 Statistics (gdsc.core.utils.Statistics)1 EJMLLinearSolver (gdsc.smlm.fitting.linear.EJMLLinearSolver)1 ErfGaussian2DFunction (gdsc.smlm.function.gaussian.erf.ErfGaussian2DFunction)1 MultiFreeCircularErfGaussian2DFunction (gdsc.smlm.function.gaussian.erf.MultiFreeCircularErfGaussian2DFunction)1 ArrayList (java.util.ArrayList)1 RandomDataGenerator (org.apache.commons.math3.random.RandomDataGenerator)1 Well19937c (org.apache.commons.math3.random.Well19937c)1