Package org.apache.commons.math.transform

Source Code of org.apache.commons.math.transform.FastFourierTransformer$MultiDimensionalComplexMatrix

/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements.  See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License.  You may obtain a copy of the License at
*
*      http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.commons.math.transform;

import java.io.Serializable;
import java.lang.reflect.Array;

import org.apache.commons.math.FunctionEvaluationException;
import org.apache.commons.math.MathRuntimeException;
import org.apache.commons.math.analysis.UnivariateRealFunction;
import org.apache.commons.math.complex.Complex;

/**
* Implements the <a href="http://mathworld.wolfram.com/FastFourierTransform.html">
* Fast Fourier Transform</a> for transformation of one-dimensional data sets.
* For reference, see <b>Applied Numerical Linear Algebra</b>, ISBN 0898713897,
* chapter 6.
* <p>
* There are several conventions for the definition of FFT and inverse FFT,
* mainly on different coefficient and exponent. Here the equations are listed
* in the comments of the corresponding methods.</p>
* <p>
* We require the length of data set to be power of 2, this greatly simplifies
* and speeds up the code. Users can pad the data with zeros to meet this
* requirement. There are other flavors of FFT, for reference, see S. Winograd,
* <i>On computing the discrete Fourier transform</i>, Mathematics of Computation,
* 32 (1978), 175 - 199.</p>
*
* @version $Revision: 790243 $ $Date: 2009-07-01 12:03:28 -0400 (Wed, 01 Jul 2009) $
* @since 1.2
*/
public class FastFourierTransformer implements Serializable {

    /** Serializable version identifier. */
    static final long serialVersionUID = 5138259215438106000L;

    /** roots of unity */
    private RootsOfUnity roots = new RootsOfUnity();

    /**
     * Construct a default transformer.
     */
    public FastFourierTransformer() {
        super();
    }

    /**
     * Transform the given real data set.
     * <p>
     * The formula is $ y_n = \Sigma_{k=0}^{N-1} e^{-2 \pi i nk/N} x_k $
     * </p>
     *
     * @param f the real data array to be transformed
     * @return the complex transformed array
     * @throws IllegalArgumentException if any parameters are invalid
     */
    public Complex[] transform(double f[])
        throws IllegalArgumentException {
        return fft(f, false);
    }

    /**
     * Transform the given real function, sampled on the given interval.
     * <p>
     * The formula is $ y_n = \Sigma_{k=0}^{N-1} e^{-2 \pi i nk/N} x_k $
     * </p>
     *
     * @param f the function to be sampled and transformed
     * @param min the lower bound for the interval
     * @param max the upper bound for the interval
     * @param n the number of sample points
     * @return the complex transformed array
     * @throws FunctionEvaluationException if function cannot be evaluated
     * at some point
     * @throws IllegalArgumentException if any parameters are invalid
     */
    public Complex[] transform(UnivariateRealFunction f,
                               double min, double max, int n)
        throws FunctionEvaluationException, IllegalArgumentException {
        double data[] = sample(f, min, max, n);
        return fft(data, false);
    }

    /**
     * Transform the given complex data set.
     * <p>
     * The formula is $ y_n = \Sigma_{k=0}^{N-1} e^{-2 \pi i nk/N} x_k $
     * </p>
     *
     * @param f the complex data array to be transformed
     * @return the complex transformed array
     * @throws IllegalArgumentException if any parameters are invalid
     */
    public Complex[] transform(Complex f[])
        throws IllegalArgumentException {
        roots.computeOmega(f.length);
        return fft(f);
    }

    /**
     * Transform the given real data set.
     * <p>
     * The formula is $y_n = (1/\sqrt{N}) \Sigma_{k=0}^{N-1} e^{-2 \pi i nk/N} x_k$
     * </p>
     *
     * @param f the real data array to be transformed
     * @return the complex transformed array
     * @throws IllegalArgumentException if any parameters are invalid
     */
    public Complex[] transform2(double f[])
        throws IllegalArgumentException {

        double scaling_coefficient = 1.0 / Math.sqrt(f.length);
        return scaleArray(fft(f, false), scaling_coefficient);
    }

    /**
     * Transform the given real function, sampled on the given interval.
     * <p>
     * The formula is $y_n = (1/\sqrt{N}) \Sigma_{k=0}^{N-1} e^{-2 \pi i nk/N} x_k$
     * </p>
     *
     * @param f the function to be sampled and transformed
     * @param min the lower bound for the interval
     * @param max the upper bound for the interval
     * @param n the number of sample points
     * @return the complex transformed array
     * @throws FunctionEvaluationException if function cannot be evaluated
     * at some point
     * @throws IllegalArgumentException if any parameters are invalid
     */
    public Complex[] transform2(UnivariateRealFunction f,
                                double min, double max, int n)
        throws FunctionEvaluationException, IllegalArgumentException {

        double data[] = sample(f, min, max, n);
        double scaling_coefficient = 1.0 / Math.sqrt(n);
        return scaleArray(fft(data, false), scaling_coefficient);
    }

    /**
     * Transform the given complex data set.
     * <p>
     * The formula is $y_n = (1/\sqrt{N}) \Sigma_{k=0}^{N-1} e^{-2 \pi i nk/N} x_k$
     * </p>
     *
     * @param f the complex data array to be transformed
     * @return the complex transformed array
     * @throws IllegalArgumentException if any parameters are invalid
     */
    public Complex[] transform2(Complex f[])
        throws IllegalArgumentException {

        roots.computeOmega(f.length);
        double scaling_coefficient = 1.0 / Math.sqrt(f.length);
        return scaleArray(fft(f), scaling_coefficient);
    }

    /**
     * Inversely transform the given real data set.
     * <p>
     * The formula is $ x_k = (1/N) \Sigma_{n=0}^{N-1} e^{2 \pi i nk/N} y_n $
     * </p>
     *
     * @param f the real data array to be inversely transformed
     * @return the complex inversely transformed array
     * @throws IllegalArgumentException if any parameters are invalid
     */
    public Complex[] inversetransform(double f[])
        throws IllegalArgumentException {

        double scaling_coefficient = 1.0 / f.length;
        return scaleArray(fft(f, true), scaling_coefficient);
    }

    /**
     * Inversely transform the given real function, sampled on the given interval.
     * <p>
     * The formula is $ x_k = (1/N) \Sigma_{n=0}^{N-1} e^{2 \pi i nk/N} y_n $
     * </p>
     *
     * @param f the function to be sampled and inversely transformed
     * @param min the lower bound for the interval
     * @param max the upper bound for the interval
     * @param n the number of sample points
     * @return the complex inversely transformed array
     * @throws FunctionEvaluationException if function cannot be evaluated
     * at some point
     * @throws IllegalArgumentException if any parameters are invalid
     */
    public Complex[] inversetransform(UnivariateRealFunction f,
                                      double min, double max, int n)
        throws FunctionEvaluationException, IllegalArgumentException {

        double data[] = sample(f, min, max, n);
        double scaling_coefficient = 1.0 / n;
        return scaleArray(fft(data, true), scaling_coefficient);
    }

    /**
     * Inversely transform the given complex data set.
     * <p>
     * The formula is $ x_k = (1/N) \Sigma_{n=0}^{N-1} e^{2 \pi i nk/N} y_n $
     * </p>
     *
     * @param f the complex data array to be inversely transformed
     * @return the complex inversely transformed array
     * @throws IllegalArgumentException if any parameters are invalid
     */
    public Complex[] inversetransform(Complex f[])
        throws IllegalArgumentException {

        roots.computeOmega(-f.length);    // pass negative argument
        double scaling_coefficient = 1.0 / f.length;
        return scaleArray(fft(f), scaling_coefficient);
    }

    /**
     * Inversely transform the given real data set.
     * <p>
     * The formula is $x_k = (1/\sqrt{N}) \Sigma_{n=0}^{N-1} e^{2 \pi i nk/N} y_n$
     * </p>
     *
     * @param f the real data array to be inversely transformed
     * @return the complex inversely transformed array
     * @throws IllegalArgumentException if any parameters are invalid
     */
    public Complex[] inversetransform2(double f[])
        throws IllegalArgumentException {

        double scaling_coefficient = 1.0 / Math.sqrt(f.length);
        return scaleArray(fft(f, true), scaling_coefficient);
    }

    /**
     * Inversely transform the given real function, sampled on the given interval.
     * <p>
     * The formula is $x_k = (1/\sqrt{N}) \Sigma_{n=0}^{N-1} e^{2 \pi i nk/N} y_n$
     * </p>
     *
     * @param f the function to be sampled and inversely transformed
     * @param min the lower bound for the interval
     * @param max the upper bound for the interval
     * @param n the number of sample points
     * @return the complex inversely transformed array
     * @throws FunctionEvaluationException if function cannot be evaluated
     * at some point
     * @throws IllegalArgumentException if any parameters are invalid
     */
    public Complex[] inversetransform2(UnivariateRealFunction f,
                                       double min, double max, int n)
        throws FunctionEvaluationException, IllegalArgumentException {

        double data[] = sample(f, min, max, n);
        double scaling_coefficient = 1.0 / Math.sqrt(n);
        return scaleArray(fft(data, true), scaling_coefficient);
    }

    /**
     * Inversely transform the given complex data set.
     * <p>
     * The formula is $x_k = (1/\sqrt{N}) \Sigma_{n=0}^{N-1} e^{2 \pi i nk/N} y_n$
     * </p>
     *
     * @param f the complex data array to be inversely transformed
     * @return the complex inversely transformed array
     * @throws IllegalArgumentException if any parameters are invalid
     */
    public Complex[] inversetransform2(Complex f[])
        throws IllegalArgumentException {

        roots.computeOmega(-f.length);    // pass negative argument
        double scaling_coefficient = 1.0 / Math.sqrt(f.length);
        return scaleArray(fft(f), scaling_coefficient);
    }

    /**
     * Perform the base-4 Cooley-Tukey FFT algorithm (including inverse).
     *
     * @param f the real data array to be transformed
     * @param isInverse the indicator of forward or inverse transform
     * @return the complex transformed array
     * @throws IllegalArgumentException if any parameters are invalid
     */
    protected Complex[] fft(double f[], boolean isInverse)
        throws IllegalArgumentException {

        verifyDataSet(f);
        Complex F[] = new Complex[f.length];
        if (f.length == 1) {
            F[0] = new Complex(f[0], 0.0);
            return F;
        }

        // Rather than the naive real to complex conversion, pack 2N
        // real numbers into N complex numbers for better performance.
        int N = f.length >> 1;
        Complex c[] = new Complex[N];
        for (int i = 0; i < N; i++) {
            c[i] = new Complex(f[2*i], f[2*i+1]);
        }
        roots.computeOmega(isInverse ? -N : N);
        Complex z[] = fft(c);

        // reconstruct the FFT result for the original array
        roots.computeOmega(isInverse ? -2*N : 2*N);
        F[0] = new Complex(2 * (z[0].getReal() + z[0].getImaginary()), 0.0);
        F[N] = new Complex(2 * (z[0].getReal() - z[0].getImaginary()), 0.0);
        for (int i = 1; i < N; i++) {
            Complex A = z[N-i].conjugate();
            Complex B = z[i].add(A);
            Complex C = z[i].subtract(A);
            //Complex D = roots.getOmega(i).multiply(Complex.I);
            Complex D = new Complex(-roots.getOmegaImaginary(i),
                                    roots.getOmegaReal(i));
            F[i] = B.subtract(C.multiply(D));
            F[2*N-i] = F[i].conjugate();
        }

        return scaleArray(F, 0.5);
    }

    /**
     * Perform the base-4 Cooley-Tukey FFT algorithm (including inverse).
     *
     * @param data the complex data array to be transformed
     * @return the complex transformed array
     * @throws IllegalArgumentException if any parameters are invalid
     */
    protected Complex[] fft(Complex data[])
        throws IllegalArgumentException {

        int i, j, k, m, N = data.length;
        Complex A, B, C, D, E, F, z, f[] = new Complex[N];

        // initial simple cases
        verifyDataSet(data);
        if (N == 1) {
            f[0] = data[0];
            return f;
        }
        if (N == 2) {
            f[0] = data[0].add(data[1]);
            f[1] = data[0].subtract(data[1]);
            return f;
        }

        // permute original data array in bit-reversal order
        j = 0;
        for (i = 0; i < N; i++) {
            f[i] = data[j];
            k = N >> 1;
            while (j >= k && k > 0) {
                j -= k; k >>= 1;
            }
            j += k;
        }

        // the bottom base-4 round
        for (i = 0; i < N; i += 4) {
            A = f[i].add(f[i+1]);
            B = f[i+2].add(f[i+3]);
            C = f[i].subtract(f[i+1]);
            D = f[i+2].subtract(f[i+3]);
            E = C.add(D.multiply(Complex.I));
            F = C.subtract(D.multiply(Complex.I));
            f[i] = A.add(B);
            f[i+2] = A.subtract(B);
            // omegaCount indicates forward or inverse transform
            f[i+1] = roots.isForward() ? F : E;
            f[i+3] = roots.isForward() ? E : F;
        }

        // iterations from bottom to top take O(N*logN) time
        for (i = 4; i < N; i <<= 1) {
            m = N / (i<<1);
            for (j = 0; j < N; j += i<<1) {
                for (k = 0; k < i; k++) {
                    //z = f[i+j+k].multiply(roots.getOmega(k*m));
                    final int k_times_m = k*m;
                    final double omega_k_times_m_real = roots.getOmegaReal(k_times_m);
                    final double omega_k_times_m_imaginary = roots.getOmegaImaginary(k_times_m);
                    //z = f[i+j+k].multiply(omega[k*m]);
                    z = new Complex(
                        f[i+j+k].getReal() * omega_k_times_m_real -
                        f[i+j+k].getImaginary() * omega_k_times_m_imaginary,
                        f[i+j+k].getReal() * omega_k_times_m_imaginary +
                        f[i+j+k].getImaginary() * omega_k_times_m_real);
                 
                    f[i+j+k] = f[j+k].subtract(z);
                    f[j+k] = f[j+k].add(z);
                }
            }
        }
        return f;
    }

    /**
     * Sample the given univariate real function on the given interval.
     * <p>
     * The interval is divided equally into N sections and sample points
     * are taken from min to max-(max-min)/N. Usually f(x) is periodic
     * such that f(min) = f(max) (note max is not sampled), but we don't
     * require that.</p>
     *
     * @param f the function to be sampled
     * @param min the lower bound for the interval
     * @param max the upper bound for the interval
     * @param n the number of sample points
     * @return the samples array
     * @throws FunctionEvaluationException if function cannot be evaluated
     * at some point
     * @throws IllegalArgumentException if any parameters are invalid
     */
    public static double[] sample(UnivariateRealFunction f,
                                  double min, double max, int n)
        throws FunctionEvaluationException, IllegalArgumentException {

        if (n <= 0) {
            throw MathRuntimeException.createIllegalArgumentException(
                    "number of sample is not positive: {0}",
                    n);
        }
        verifyInterval(min, max);

        double s[] = new double[n];
        double h = (max - min) / n;
        for (int i = 0; i < n; i++) {
            s[i] = f.value(min + i * h);
        }
        return s;
    }

    /**
     * Multiply every component in the given real array by the
     * given real number. The change is made in place.
     *
     * @param f the real array to be scaled
     * @param d the real scaling coefficient
     * @return a reference to the scaled array
     */
    public static double[] scaleArray(double f[], double d) {
        for (int i = 0; i < f.length; i++) {
            f[i] *= d;
        }
        return f;
    }

    /**
     * Multiply every component in the given complex array by the
     * given real number. The change is made in place.
     *
     * @param f the complex array to be scaled
     * @param d the real scaling coefficient
     * @return a reference to the scaled array
     */
    public static Complex[] scaleArray(Complex f[], double d) {
        for (int i = 0; i < f.length; i++) {
            f[i] = new Complex(d * f[i].getReal(), d * f[i].getImaginary());
        }
        return f;
    }

    /**
     * Returns true if the argument is power of 2.
     *
     * @param n the number to test
     * @return true if the argument is power of 2
     */
    public static boolean isPowerOf2(long n) {
        return (n > 0) && ((n & (n - 1)) == 0);
    }

    /**
     * Verifies that the data set has length of power of 2.
     *
     * @param d the data array
     * @throws IllegalArgumentException if array length is not power of 2
     */
    public static void verifyDataSet(double d[]) throws IllegalArgumentException {
        if (!isPowerOf2(d.length)) {
            throw MathRuntimeException.createIllegalArgumentException(
                    "{0} is not a power of 2, consider padding for fix",
                    d.length);
        }      
    }

    /**
     * Verifies that the data set has length of power of 2.
     *
     * @param o the data array
     * @throws IllegalArgumentException if array length is not power of 2
     */
    public static void verifyDataSet(Object o[]) throws IllegalArgumentException {
        if (!isPowerOf2(o.length)) {
            throw MathRuntimeException.createIllegalArgumentException(
                    "{0} is not a power of 2, consider padding for fix",
                    o.length);
        }      
    }

    /**
     * Verifies that the endpoints specify an interval.
     *
     * @param lower lower endpoint
     * @param upper upper endpoint
     * @throws IllegalArgumentException if not interval
     */
    public static void verifyInterval(double lower, double upper)
        throws IllegalArgumentException {

        if (lower >= upper) {
            throw MathRuntimeException.createIllegalArgumentException(
                    "endpoints do not specify an interval: [{0}, {1}]",
                    lower, upper);
        }      
    }
   
    /**
     * Performs a multi-dimensional Fourier transform on a given array.
     * Use {@link #inversetransform2(Complex[])} and
     * {@link #transform2(Complex[])} in a row-column implementation
     * in any number of dimensions with O(N&times;log(N)) complexity with
     * N=n<sub>1</sub>&times;n<sub>2</sub>&times;n<sub>3</sub>&times;...&times;n<sub>d</sub>,
     * n<sub>x</sub>=number of elements in dimension x,
     * and d=total number of dimensions.
     *
     * @param mdca Multi-Dimensional Complex Array id est Complex[][][][]
     * @param forward inverseTransform2 is preformed if this is false
     * @return transform of mdca as a Multi-Dimensional Complex Array id est Complex[][][][]
     * @throws IllegalArgumentException if any dimension is not a power of two
     */
    public Object mdfft(Object mdca, boolean forward)
        throws IllegalArgumentException {
        MultiDimensionalComplexMatrix mdcm = (MultiDimensionalComplexMatrix)
                new MultiDimensionalComplexMatrix(mdca).clone();
        int[] dimensionSize = mdcm.getDimensionSizes();
        //cycle through each dimension
        for (int i = 0; i < dimensionSize.length; i++) {
            mdfft(mdcm, forward, i, new int[0]);
        }
        return mdcm.getArray();
    }
   
    /**
     * Performs one dimension of a multi-dimensional Fourier transform.
     *
     * @param mdcm input matrix
     * @param forward inverseTransform2 is preformed if this is false
     * @param d index of the dimension to process
     * @param subVector recursion subvector
     * @throws IllegalArgumentException if any dimension is not a power of two
     */
    private void mdfft(MultiDimensionalComplexMatrix mdcm, boolean forward,
                       int d, int[] subVector)
        throws IllegalArgumentException {
        int[] dimensionSize = mdcm.getDimensionSizes();
        //if done
        if (subVector.length == dimensionSize.length) {
            Complex[] temp = new Complex[dimensionSize[d]];
            for (int i = 0; i < dimensionSize[d]; i++) {
                //fft along dimension d
                subVector[d] = i;
                temp[i] = mdcm.get(subVector);
            }
           
            if (forward)
                temp = transform2(temp);
            else
                temp = inversetransform2(temp);
           
            for (int i = 0; i < dimensionSize[d]; i++) {
                subVector[d] = i;
                mdcm.set(temp[i], subVector);
            }
        } else {
            int[] vector = new int[subVector.length + 1];
            System.arraycopy(subVector, 0, vector, 0, subVector.length);
            if (subVector.length == d) {
                //value is not important once the recursion is done.
                //then an fft will be applied along the dimension d.
                vector[d] = 0;
                mdfft(mdcm, forward, d, vector);
            } else {
                for (int i = 0; i < dimensionSize[subVector.length]; i++) {
                    vector[subVector.length] = i;
                    //further split along the next dimension
                    mdfft(mdcm, forward, d, vector);
                }
            }
        }
        return;
    }

    /**
     * Complex matrix implementation.
     * Not designed for synchronized access
     * may eventually be replaced by jsr-83 of the java community process
     * http://jcp.org/en/jsr/detail?id=83
     * may require additional exception throws for other basic requirements.
     */
    private static class MultiDimensionalComplexMatrix
        implements Cloneable {

        /** Size in all dimensions. */
        protected int[] dimensionSize;

        /** Storage array. */
        protected Object multiDimensionalComplexArray;

        /** Simple constructor.
         * @param multiDimensionalComplexArray array containing the matrix elements
         */
        public MultiDimensionalComplexMatrix(Object multiDimensionalComplexArray) {

            this.multiDimensionalComplexArray = multiDimensionalComplexArray;

            // count dimensions
            int numOfDimensions = 0;
            for (Object lastDimension = multiDimensionalComplexArray;
                 lastDimension instanceof Object[];) {
                final Object[] array = (Object[]) lastDimension;
                numOfDimensions++;
                lastDimension = array[0];
            }

            // allocate array with exact count
            dimensionSize = new int[numOfDimensions];

            // fill array
            numOfDimensions = 0;
            for (Object lastDimension = multiDimensionalComplexArray;
                 lastDimension instanceof Object[];) {
                final Object[] array = (Object[]) lastDimension;
                dimensionSize[numOfDimensions++] = array.length;
                lastDimension = array[0];
            }

        }

        /**
         * Get a matrix element.
         * @param vector indices of the element
         * @return matrix element
         * @exception IllegalArgumentException if dimensions do not match
         */
        public Complex get(int... vector)
            throws IllegalArgumentException {
            if (vector == null) {
                if (dimensionSize.length > 0) {
                    throw MathRuntimeException.createIllegalArgumentException(
                            "some dimensions don't match: {0} != {1}",
                            0, dimensionSize.length);
                }
                return null;
            }
            if (vector.length != dimensionSize.length) {
                throw MathRuntimeException.createIllegalArgumentException(
                        "some dimensions don't match: {0} != {1}",
                        vector.length, dimensionSize.length);
            }
           
            Object lastDimension = multiDimensionalComplexArray;
           
            for (int i = 0; i < dimensionSize.length; i++) {
                lastDimension = ((Object[]) lastDimension)[vector[i]];
            }
            return (Complex) lastDimension;
        }
       
        /**
         * Set a matrix element.
         * @param magnitude magnitude of the element
         * @param vector indices of the element
         * @return the previous value
         * @exception IllegalArgumentException if dimensions do not match
         */
        public Complex set(Complex magnitude, int... vector)
            throws IllegalArgumentException {
            if (vector == null) {
                if (dimensionSize.length > 0) {
                    throw MathRuntimeException.createIllegalArgumentException(
                            "some dimensions don't match: {0} != {1}",
                            0, dimensionSize.length);
                }
                return null;
            }
            if (vector.length != dimensionSize.length) {
                throw MathRuntimeException.createIllegalArgumentException(
                        "some dimensions don't match: {0} != {1}",
                        vector.length,dimensionSize.length);
            }

            Object[] lastDimension = (Object[]) multiDimensionalComplexArray;
            for (int i = 0; i < dimensionSize.length - 1; i++) {
                lastDimension = (Object[]) lastDimension[vector[i]];
            }

            Complex lastValue = (Complex) lastDimension[vector[dimensionSize.length - 1]];
            lastDimension[vector[dimensionSize.length - 1]] = magnitude;

            return lastValue;
        }

        /**
         * Get the size in all dimensions.
         * @return size in all dimensions
         */
        public int[] getDimensionSizes() {
            return dimensionSize.clone();
        }

        /**
         * Get the underlying storage array
         * @return underlying storage array
         */
        public Object getArray() {
            return multiDimensionalComplexArray;
        }

        /** {@inheritDoc} */
        @Override
        public Object clone() {
            MultiDimensionalComplexMatrix mdcm =
                    new MultiDimensionalComplexMatrix(Array.newInstance(
                    Complex.class, dimensionSize));
            clone(mdcm);
            return mdcm;
        }
       
        /**
         * Copy contents of current array into mdcm.
         * @param mdcm array where to copy data
         */
        private void clone(MultiDimensionalComplexMatrix mdcm) {
            int[] vector = new int[dimensionSize.length];
            int size = 1;
            for (int i = 0; i < dimensionSize.length; i++) {
                size *= dimensionSize[i];
            }
            int[][] vectorList = new int[size][dimensionSize.length];
            for (int[] nextVector: vectorList) {
                System.arraycopy(vector, 0, nextVector, 0,
                                 dimensionSize.length);
                for (int i = 0; i < dimensionSize.length; i++) {
                    vector[i]++;
                    if (vector[i] < dimensionSize[i]) {
                        break;
                    } else {
                        vector[i] = 0;
                    }
                }
            }
           
            for (int[] nextVector: vectorList) {
                mdcm.set(get(nextVector), nextVector);
            }
        }
    }
   
   
    /** Computes the n<sup>th</sup> roots of unity.
     * A cache of already computed values is maintained.
     */
    private static class RootsOfUnity implements Serializable {

      /** Serializable version id. */
      private static final long serialVersionUID = 6404784357747329667L;

      /** Number of roots of unity. */
      private int      omegaCount;

      /** Real part of the roots. */
      private double[] omegaReal;

      /** Imaginary part of the roots for forward transform. */
      private double[] omegaImaginaryForward;

      /** Imaginary part of the roots for reverse transform. */
      private double[] omegaImaginaryInverse;

      /** Forward/reverse indicator. */
      private boolean  isForward;

      /**
       * Build an engine for computing then <sup>th</sup> roots of unity
       */
      public RootsOfUnity() {
       
        omegaCount = 0;
        omegaReal = null;
        omegaImaginaryForward = null;
        omegaImaginaryInverse = null;
        isForward = true;
       
      }

      /**
       * Check if computation has been done for forward or reverse transform.
       * @return true if computation has been done for forward transform
       * @throws IllegalStateException if no roots of unity have been computed yet
       */
      public synchronized boolean isForward() throws IllegalStateException {
         
        if (omegaCount == 0) {
          throw MathRuntimeException.createIllegalStateException(
                  "roots of unity have not been computed yet");
        }       
        return isForward;
       
      }
     
      /** Computes the n<sup>th</sup> roots of unity.
       * <p>The computed omega[] = { 1, w, w<sup>2</sup>, ... w<sup>(n-1)</sup> } where
       * w = exp(-2 &pi; i / n), i = &sqrt;(-1).</p>
       * <p>Note that n is positive for
       * forward transform and negative for inverse transform.</p>
       * @param n number of roots of unity to compute,
       * positive for forward transform, negative for inverse transform
       * @throws IllegalArgumentException if n = 0
       */
      public synchronized void computeOmega(int n) throws IllegalArgumentException {

        if (n == 0) {
          throw MathRuntimeException.createIllegalArgumentException(
                  "cannot compute 0-th root of unity, indefinite result");
        }

        isForward = (n > 0);
       
        // avoid repetitive calculations
        final int absN = Math.abs(n);
       
        if (absN == omegaCount) {
            return;
        }

        // calculate everything from scratch, for both forward and inverse versions
        final double t    = 2.0 * Math.PI / absN;
        final double cosT = Math.cos(t);
        final double sinT = Math.sin(t);
        omegaReal             = new double[absN];
        omegaImaginaryForward = new double[absN];
        omegaImaginaryInverse = new double[absN];
        omegaReal[0]             = 1.0;
        omegaImaginaryForward[0] = 0.0;
        omegaImaginaryInverse[0] = 0.0;
        for (int i = 1; i < absN; i++) {
          omegaReal[i] =
            omegaReal[i-1] * cosT + omegaImaginaryForward[i-1] * sinT;
          omegaImaginaryForward[i] =
             omegaImaginaryForward[i-1] * cosT - omegaReal[i-1] * sinT;
          omegaImaginaryInverse[i] = -omegaImaginaryForward[i];
        }
        omegaCount = absN;

      }

      /**
       * Get the real part of the k<sup>th</sup> n<sup>th</sup> root of unity
       * @param k index of the n<sup>th</sup> root of unity
       * @return real part of the k<sup>th</sup> n<sup>th</sup> root of unity
       * @throws IllegalStateException if no roots of unity have been computed yet
       * @throws IllegalArgumentException if k is out of range
       */
      public synchronized double getOmegaReal(int k)
        throws IllegalStateException, IllegalArgumentException {
       
        if (omegaCount == 0) {
            throw MathRuntimeException.createIllegalStateException(
                    "roots of unity have not been computed yet");
        }
        if ((k < 0) || (k >= omegaCount)) {
            throw MathRuntimeException.createIllegalArgumentException(
                    "out of range root of unity index {0} (must be in [{1};{2}])",
                    k, 0, omegaCount - 1);
        }
       
        return omegaReal[k];
       
      }

      /**
       * Get the imaginary part of the k<sup>th</sup> n<sup>th</sup> root of unity
       * @param k index of the n<sup>th</sup> root of unity
       * @return imaginary part of the k<sup>th</sup> n<sup>th</sup> root of unity
       * @throws IllegalStateException if no roots of unity have been computed yet
       * @throws IllegalArgumentException if k is out of range
       */
      public synchronized double getOmegaImaginary(int k)
        throws IllegalStateException, IllegalArgumentException {
     
        if (omegaCount == 0) {
            throw MathRuntimeException.createIllegalStateException(
                    "roots of unity have not been computed yet");
        }
        if ((k < 0) || (k >= omegaCount)) {
          throw MathRuntimeException.createIllegalArgumentException(
                  "out of range root of unity index {0} (must be in [{1};{2}])",
                  k, 0, omegaCount - 1);
        }

        return (isForward) ?
            omegaImaginaryForward[k] : omegaImaginaryInverse[k];
       
      }

    }
   
}
TOP

Related Classes of org.apache.commons.math.transform.FastFourierTransformer$MultiDimensionalComplexMatrix

TOP
Copyright © 2018 www.massapi.com. All rights reserved.
All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.