RealDuplication.java
/*
* Licensed to the Hipparchus project under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The Hipparchus project 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
*
* https://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.hipparchus.special.elliptic.carlson;
import org.hipparchus.util.FastMath;
/** Duplication algorithm for Carlson symmetric forms.
* <p>
* The algorithms are described in B. C. Carlson 1995 paper
* "Numerical computation of real or complex elliptic integrals", with
* improvements described in the appendix of B. C. Carlson and James FitzSimons
* 2000 paper "Reduction theorems for elliptic integrands with the square root
* of two quadratic factors". They are also described in
* <a href="https://dlmf.nist.gov/19.36#i">section 19.36(i)</a>
* of Digital Library of Mathematical Functions.
* </p>
* @since 2.0
*/
abstract class RealDuplication {
/** Max number of iterations. */
private static final int M_MAX = 16;
/** Symmetric variables of the integral, plus mean point. */
private final double[] initialVA;
/** Convergence criterion. */
private final double q;
/** Constructor.
* @param v symmetric variables of the integral
*/
RealDuplication(final double... v) {
final int n = v.length;
initialVA = new double[n + 1];
System.arraycopy(v, 0, initialVA, 0, n);
initialMeanPoint(initialVA);
double max = 0;
final double a0 = initialVA[n];
for (final double vi : v) {
max = FastMath.max(max, FastMath.abs(a0 - vi));
}
this.q = convergenceCriterion(FastMath.ulp(1.0), max);
}
/** Get the i<sup>th</sup> symmetric variable.
* @param i index of the variable
* @return i<sup>th</sup> symmetric variable
*/
protected double getVi(final int i) {
return initialVA[i];
}
/** Compute initial mean point.
* <p>
* The initial mean point is put as the last array element
* </>
* @param va symmetric variables of the integral (plus placeholder for initial mean point)
*/
protected abstract void initialMeanPoint(double[] va);
/** Compute convergence criterion.
* @param r relative tolerance
* @param max max(|a0-v[i]|)
* @return convergence criterion
*/
protected abstract double convergenceCriterion(double r, double max);
/** Update reduced variables in place.
* <ul>
* <li>vₘ₊₁|i] ← (vₘ[i] + λₘ) / 4</li>
* <li>aₘ₊₁ ← (aₘ + λₘ) / 4</li>
* </ul>
* @param m iteration index
* @param vaM reduced variables and mean point (updated in place)
* @param sqrtM square roots of reduced variables
* @param fourM 4<sup>m</sup>
*/
protected abstract void update(int m, double[] vaM, double[] sqrtM, double fourM);
/** Evaluate integral.
* @param va0 initial symmetric variables and mean point of the integral
* @param aM reduced mean point
* @param fourM 4<sup>m</sup>
* @return integral value
*/
protected abstract double evaluate(double[] va0, double aM, double fourM);
/** Compute Carlson elliptic integral.
* @return Carlson elliptic integral
*/
public double integral() {
// duplication iterations
final int n = initialVA.length - 1;
final double[] vaM = initialVA.clone();
final double[] sqrtM = new double[n];
double fourM = 1.0;
for (int m = 0; m < M_MAX; ++m) {
if (m > 0 && q < fourM * FastMath.abs(vaM[n])) {
// convergence reached
break;
}
// apply duplication once more
// (we know that {Field}Complex.sqrt() returns the root with nonnegative real part)
for (int i = 0; i < n; ++i) {
sqrtM[i] = FastMath.sqrt(vaM[i]);
}
update(m, vaM, sqrtM, fourM);
fourM *= 4;
}
return evaluate(initialVA, vaM[n], fourM);
}
}