1 /* 2 * Licensed to the Apache Software Foundation (ASF) under one or more 3 * contributor license agreements. See the NOTICE file distributed with 4 * this work for additional information regarding copyright ownership. 5 * The ASF licenses this file to You under the Apache License, Version 2.0 6 * (the "License"); you may not use this file except in compliance with 7 * the License. You may obtain a copy of the License at 8 * 9 * http://www.apache.org/licenses/LICENSE-2.0 10 * 11 * Unless required by applicable law or agreed to in writing, software 12 * distributed under the License is distributed on an "AS IS" BASIS, 13 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 * See the License for the specific language governing permissions and 15 * limitations under the License. 16 */ 17 18 package org.apache.commons.math4.legacy.analysis.solvers; 19 20 /** 21 * Implements the <em>Regula Falsi</em> or <em>False position</em> method for 22 * root-finding (approximating a zero of a univariate real function). It is a 23 * modified {@link SecantSolver <em>Secant</em>} method. 24 * 25 * <p>The <em>Regula Falsi</em> method is included for completeness, for 26 * testing purposes, for educational purposes, for comparison to other 27 * algorithms, etc. It is however <strong>not</strong> intended to be used 28 * for actual problems, as one of the bounds often remains fixed, resulting 29 * in very slow convergence. Instead, one of the well-known modified 30 * <em>Regula Falsi</em> algorithms can be used ({@link IllinoisSolver 31 * <em>Illinois</em>} or {@link PegasusSolver <em>Pegasus</em>}). These two 32 * algorithms solve the fundamental issues of the original <em>Regula 33 * Falsi</em> algorithm, and greatly out-performs it for most, if not all, 34 * (practical) functions. 35 * 36 * <p>Unlike the <em>Secant</em> method, the <em>Regula Falsi</em> guarantees 37 * convergence, by maintaining a bracketed solution. Note however, that due to 38 * the finite/limited precision of Java's {@link Double double} type, which is 39 * used in this implementation, the algorithm may get stuck in a situation 40 * where it no longer makes any progress. Such cases are detected and result 41 * in a {@code ConvergenceException} exception being thrown. In other words, 42 * the algorithm theoretically guarantees convergence, but the implementation 43 * does not.</p> 44 * 45 * <p>The <em>Regula Falsi</em> method assumes that the function is continuous, 46 * but not necessarily smooth.</p> 47 * 48 * <p>Implementation based on the following article: M. Dowell and P. Jarratt, 49 * <em>A modified regula falsi method for computing the root of an 50 * equation</em>, BIT Numerical Mathematics, volume 11, number 2, 51 * pages 168-174, Springer, 1971.</p> 52 * 53 * @since 3.0 54 */ 55 public class RegulaFalsiSolver extends BaseSecantSolver { 56 57 /** Construct a solver with default accuracy (1e-6). */ 58 public RegulaFalsiSolver() { 59 super(DEFAULT_ABSOLUTE_ACCURACY, Method.REGULA_FALSI); 60 } 61 62 /** 63 * Construct a solver. 64 * 65 * @param absoluteAccuracy Absolute accuracy. 66 */ 67 public RegulaFalsiSolver(final double absoluteAccuracy) { 68 super(absoluteAccuracy, Method.REGULA_FALSI); 69 } 70 71 /** 72 * Construct a solver. 73 * 74 * @param relativeAccuracy Relative accuracy. 75 * @param absoluteAccuracy Absolute accuracy. 76 */ 77 public RegulaFalsiSolver(final double relativeAccuracy, 78 final double absoluteAccuracy) { 79 super(relativeAccuracy, absoluteAccuracy, Method.REGULA_FALSI); 80 } 81 82 /** 83 * Construct a solver. 84 * 85 * @param relativeAccuracy Relative accuracy. 86 * @param absoluteAccuracy Absolute accuracy. 87 * @param functionValueAccuracy Maximum function value error. 88 */ 89 public RegulaFalsiSolver(final double relativeAccuracy, 90 final double absoluteAccuracy, 91 final double functionValueAccuracy) { 92 super(relativeAccuracy, absoluteAccuracy, functionValueAccuracy, Method.REGULA_FALSI); 93 } 94 }