/* -*- mode: c++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */ /* Copyright (C) 2015 Andres Hernandez This file is part of QuantLib, a free-software/open-source library for financial quantitative analysts and developers - http://quantlib.org/ QuantLib is free software: you can redistribute it and/or modify it under the terms of the QuantLib license. You should have received a copy of the license along with this program; if not, please email . The license is also available online at . This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the license for more details. */ /*! \file isotropicrandomwalk.hpp \brief Isotropic random walk */ #ifndef quantlib_isotropic_random_walk_hpp #define quantlib_isotropic_random_walk_hpp #include #include #include #include namespace QuantLib { //! Isotropic random walk /*! A variate is used to draw from a random element of a probability distribution. The draw corresponds to the radius of a d-dimensional sphere. The position on the surface of the d-dimensional sphere is randomly chosen with all points on the surface having the same probability, i.e. all directions are isotropic and the step is randomly drawn from the given variate. */ template class IsotropicRandomWalk { public: typedef boost::variate_generator VariateGenerator; IsotropicRandomWalk(const Engine& eng, Distribution dist, Size dim, const Array& weights = Array(), unsigned long seed = 0) : variate_(eng, dist), rng_(seed), weights_(weights), dim_(dim) { if (weights_.empty()) weights_ = Array(dim, 1.0); else QL_REQUIRE(dim_ == weights_.size(), "Invalid weights"); } template inline void nextReal(InputIterator first) const { Real radius = variate_(); Array::const_iterator weight = weights_.begin(); if (dim_ > 1) { //Isotropic random direction Real phi = M_PI*rng_.nextReal(); for (Size i = 0; i < dim_ - 2; i++) { *first++ = radius*cos(phi)*(*weight++); radius *= sin(phi); phi = M_PI*rng_.nextReal(); } *first++ = radius*cos(2.0*phi)*(*weight++); *first = radius*sin(2.0*phi)*(*weight); } else { if (rng_.nextReal() < 0.5) *first = -radius*(*weight); else *first = radius*(*weight); } } inline void setDimension(Size dim) { dim_ = dim; weights_ = Array(dim, 1.0); } inline void setDimension(Size dim, const Array& weights) { QL_REQUIRE(dim == weights.size(), "Invalid weights"); dim_ = dim; weights_ = weights; } /*! The isotropic random walk will not adjust its draw to be within the lower and upper bounds, but if the limits are provided, they are used to rescale the sphere so as to make it to an ellipsoid, with different radius in different dimensions. */ inline void setDimension(Size dim, const Array& lowerBound, const Array& upperBound) { QL_REQUIRE(dim == lowerBound.size(), "Incompatible dimension and lower bound"); QL_REQUIRE(dim == upperBound.size(), "Incompatible dimension and upper bound"); //Find largest bound Array bounds = upperBound - lowerBound; Real maxBound = bounds[0]; for (Size j = 1; j < dim; j++) { if (bounds[j] > maxBound) maxBound = bounds[j]; } //weights by dimension is the size of the bound //divided by the largest bound maxBound = 1.0 / maxBound; bounds *= maxBound; setDimension(dim, bounds); } protected: mutable VariateGenerator variate_; MersenneTwisterUniformRng rng_; Array weights_; Size dim_; }; } #endif