/* round2n.c (round floating-point number to nearest power of two) */ /*********************************************************************** * This code is part of GLPK (GNU Linear Programming Kit). * * Copyright (C) 2000-2013 Andrew Makhorin, Department for Applied * Informatics, Moscow Aviation Institute, Moscow, Russia. All rights * reserved. E-mail: . * * GLPK is free software: you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by * the Free Software Foundation, either version 3 of the License, or * (at your option) any later version. * * GLPK 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 GNU General Public * License for more details. * * You should have received a copy of the GNU General Public License * along with GLPK. If not, see . ***********************************************************************/ #include "env.h" #include "misc.h" /*********************************************************************** * NAME * * round2n - round floating-point number to nearest power of two * * SYNOPSIS * * #include "misc.h" * double round2n(double x); * * RETURNS * * Given a positive floating-point value x the routine round2n returns * 2^n such that |x - 2^n| is minimal. * * EXAMPLES * * round2n(10.1) = 2^3 = 8 * round2n(15.3) = 2^4 = 16 * round2n(0.01) = 2^(-7) = 0.0078125 * * BACKGROUND * * Let x = f * 2^e, where 0.5 <= f < 1 is a normalized fractional part, * e is an integer exponent. Then, obviously, 0.5 * 2^e <= x < 2^e, so * if x - 0.5 * 2^e <= 2^e - x, we choose 0.5 * 2^e = 2^(e-1), and 2^e * otherwise. The latter condition can be written as 2 * x <= 1.5 * 2^e * or 2 * f * 2^e <= 1.5 * 2^e or, finally, f <= 0.75. */ double round2n(double x) { int e; double f; xassert(x > 0.0); f = frexp(x, &e); return ldexp(1.0, f <= 0.75 ? e-1 : e); } /* eof */