# Copyright(c) 1986 Association of Universities for Research in Astronomy Inc. include # FP_NORMD -- Normalize a double precision number x to the value NORMX, in the # range [1-10). EXPON is returned such that x = normx * (10.0d0 ** expon). procedure fp_normd (x, normx, expon) double x # number to be normalized double normx # X normalized to the range 1-10 (output) int expon # exponent of normalized X double absx, tol begin tol = EPSILOND * 10.0D0 absx = abs (x) expon = 0 if (absx > 0) { while (absx < (1.0D0 - tol)) { absx = absx * 10.0D0 expon = expon - 1 if (absx == 0.0D0) { # check for underflow to zero normx = 0 expon = 0 return } } while (absx >= (10.0D0 + tol)) { absx = absx / 10.0D0 expon = expon + 1 } } if (x < 0) normx = -absx else normx = absx end