From: SourceForge.net <noreply@so...>  20091029 23:29:46

Bugs item #2889126, was opened at 20091030 00:29 Message generated for change (Tracker Item Submitted) made by crategus You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=104933&aid=2889126&group_id=4933 Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: Lisp Core  Simplification Group: None Status: Open Resolution: None Priority: 5 Private: No Submitted By: Dieter Kaiser (crategus) Assigned to: Nobody/Anonymous (nobody) Summary: Numerical evaluation of Bessel functions inconsistent Initial Comment: The numerical evaluation of the Bessel functions is not implemented consistently. These expressions evaluate numerically as expected: (%i28) bessel_j(1,0.5); (%o28) .2422684576748739 (%i29) bessel_j(1,0.5+%i); (%o29) .5124137767280906 %i + .3392601907198862 (%i30) bessel_j(0.5,%i); (%o30) .6630362720267232 %i + .6630362720267233 But these expressions should evaluate numerically too: (%i33) bessel_j(1/2,0.5); (%o33) bessel_j(1/2,0.5) (%i34) bessel_j(1,0.5+1/2*%i); (%o34) bessel_j(1,%i/2+0.5) (%i35) bessel_j(0.5,0.5+1/2*%i); (%o35) bessel_j(0.5,%i/2+0.5) The convention is, that the function evaluates numerically if all arguments are numbers (including Maxima rationals and bigfloats) and at least one of the numbers is a float or a bigfloat or the flag $numer is TRUE. The problem is that the function besselnumericalevalp does not count Maxima rationals as numbers. We have the function floatnumericalevalp and friends, which check more carefully if a function should be evaluated numerically. With the help of these functions we can implement the numerical evaluation of the Bessel functions more consistent. Dieter Kaiser  You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=104933&aid=2889126&group_id=4933 