MantissaExponent
- To: mathgroup at yoda.physics.unc.edu
- Subject: MantissaExponent
- From: jesroj at wmatem.eis.uva.es
- Date: Tue, 29 Mar 94 14:10:13 +0100
While working whith the MantissaExponent function I have
found the calculation it carries out to be totally correct for
the interests of Numerical Analysis, but not for those of the
Symbolic Calculus, which requires an exactly accurate
calculation. For example, I have used it in a rounding off
operation where the difference between
1.005 and 1.00499999...
is crucial. I solved the problem by using the RealDigits function
which is not so convenient. I believe that you could easily build
a MantissaExponent with a correct code using this same function.
The following is a short and simple example of what I mean:
In[1] := MantissaExponent[1.005]
Out[1] = {0.1005, 1}
In[2] := MantissaExponent[1.005][[1]] - 0.1005
Out[2] = -(1.387778780781446*10^-17)
Many thanks,
Jesus Rojo
E.T.S. de Ing. Industriales
Paseo del Cauce s/n
47011 VALLADOLID
( SPAIN )
tel: SPAIN.83.42 33 90
fax: SPAIN.83.42 34 06
e-mail: jesroj at wmatem.eis.uva.es