[SciPy-dev] tanh(j*pi/2)

Robert Kern robert.kern@gmail....
Wed Jun 10 02:13:54 CDT 2009


On Wed, Jun 10, 2009 at 01:49, David Goldsmith<d_l_goldsmith@yahoo.com> wrote:
>
> --- On Tue, 6/9/09, Robert Kern <robert.kern@gmail.com> wrote:
>
>> Umm, explaining the phenomenon necessarily entails showing
>> examples of
>> the library functions it applies to. This particular
>
> Good, we're on the same page. :-)
>
>> phenomenon is
>> *about* how the standard, furnished mathematical functions
>> behave,
>> nothing else. But I don't think a full, explicit list is
>
> "Nothing else"?  It doesn't apply to user-written, rational functions and algorithms that may divide by small-modulus numbers that should theoretically be zero?

Ah, that's not the real issue here that I am suggesting to be
documented. The result of cos(np.pi/2) is *not* theoretically 0. This
is because np.pi is not exactly the transcendental real number pi.
This has nothing to do with the quality of implementation of np.cos().
The answer that is given is more accurate than 0.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
  -- Umberto Eco


More information about the Scipy-dev mailing list