MathGroup Archive 2006

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Real numerical computations

  • To: mathgroup at smc.vnet.net
  • Subject: [mg71812] [mg71797] Re: Real numerical computations
  • From: "David W. Cantrell" <DWCantrell at sigmaxi.net>
  • Date: Thu, 30 Nov 2006 06:05:02 -0500 (EST)
  • References: <ekh78f$sa9$1@smc.vnet.net>

"Mark Harder" <harderm at onid.orst.edu> wrote:
> Jose,
>         There are 3 cube roots of unity, all of which can be found using
> the Root[] function (look it up).  Same thing goes for cube roots of -1.
> Run the following:
>
> Thread[Root[#^3 - 1 &, {1, 2, 3} ] ]
>
> And
>
> Thread[Root[#^3 + 1 &, {1, 2, 3} ] ]
>
> The first gives the cube roots of 1 (i.e. x^3 - 1 ==0).  The second gives
> the cube roots of -1, so you need only do Root[#^3+1&,1] to get the real
> root.
> Also, if you use exact math --  (-1)^(1/3)  --   you will get the real
> root.

No, you don't. You get the principal root.

FWIW, if you're wanting to work with just real cube roots of reals, rather
than bothering to load a package like RealOnly, you could simply define
your own function:

RealCbrt[x_]:= Sign[x]*Abs[x]^(1/3)

David W. Cantrell


> Maybe someone else understands why the finite precision math
> returns only a complex root, and a rather inaccurate one at that -- try
> raising the answer you got (using default precision) to the 3rd power to
> see how inaccurate. Play around with extra precision in your expressions
> to see how to make this better.
>
> Mark Harder
>
> > -----Original Message-----
> > From: José Carlos Santos [mailto:jcsantos at fc.up.pt]
> > Sent: Monday, November 27, 2006 1:04 AM
> > To: mathgroup at smc.vnet.net
> > Subject:  Real numerical computations
> >
> > Hi all,
> >
> > With Mathematica, if I type
> >
> > N[(-1.)^(1/3)]
> >
> > I get
> >
> > 0.5 + 0.866025 i
> >
> > This is correct, of course, but I would like to get -1. instead. How do
> > I tell Mathematica that I want a real result (if there is one)?
> >
> > Best regards,
> >
> > Jose Carlos Santos
> >


  • Prev by Date: something is "wrong" with the Fills option in InequalityGraphics
  • Next by Date: Re: Strange empty set of solutions
  • Previous by thread: Re: Real numerical computations
  • Next by thread: Please help carry out the integral