MathGroup Archive 2008

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

Search the Archive

Re: Method Option

  • To: mathgroup at smc.vnet.net
  • Subject: [mg90931] Re: Method Option
  • From: mark mcclure <mcmcclur at unca.edu>
  • Date: Wed, 30 Jul 2008 03:52:24 -0400 (EDT)
  • References: <g6h5dc$h0u$1@smc.vnet.net> <g6matp$i1k$1@smc.vnet.net>

On Jul 29, 6:11 am, Peter Breitfeld <ph... at t-online.de> wrote:
> this is a good first try to find the possible methods, but not
> exhausting.
> I posted this message to get more insight when reading a post from
> Jens Peer Kuska who put a method option to ContourPlot like this:
> plt1 = ContourPlot[Sin[x*y], {x, -Pi, Pi}, {y, -Pi, Pi},
>    MaxRecursion -> 0, PlotPoints -> 64,
>    Method -> {"Refinement" -> {"CellDecomposition" -> "Quad"}}]
>
> So I believe, Methods to be defined as strings are even harder to find.

Interesting.  On the other hand:

In[1]:= Eigenvalues[RandomReal[{0,1},{20,20}], 1, Method -> Bogus]
Eigenvalues::emeth:
   The method specified by Method -> Bogus
     should be either Automatic or Arnoldi.

Note, however, that Arnoldi is not a built in object; Method-
>"Arnoldi"
is the appropriate way to call it.  Furthermore, there area quite a few
other possible Method options for Eigenvalues, as documented in
the linear algebra documentation.  The trick I mentioned, incidentally,
comes from one of Michael Trott's books.  So, I'm beginning to
wonder if the internal handling of Bogus options has been relaxed.

Mark McClure


  • Prev by Date: SEM: New third-party Mathematica application for supercomputing
  • Next by Date: Re: remove higher orders terms from mathemtica's result
  • Previous by thread: Re: Method Option
  • Next by thread: ParametricPlot precision problem