MathGroup Archive 2009

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

Search the Archive

Re: LogLinearPlot strange "features"

  • To: mathgroup at smc.vnet.net
  • Subject: [mg101424] Re: LogLinearPlot strange "features"
  • From: Fred Bartoli <""@news.free.fr>
  • Date: Tue, 7 Jul 2009 05:04:55 -0400 (EDT)
  • References: <16474964.1246447235743.JavaMail.root@n11> <h2pplb$qqr$1@smc.vnet.net>
  • Reply-to: myname_with_a_dot_inbetween at free.fr

David Park a =E9crit :
> You essentially found the answer. The separately plotted Points have to be
> converted to the underlying log plot coordinates. (And these are not the
> same as the labeled values.)
>
>> "It'd be nice to have all the graphics primitives in the same coordinate,
>> wouldn't it?"
>
> Yes. Here is one way to work from the same basic data, in fact the same
> basic plot primitives, for all four plot types. I'm using the Presentations
> package DrawingTransform to transform a set of pre-computed graphics
> primitives to the various log plot types. The one down side of this method
> is that we have to use CustomTicks and CustomGridLines to specify the proper
> ticks and grids. However, especially with the Grids, the Mathematica choice
> is quite poor anyway. Besides a poor choice of values, they make the Grid
> lines way too dark. Grid lines should be barely visible so they don't stomp
> all over the data. (The following plots don't generate any "out of domain"
> messages, but you could just use Quiet on your plots anyway. If Mathematica
> can normally ignore non-real plot points, why can't they also ignore points
> outside an InterpolatingFunction domain?)
>
<>
>> I also keep getting that message:
>> InterpolatingFunction::dmval: Input value {-0.634829} lies outside the
>> range of data in the interpolating function. Extrapolation will be used.
>>
>> which I guess is plain wrong since
>> Exp[-0.6348286996935327`] = 0.5300262742047284`
>>

Thanks David. Unfortunately I don't have the Presentations package.

My point about the out of domain message is that it seems to be a bug,
because the values given back by mathematica *are* in the interpolating
function domain.
Sure I can Quiet it (and this is what I do).


--
Thanks,
Fred.


  • Prev by Date: Re: including specific values in parametric plots
  • Next by Date: Re: Fast calculation of pair correlation function
  • Previous by thread: Re: LogLinearPlot strange "features"
  • Next by thread: Re: LogLinearPlot strange "features"