On May 30, 4:02 pm, Lewin Boehnke <n...@[EMAIL PROTECTED]
> wrote:
> Tim wrote:
> > On May 28, 6:53 pm, goog...@[EMAIL PROTECTED]
wrote:
> >> Tim wrote:
> >> > Hi All,
>
> >> > I've looked in the do***entation and online, and I can't seem to
find
> >> > any way to do Gaussian smoothing.
>
> >> > Normally, I'd simply write a function for it, but the problem is
that
> >> > this type of smoothing of course requires as many dimensions as
there
> >> > are data points. That is, each point of data is scaled as a
function
> >> > of the height of all other points. Since (I think) GNUPLOT reads
one
> >> > line at a time from data files, I don't know if it's possible.
>
> >> > That said, GNUPLOT is so powerful, I can't see it not being able to
do
> >> > something like this. Could somebody please point me in the right
> >> > direction?
>
> >> > Cheers,
>
> >> > Tim.
>
> >> Gnuplot is primarily a plotting tool . This is probably the sort of
> >> data processing task you could do externally , either before calling
> >> gnuplot of by using gnuplot's system() command.
>
> >> Having said that someone just posted a patch to do something like
this
> >> to gnuplot-beta mailing list. You may like to get current cvs
> >> gnuplot , add that patch and test it.
>
> >> BTW if you want to write some code you can do gaussian smoothing by
> >> applying a finite kernel filter. In practice this is just a few
values
> >> depending on the width of your gaussian filter, not the all your data
> >> points.
>
> > I don't really want to keep a lot of extra numbers around. These files
> > will be somewhat large and numerous, so it could become confusing very
> > fast.
>
> > If I write a program which outputs the smoothed data to stdout, could
> > I somehow run it within gnuplot and directly read the output? Not that
> > it's a big issue to write to a tem****ary file, but I would somehow
> > like something cleaner.
>
> As I was teached in this newsgroup just a few weeks ago, "plot '<
myscript'"
> does just that. So "plot '< smooth mydata.dat'" is probably what you
want.
>
> Lewin
Thanks, that sounds great. Exactly what I need.
Cheers,
Tim.
|
