MathGroup Archive 2011

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

Search the Archive

Re: Delete elements from list..

  • To: mathgroup at smc.vnet.net
  • Subject: [mg116773] Re: Delete elements from list..
  • From: "Maarten van der Burgt" <Maarten.VanDerBurgt at kla-tencor.com>
  • Date: Sun, 27 Feb 2011 04:34:31 -0500 (EST)

Ray,

You are correct:  my rule would reduce {1, 9, 2, 3, 4, 5, 6, 7, 8} to
{1, 9}.
But as the y_i data are generally strictly increasing with only a few %
or less of the points where the y_i make a dip, this is not a problem.

Maarten

-----Original Message-----
From: Ray Koopman [mailto:koopman at sfu.ca]
Sent: Friday, 25 February, 2011 12:37
To: mathgroup at smc.vnet.net
Subject: [mg116773] [mg116732] Re: Delete elements from list..

On Feb 24, 3:29 am, "Maarten van der Burgt" <Maarten.VanDerBu...@kla-
tencor.com> wrote:
> Hallo,
>
> Thanks everybody who replied to my questions.
>
> The real problem I have is just a bit more complex than my
> simplified example. My list is in fact a numerical 2D list like
> mylist1 ==== {{x_0, y_0}, {x_1,y_1},... {x_i, y_i}, ...{x_N, y_N}}.
> The xi are strictly increasing and the yi should be as well.
> Due to some measurement errors it can happen that this is not the
> case. I simply want to delete the {xi, yi} pairs where
> y_i <==== y_i-1. That way I end up with a list, mylist2,
> where also the y_i are strictly increasing.
> (that way I can make an Interpolation[Reverse/@mylist2]
> in order to have a function x_i(y_i)).
>
> I have not had the time to study your answers in this view,
> but from a first look and the variety of the answers it seems
> that there is definitely something which should help.
>
> Thanks for your help.
>
> Maarten

Your rule would you reduce {1, 9, 2, 3, 4, 5, 6, 7, 8} to {1, 9}
which I don't think you would want to do.

Wouldn't it make more sense to delete only the 9?

(If we work from right to left instead of left to right,
deleting the current y if it's >== min[all previous kept y_i],
we do delete only the 9.)

Shouldn't the question be more like "What is the smallest set of
points that must be deleted to make y monotone increasing in x?"

Or, considering that all the y_i may contain error, you could find
the vector z that is closest to y (in some sense that depends on the
assumed nature of the errors) and is also monotone increasing in x,
and then do inverse interpolation.


  • Prev by Date: Numerical Convolution
  • Next by Date: Re: outputting C code
  • Previous by thread: Re: Delete elements from list..
  • Next by thread: Re: Delete elements from list..