| Re: More Kriging Problems [message #86198 is a reply to message #86197] |
Thu, 17 October 2013 06:24  |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
David Fanning writes:
> There are a couple of other differences (which may explain the time
> difference). cgKrig2D preserves the grid size of the original data set,
> while the original Krig2D code and the Krig2D_Fast code default to their
> regular 26x26 size grid. (Humm, if this is true, then if I make the grid
> 31x31 then cgKrig2D should be slower than Krig2D_Fast, and this in fact
> happens.)
>
> Elapsed Time: 5.616000 ; Original Krig2D code
> Elapsed Time: 2.137000 ; cgKrig2D
> Elapsed Time: 2.106000 ; Krig2D_Fast
>
> Also, now the original Krig2D routine produces a CRAZY surface! Not at
> all like the original surface. This wasn't happening last night, but you
> know how things get when it is late in the day and you have been working
> 12 hours straight trying to make sense of things. All kinds of
> strangeness sets in. Anyway, based on today's tests, I wouldn't use the
> original Krig2D routine.
Well, HERE is an interesting thing!
Just for the hell of it, I changed the number in my test program to 26,
to make a 26x26 array. I wanted to see if cgKrig2D and Krig2D_Fast would
be exactly the same speed in that case. In fact, they are, pretty much:
Elapsed Time: 2.482000 ; Original Krig2D code
Elapsed Time: 0.708000 ; cgKrig2D
Elapsed Time: 0.710000 ; Krig2D_Fast
But, the interesting thing here is that NOW the surface plot of the
original Krig2D plot is NORMAL!!!! Not crazy at all. Apparently, this
code only goes crazy if there is a mismatch between the input data size
and the output data size. Who woulda known! ;-)
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.idlcoyote.com/
Sepore ma de ni thue. ("Perhaps thou speakest truth.")
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
