Home »
Public Forums »
archive »
Re: range of Z in surface ...
| Re: range of Z in surface ... [message #4791] |
Tue, 25 July 1995 00:00 |
Serdar Manizade
Messages: 15
Registered: December 1994
|
Junior Member |
|
|
surinder@eng.umd.edu (Surinder P. Singh) wrote:
> I am having this problem again and again
> with idl that in the surface command if
> Z variable is below 10^-60 it gives underflow and
> plots it as zeroes, eventhough the Z variable is
> a double array.
>
> I thought the double array should be comfortable
> with ranges of 10^-200 to 10^200 or so.
>
> Why am i getting these annoying errors ?
> Surinder
I suspect (but do not know for sure) that many IDL routines carry
out intermediate calculations using floating point precision. I have
run across this in a number of contexts, here are two examples:
1) Plotting long integer data in 2-D plot when the magnitude of the
coordinates are much larger than the range of the axis-- the data
acquires a granularity whose increment size is far larger than the
precision of the long integer data.
2) Creating the interpolation grid in TRIGRID using double precision
data-- for a given grid spacing (the GS keyword), the number of elements
in the output (interpolated) array will change as the magnitudes of
the pertinent inputs are scaled. A few calculations suggest that this
is due to the precision and resultant rounding of intermediate calculations.
Both of the above cases point to a reduction of precision internal to
each of the routines. I suspect that the effect you are seeing has
a related cause. I do not have a fix, aside from writing my own routines
to replace the ones supplied by RSI. I can reinvent TRIGRID, but I
can't get around the graphic interface that IDL is using. Perhaps
somebody does know a way, but then what would be the point of using
IDL in the first place?
I empathize with your annoyance. I regret the only solution I can
suggest is to scale your data to accommodate IDL's limitations. IDL
ought to accommodate your data instead.
Anyone else have insight into the problem of getting the desired
precision from IDL routines?
Serdar Manizade
EG&G/ NASA Wallops Flight Facility, Virginia
|
|
|
|
Current Time: Sat Oct 11 10:26:34 PDT 2025
Total time taken to generate the page: 1.60071 seconds
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
