Home »
Public Forums »
archive »
Re: Searching for fast linear interpolation routine
| Re: Searching for fast linear interpolation routine [message #8551] |
Fri, 04 April 1997 00:00 |
Wayne Landsman
Messages: 117
Registered: January 1997
|
Senior Member |
|
|
Roger J. Dejus wrote:
>
> Did someone write a fast linear interpolation routine for irregular one
> dimensional arrays (monotonically ascending or descending abscissas)
> similar in functionality to the INTERPOL.PRO routine from RSI?
>
The INTERPOL function consists of two steps: (1) finding the effective
index of the interpolation value (e.g. it is located between indicies 12
and 13 in the abscissa), and (2) performing the interpolation. I know
of at least 3 ways in which the speed of the INTERPOL function can be
improved:
(1) INTERPOL uses a incremental search algorithm to find the
effective index, whereas the quickest way to search a monotonic array is
a binary search (divide and conquer).
(2) the effective index search is not vectorized
(3) the intrinsic INTERPOLATE function (available since V2.2) is not
used to do the interpolation
The program LINTERP in the IDL Astronomy Library incorporates these
three improvements and I get a factor of four improvement in speed on my
Ultra-2.
http://idlastro.gsfc.nasa.gov/ftp/pro/math/linterp.pro
LINTERP calls the program TABINV to find the effective index
http://idlastro.gsfc.nasa.gov/ftp/pro/math/tabinv.pro
These procedures also call ISARRAY (from the JHUAPL library) and
ZPARCHECK
http://idlastro.gsfc.nasa.gov/ftp/pro/jhuapl/isarray.pro
http://idlastro.gsfc.nasa.gov/ftp/pro/misc/zparcheck.pro
|
|
|
|
Current Time: Wed Oct 08 15:22:37 PDT 2025
Total time taken to generate the page: 0.00431 seconds
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
