| Re: polar interpolation [message #33671 is a reply to message #33625] |
Sun, 19 January 2003 19:33   |
Thomas Gutzler
Messages: 44
Registered: November 2002
|
Member |
|
|
Hi,
Rick Towler wrote:
>
> Interpolate between points a and b:
>
> iFactor = (Ti - Ta) / (Tb - Ta)
> Ri = Ra + (iFactor * (Rb - Ra))
> Where Ti is the Theta value where you are interpolating your radius Ri.
I stole this from interpol.pro. The function is the same, just other
names for variables:
; ,----------------------------------------------------------- -----
; | Purpose:
; | linear, polar interpolation for irregular grids
; | T: The theta values (in degrees) for R (radius). This vector
; | must have same # of elements as R. The values MUST be
; | "monotonically ascending" or "descending" (see problem below).
; | U: The theta values for the result. The result will have
; | the same number of elements as U. U does not need to be
; | monotonic. If U is outside the range of T, then the
; | closest two endpoints of (T,R) are linearly extrapolated.
; `----------------------------------------------------------- ------
FUNCTION polar_interpol, R, T, U
m = N_elements(R) ; # of input pnts
s = VALUE_LOCATE(T, U) > 0L < (m-2) ; Subscript intervals.
p = ( U - T[s] ) * ( R[s+1] - R[s] ) / ( T[s+1] - T[s] ) + R[s]
RETURN, p
END
But what to do, if an array like this wants to be interpolated:
R: [1,2,3]
T: [340,350,20]
U: [355,5,15,25]
I can see 2 solutions at the moment:
1) - find the first index (i) after crossing the magic 0-line.
In this case: i_T = 2 and i_U = 1
- T[i_T:*] = T[i_T:*] + 360
- U[i_U:*] = U[i_U:*] + 360
- interpolate T: [340,350,380], U: [355,365,375,385]
- RESULT = [2.16667, 2.50000, 2.83333, 3.16667]
2) - split the vectors
R1: [1,2,2.33333] (last value must be interpolated)
T1: [340,350,360]
U1: [355]
R2: [2.33333,3]
T2: [0,20]
U2: [5,15,25]
- interpolate separately
RESULT1: [2.16667]
RESULT2: [2.50000, 2.83333, 3.16667]
- return [RESULT1, RESULT2]
Is there a trick to do this easier, using some strange functions
(VALUE_LOCATE was strange for me) ?
Keep in mind that the next step would be a quadratic interpolation.
thanks, Tom
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
