| Re: Interpolation to conserve integrated flux [message #71164 is a reply to message #71162] |
Sun, 30 May 2010 05:03   |
jkj
Messages: 48
Registered: April 2007
|
Member |
|
|
On May 29, 8:13 pm, John Shaw <jds...@udel.edu> wrote:
> Hi all,
>
> I am hoping someone out there can help me with this.
>
> I am looking to combine spectra such that the integrated flux between
> any arbitrary set of points is conserved.
>
> Consider a set of two equal length float arrays, one containing
> wavelength values (call this x1, x2, x3, ...) and one containing flux
> values (call these y1, y2, y3,...). So that y1 goes with x1 and y2
> goes with x2, and so on.
>
> Unfortunately, the wavelength arrays (x1, x2, x3, ...) have similar
> values but not the same, e.g.:
>
> x1 = [3212.7, 3215.1, 3217.5, 3219.9, ...]
> x2 = [3213.1, 3215.4, 3217.7, 3220.0, ...]
>
> What I want to do is combine x1 and x2 to x_comb and y1 and y2 to
> y_comb such that if one were to sum between two wavelengths the in the
> new arrays (x_comb and y_comb) you would get the same as doing them in
> the original (x1,y1 and x2,y2) and adding the results. That is, if I
> need to sum up values in y when x is between 3214.0 and 3219.0. For
> x1, I would linearly interpolate the values in y1 for the two
> wavelengths and sum under the curve. I then do the same for x2 and
> y2. I can already do this - that part is simple. What I want to do
> is have new arrays x_comb and y_comb such that I can do the same sum
> across such that the same (summed region in y1) + (summed region in
> y2) = (summed region in y_comb).
>
> Is there an algorithm that anyone knows that will do this?
>
> Or, as an intermediate step, is there a way interpolate the values in
> x2 to x1, and y2 to an new set of values (y2_new) that keeps the sums
> of y2 and y2_new the same between any two arbitray points (as
> described above)?
>
> Thanks - John.
You should use the spline function to translate one spectrum to the
same x-values as the other. We use the spline function with electron
spectra quite often. The simplest thing to do is use the INTERPOL
function with SPLINE set.
y_comb = interpol(y_orig, x_orig, x_other, /spline)
where "x_orig" and "y_orig" are the values of one spectrum or the
other and "x_other" are the wavelengths of the other spectrum, the
wavelengths that you wish the original spectrum was defined in so that
you could directly compare their integrated values. Simply overplot
the "orig" and "comb" [x_comb, of course, is the same as x_other] sets
of spectrum to convince yourself that the spline function did a good
job of translating from one to the other.
-Kevin
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
