Re: K-Correction histogram

From: Lifan Wang (lifan@panisse.lbl.gov)
Date: Thu Apr 01 2004 - 16:11:44 PST

  • Next message: Lifan Wang: "Re: K-Correction histogram"

    On Thu, 1 Apr 2004, Robert A. Knop Jr. wrote:

    > On Thu, Apr 01, 2004 at 03:29:09PM -0800, Lifan Wang wrote:
    > > I might have found the reason for the difference between my K-corr and
    > > the Serena's. In my fit, I only normalize the spectra to the observed
    > > magnitudes close to the region of interest, and use a smooth function for
    > > the normalization. When I try to include all the colors, I do end up
    > > something closer to what I can get from Serena's template, but I think
    > > this is unphysical.
    >
    > ???
    >
    > How can that be unphysical?
    >
    > What's happening is that your smooth function is unanchored when you
    > don't use the other colors, and you're getting a different continuum
    > from what Serena is getting. I don't see how that could be unphysical;
    > it simply sounds like a matter of whether or not you're consistent with
    > the observed data.
    >
      Rob, by unphysical, I mean there should be no wiggles in the
    normalization function. Spline, by definition, might be too none-linear
    with too many ups and downs.
    > > Note that contrary to what's in Rob's email, that in my approach, I
    did
    > > not "applying the function that pushes it a little up here, a little down
    > > there, in comparison to what Serena did".
    > >
    > > The normalization function I use has NO wiggles, and as it seems, we all
    > > agree that this is important.
    >
    > That's not what I meant. *ANY* normalization that isn't a single
    > constant is going to push the spectrum up a little in one place, down a
    > little in another. This isn't wiggles, this is just what you have to do
    > if you're normalizing more than one color.
    >
      Put it in a mathematical term, by no wiggles I mean the second
    order derivatives should be positive or negative in all places.
    > It's probably possible that there are degeneracies-- that two different
    > normalization functions could yield the same photometry. This is
    > another reason to have as many photometry poitns as possible. (Indeed,
    > we'd be better off if we had Z and J photometry to anchor the spectrum
    > to the red of the I-band.)
    >
    > -Rob
    >
    >

    -- 
    Lifan Wang           (510) 495 2733 (o)   (510) 704 0119 (h)
    


    This archive was generated by hypermail 2.1.4 : Thu Apr 01 2004 - 16:11:45 PST