Re: snminuit fits to subaru data.

From: Don Groom (deg@panisse.lbl.gov)
Date: Thu Dec 12 2002 - 18:00:59 PST

  • Next message: Lifan Wang: "Re: snminuit fits to subaru data."

    Gerson,

    Thanks for posting it. People can see the entire show at
    /home/sierra1/deg/idl/SCPfits02/Output

    A couple of observations:

    1. As we all know, it's basically incorrect to start with magnitudes. We
    should get data in the observed form, counts. In particular, errors are
    symmetric in counts (which for all practical purposes are Gaussian), and
    except in the limiting case of very small errors are not even symmetric in
    magnitude. On top of this, it is the case that for references the number
    of counts is negative as often as positive, if sky has been subtracted
    properly. We occasionally see really biased results when people discard
    the negative results and fit in magnitudes, as happened e.g. in recent
    papers on SNe risetime.

    2. In general, the chi^2's are about a factor of 4 too high, suggesting
    that the errors are underestimated by about a factor of two.

    3. I'll repeat this more carefully in the AM. In particular, the choices
    about the appropriate kcorr and whether to fix stretch need revisited and
    checked. Some of the low-stretch guys might be artifacts because of this.

    4. There are still a couple of program bugs in the graphics display, most
    importantly that the reported kcorr on line 3 comes out as e.g.
    splinefit_b_i even if another kcorr was in the command file. Mana\~na.
    The chi^2 reporting is garbled but correct.

    5. We need better K corrections. I winged it, and made I --> B for the
    high-z objects and I -> V for the lower redshift objects. In the absence
    of good R or I templates not much else could be done. BUT: the I --> B etc
    kcorr's were generated from l.s. spline fits to real data. Peter, did you
    artifically cut the b_i data at z = 0.500(-0.002)0.998 at the top end? (I
    used the highest-z kcorr for z ge 1.000.) Could we slightly extend it by
    the same method? (I mean with the individual sn data points, not the
    ueberspectrum.)

    6. I'm really, really impressed with the quality of the SUBARU data. It
    looked like HST results at first.

    7. Gerson's lightcurve files could be run with no further screwing with
    format. Great!

    --Don

     Thu, 12 Dec 2002, Gerson Goldhaber wrote:

    > Hi all,
    > I have taken the table from Naoki of Dec 10,and brought it into the
    > format needed for snminuit fits.
    > For the z > 1. SNe the k-corrections are not correct. I have left out
    > points with negative magnitude. When all 7 data points are included the
    > first one is Nov 3 and the last one Dec 8. Time is in Observer days.
    > This was a test of Don's new program, run with Don's help.
    > I think we have too many stretch < 1 cases. So remember this is crude. I
    > am enclosing 2 cases as sample. The rest will be on the Web shortly.
    > enjoy Gerson
    >

    |-+|-+|-+|-+|-+|-+|-+|-+|-+|-+|-+|-+|-+|-+|-+|-+|-+|-+|-+|-+|-+|-+|-+|-+|
    Don Groom (Particle Data Group, Supernova Cosmology Project)
    DEGroom(at)lbl.gov www-ccd.lbl.gov Voice: 510/486-6788 FAX: 510/486-4799
    Analog: 50-308//Berkeley Lab//Berkeley, CA 94720



    This archive was generated by hypermail 2.1.4 : Thu Dec 12 2002 - 18:01:16 PST