[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: bogus values in photometric solution ... solved
Michael and all,
This is great news. I look forward to getting the updated code. I am
eager to process a bunch of data so that I can look at things like long
term drift.
The present data seems to have a floor around 0.015 mag. However, when you
look at the data there it would only be about 0.005 mag if systematic
trends could somehow be removed. Part of this is due to changes in the
color terms over the two hour run. I want to study this, but there is not
much sense in doing it until the new version of photom is available. It
just might go away.
Looking at papers like "Study in IR of stars in Chamaeleon" that Michael
referenced, we are at about the same accuracy. So I think the engineering
run data is worth working up for a publication. I am ready to do that.
Tom Droege
At 09:03 AM 4/30/02 -0400, you wrote:
> Tom Droege and Rich Knowles pointed out that the photometric
>calibration portion of the pipeline would sometimes fail.
>I've finally tracked down the source of the error: it is the
>library of numerical routines, called "CCMATH", which I had been
>using to solve a big least-squares equation.
>
> Today, I replaced the CCMATH routines with functions from
>another library of numerical routines, the "GNU Scientific
>Library". The program runs without error and yields a good
>solution for all images on the night 020206; with the CCMATH
>routines, it would give bogus results for the entire night,
>and for certain subsets of the them.
>
> I need to produce a new version of the "photom" program,
>which calls the GSL routines. It will take me some time to
>write and document the new version. At the same time, I'll
>fix the small error in the matching routines noted earlier,
>and create a larger subset of the Tycho data for use in
>matching. It will probably take me a couple of weeks.
>
> Thanks to all who helped in describing this problem.
>
> Michael Richmond