Current Data Information

From NYU CCPP Wiki

PRIMUS Data Reduction Progress / Information:

In an effort to keep everyone working with the most up to date version of the reductions, this page will serve as a clearing house of information on the current versions of the reductions and provide some documentation for each rerun number.

In general, reductions of the two-dimensional data - extracting the raw frames to 1d spectra for each objects - are numbered sequentially. In general, each new version is supplemented with a tagged version of primus. Fitting the one-dimensional spectra to obtain redshifts - the 1d pipeline - is often run several time per 2d reduction as various changes are implemented and tested. In general, although there are some exceptions already, 1d reductions are number sequentially with a four digit number TTXX where TT is the two digit number indicating the 2d version and XX is the counter for the 1d version within a given 2d reduction.

Bold indicate the current version.

Two-Dimensional Reductions

0015 - Extracted with new halo correction and a subhalo correction. Only test plates have been run.

0016 - Rerun based off of rerun 0015 with some test revisions. Some of the changes were rolled back from this version, so this rerun is not our primary rerun.

0011 - Extractions including a new halo model which includes a dipole-like component to the halo kernel and thus mimics the effects of a wavelength dependent halo kernel. Also, the strength of the scattering halo is allowed to vary with position on the focal plane. This version should treat all data properly in the sense that data from 2007 (with the objects in the inner traces) and 2008 (objects in the outer traces) should all extract and calibrate properly. No changes in the optical model assumed for our wavelengths has been made (all of the reductions use the same model). This is something that likely should be tested. Currently, only the ~30 masked used for tests of the code have been run through this code

0010 - A test reduction which extends the tracelength to 200 pixels and only extracts objects targeted as blue standards. This results in many slit collisions during extraction and thus this reduction should only be used after contacting Richard to be sure proper precautions are used.

0009 - An update to rerun 0008 (i.e. none of the spectra were re-extracted the output from rerun 0008 was modified to create a new rerun). This rerun included a shift to the wavelengths of each object based on an offset and rotation determined by measuring the offset in the spatial direction of the slit. Since offsets in the spectral direction are only shown when you look for offsets in the spectral features of a understood populations, this offset can be systematically offset in the spectra direction.

0008 - The first full extraction of PRIMUS data from pre-2008 including a full halo subtraction, optimal extraction.

Redshift Fitting Reruns

Options involved in fitting :

1510 - 1d rerun that is currently being generated for data from 2d rerun 15. This is the first 1d run with all of the photometry included. Like the 2d rerun, this only includes test plates. This was run with the new reweighting (cutting the blue at 4500 and then downweighting pixels in the red). The grid used was based upon the "ages_basis_080905_1.0.fits.gz" file and was created using a set of basis vectors with >1 chi-squared between them.

1100 - Base reductions for 2d rerun 0011. Options : /grid, /constpsf, /chooseslit. Note: Previously reweight was used when fitting but tests of fakedata show this leads to many problems at inferred redshifts of <0.2. This should be looked into and possibly modified (possibly so only the red is downweighted but the blue is preserved).

0850 - Base reduction for rerun 0009 (note data model error). Options : /grid, /constpsf, /chooseslit, /reweight

Both rerun 1100 and 0850 use a tweaked version of the young template with a hand-tuned about of Oii emission added to the continuum template. No full rerun has been preformed since 0850 which does not include this change. Note that reruns prior to 0850 are antiquated and should not be used for comparisons.

grid - use a precomputed grid of models rather than running template2primus for each model for each slit. This saves time, but may introduce its own errors so it will be tracked separately.

reweight - downweight the blue and red portions of the spectrum when computing chi2 for each model. This was originally instated to try and distinguish more robustly between stars and galaxies (small errors in the red would lead to erros when choosing the best template).

constpsf - Dont try and account for wavlength dependant object and instrumental PSF size. This is used implicitly when you use /grid.

chooseslit - If either A or B is flagged as bad, then only use the other.z