next up previous contents
Next: Database Up: Data Reduction Previous: Installing this software   Contents

Pitfalls and features

rctred:
  1. Rctred uses quite a bit of memory and this is possibly too much for some computers. This problem might kill the program before it starts, in the middle of working, or simply slow the computer to a crawl if it uses swap memory.
  2. Rctred may fail while reducing certain images with "funny" properties. For instance, it is set up to accept images of up to a certain size (currently 2300x2300 pixels). Exceeding this size limit (e.g., by choosing an overscan region that is too large) will cause problems.
  3. Rctred reads in image headers containing information about the way the image was read out. If you monkey around with these image header values so that they no longer make sense, it will cause problems.
  4. Reducing a one-dimensional image, or even one with very few rows, might cause problems.
  5. Rctred expects the images it is reducing to have the same size as any calibration images applied once they reach the particular step in the process. The user must make sure that this is the case.
  6. Rctred behaves in an unexpected manner in this circumstance: if too many pixels are rejected from a row of the overscan region such that no pixels in that row are left to be averaged, then the pixel rejection gets ignored and all pixels in that row are counted in the calculation of the overscan bias.
  7. Rctred behaves in an unexpected manner in this circumstance: if the overscan region pixel rejection is of type 'sigclip', but there are fewer than 3 columns in the overscan region, no pixel rejection is performed.
  8. One limitation of rctred is that currently the program attempts to use all columns in the overscan region to calculate the overscan bias (disregarding whichever pixels get rejected by any rejection algorithm used). Perhaps it will be necessary in the future to allow the user to specify that only certain overscan columns be used, but note that since this CCD has the option of the user specifying both the width of the overscan region and an "offset" (which I think means there can be overscan pixels readout between the image pixels and the overscan pixels that actually get written to the image), the present situation may work well and be much simpler.
rctcom:
  1. Rctcom will quit with an error if there are blank lines in the list of input filenames.
  2. Rctcom expects the images it is combining to have the same size. The user must make sure that this is the case.
  3. The fortran version of rctcom may fail if you attempt to reject too many pixels given the number of images being combined.
rctnight:
  1. Rctnight uses both rctred and rctcom directly and so is subject to some of the same problems: images that are too large for the image buffer or computer RAM, that have the wrong dimension, that are too small(?), that have too few overscan columns, or that have nonsense image header values will most likely cause problems.
  2. The input parameters to rctnight will be expanded (by the shell?) if they look like filename templates, but are NOT enclosed in single quotes. This will probably cause problems. This argument expansion is the reason that template-type arguments must be enclosed in quotes. If you don't have any files or directories that have a '?' or '*' character in their names (why would you?), there shouldn't be any problem distinguishing between a specific file/directory name and a template and no reason to experience this problem.

next up previous contents
Next: Database Up: Data Reduction Previous: Installing this software   Contents
Louis-Gregory Strolger 2012-01-09