fwhigh / stellar-locus-regression Goto Github PK

Steps that reproduce the problem:
1. Run SLR on a new colortable.  A .sav file is created based on the ascii
table.
2. Run again, and SLR goes faster (if force is not set).  This is because
the .sav file was read instead of the ascii table.
3. Now you've remade the colortable in your upstream analysis, and so you
overwrite the old one.
4. Run SLR again, and it uses the old .sav file.  It doesn't create a new
one automatically!

We should be making new .sav files every time a new ascii colortable for a
given field is recreated.  This probably involves checking the .sav file
timestamp against the ascii colortable timestamp.  I haven't yet figured
out how to do this in a platform-independent way, after initially searching
through the native IDL routines.

Steps that reproduce the problem
1. Run the slr.csh example in the install instructions.
2. Check out the output colortable

The new calibrated colors appear in the last few columns, but the column
headings in the first line are not aligned with the data.

Shouldn't be hard to fix in IDL using format='(A10)' or whatever.

easy fix

Example usage of slr_get_data where this matters:
1. Call slr_get_data on a huge data set
2. Call it again, and you get the data faster

This is great.  However, right now the hard limits (eg, signal-to-noise
threshold) are placed on the data *before* the savefile is saved, so that
the second time you call it it won't give data with *new* limits if the
limits are changed in the meantime.

What steps will reproduce the problem?
1. Add a blank line randomly in a config file
2. Try to run SLR


What is the expected output? What do you see instead?
It should run properly, but it complains, probably because I'm not throwing
out whitespace like I should.

What steps will reproduce the problem?
1. You don't have the sfd maps (which should be fine)
2. You run the SLR demos out of the box

slr_get_data_array complains that there is no galactic extinction tag in
the data structure.

It is supposed to be ok if there is no galactic extinction tag in the
structure, as long as you don't want to deredden your colors.  This was
never actually tested, but it will now be fixed

Version 1.0 of SLR depends on IDL analyst, but the only routines it uses are
1. imsl_lnormregress
2. imsl_random
3. imsl_inv

These can all be replaced easily.  Considering that Analyst is an extra
couple hundred bucks a pop, we should.

[deleted issue]

What steps will reproduce the problem?
1. set postscript and plot to 1 in the config file
2. run SLR

What is the expected output? What do you see instead?
Should produce EPS files, but this hasn't been tested on the public code.

What steps will reproduce the problem?
1. Compare color errors to quadrature sum of instrumental mag statistical
errors


What is the expected output? What do you see instead?
Want them to be equal.  It is possible to add the SLR color calibration
errors to these in the catalog, but because this is systematic and
constant, it should be kept separate.