How can we make the best use of non-detections and noisy data when we don't believe the uncertainty or upper limit informaiton being reported by the data provider?

• Nat Butler (ASU)
• Dan Foreman-Mackey (NYU)
• David W. Hogg (NYU)
• James Long (Berkeley)
• Joey Richards (Berkeley)

Copyright 2011, 2012 the authors. All rights reserved.

• mira_sims: light curves for simulated mira variables. LCs are of ASAS 235627-4947.2 with varying fractions of the original flux, e.g., 1, 0.5, 0.25, 0.01

• miras: ASAS light curves for the 1720 mira candidates, with non-detections denoted with 29.999

• mira_features.dat: features for 1720 mira candidates, including all Lomb-Scargle and non-LS features

• rrl: ASAS light curves for the 1029 RR Lyrae, FM candidates, with non-detections denoted with 29.999

• rrl_features.dat: features for 1029 RR Lyrae, FM candidates, including all Lomb-Scargle and non-LS features.

mira_simulated.pdf: simulated mira light curve with different flux fractions

the code

• Makefile: file to compile Hogg's original note

• censored_catalog.tex: tex for Hogg's original note

• ms.tex: manuscript tex file

• non_detect.bib: manuscript bib file

• apj.bst: needed for ms.tex to compile