search

LSSTDESC / Monitor

Extract light curves for time-variable cosmological objects
BSD 3-Clause "New" or "Revised" License
3 stars 1 forks source link

Light curve quality metric #9

Open drphilmarshall opened 8 years ago

drphilmarshall commented 8 years ago

Reissuing from https://github.com/DarkEnergyScienceCollaboration/Twinkles/issues/143 :

I suggest we use something like the rms difference between observed and true light curves, weighted by observation errors, to make a $\chi^2$-like quantity. Then, we can compute the number of sigma offset between the observation and the truth using the Gaussian (Fisher) approximation to the chi-squared distribution, to get $N_{\sigma} = \sqrt(2\chi^2) - \sqrt(2N_d)$ where $N_d$ is the number of datapoints.

we.compare_with_truth(from='reference.fits')
we.compute_lightcurve_quality()
print("Observed light curve is",we.quality,"sigma from truth.")
drphilmarshall commented 8 years ago

Before we get to the observed vs truth metrics, we should code up some simpler ones. N5 (the number of light curve points with SNR > 5) could be one. How about we make a validate_lightcurve method in the LightCurve class, and have it add metric values to a dictionary called self.metrics?