These three light curve files are for the three millimagnitude RR 
Lyrae variables (mmRR) described in [insert reference when it exists]. They 
are released under the Creative Commons Attribution 4.0 License
(please see https://creativecommons.org/licenses/by/4.0/ for details).  Part
of the license requirements is attribution for any use of the licensed data;
any use of this data should reference the description paper, with citation
details found at [insert when it exists].

The files are CSV files, with a header on the first row and the data on the 
remaining rows.  Descriptions of the five columns can be found below:


"BJD-2454833.0"   - Barycentric Julian date, minus 2454833.0 as is standard
                    for Kepler/K2 data.

"raw"             - The "raw" light curve data, expressed in Kepler magnitudes
                    (Kp), as extracted from the K2 images, without any 
                    additional corrections.

"post_decorr"     - The light curve data, in Kp, after undergoing the roll 
                    decorrelation, as described in the paper referenced above.

"post_tfa"        - The light curve data, in Kp, after undergoing the roll
                    decorrelation and subsequently the trend filtering 
                    algorithm. These are the final light curves points that 
                    were used in our analysis and represent the most "cleaned 
                    up" light curves we generated.

"err"             - Photometric error bar size, in Kp, as calculated by the 
                    code used to generate the raw light curve data (see paper
                    for details).  In our experience across all the light
                    curves we generated, the error values were usually 
                    overestimated.

*** Important note: the first 49 rows of data for the "post_decorr" and
    "post_tfa" columns are empty.  These first observations had a very
    different telescope pointing than the rest of the observations and so
    were unable to be effectively cleaned up by the roll decorrelation, 
    and we dropped them from our analysis.