Andrei, Sheila, Naoki

We analyzed the impact of ZM4 and ZM5 curvatures on squeezing level. We choose the channels H1:SQZ-DCPD_RATIO_#_DB_MON to analyze your squeezing.  Channel H1:GRD-IFO_OK was used to make a mask that will filter out all the data points that correspond to times when the measurements were not performed. The DBSCAN method from sklearn.cluster is used to cluster spatially separated data. Subsequent operations check for temporal cluster overlap and separate them.

Presumably, histogram plots can help determine the optimal parameters to achieve the best squeezing.

However, obviously, not only changes in ZM4 and ZM5 were made over the 200 days of observation. As seen from Figure_1.png, several peaks in histograms were likely caused by changes in other parameters. In the future, it would be nice to implement an additional parameter for clustering based on the squeezer phase or others.

Link to code: git.ligo.org/andrei.danilin/zm4-and-zm5-curvature-impact-on-squeezing-level

ZM4 and ZM5 curvatures channels could be easily swappped to others.