This talk will cover two topics, both of which concern the development and use of computational algorithms to enhance our interferometers.
First, I will discuss a scheme developed at the 40m prototype to automatically and dynamically blend error signals in an optimal way. The motivating application is lock acquisition, where commissioner time and experience is often spent determining when and how to blend and handoff the interferometer control signals. I will describe testing of this algorithm at the 40m that successfully shows its ability to smoothly transition control of an arm cavity between three error signals as it scans through resonance, in a manner analogous to the aLIGO lock acquisition process.
Second, I will discuss ongoing efforts to regress noise out of our interferometer output; in particular during the events observed during O1. I will show how an initial pass of linear subtraction of known noise sources (such as angular control noise and input periscope jitter) can improve the uncertainties on our estimates of the binary source parameters. Finally, I will describe the efforts of a group of people investigating nonlinear regression methods that have potential to recover complex noise couplings, such as the influence of scattered light.
We plan to broadcast these talks using TeamSpeak. Use a sub-channel of LIGO Lab called "LIGO Seminar", which is not password protected. NOTE: These and all other scheduled LIGO seminars are listed on the LIGO Laboratory seminar calendar for convenient reference.