Effect of squeezing on parameter estimation of gravitational waves emitted by compact binary systems

The LIGO gravitational wave (GW) detectors will begin collecting data in 2015, with Virgo following shortly after. These detectors are expected to reach design sensitivity before the end of the decade, and yield the first direct detection of GWs before then. The use of squeezing has been proposed as a way to reduce the quantum noise without increasing the laser power, and has been successfully tested at one of the LIGO sites and at GEO in Germany. When used in Advanced LIGO without a filter cavity, the squeezer improves the performances of detectors above $\sim 100\mathrm{Hz}$, at the cost of a higher noise floor in the low-frequency regime. Frequency-dependent squeezing, on the other hand, will lower the noise floor throughout the entire band. Squeezing technology will have a twofold impact: it will change the number of expected detections and it will impact the quality of parameter estimation for the detected signals. In this work we consider three different GW detector networks, each utilizing a different type of squeezer—all corresponding to plausible implementations. Using LALInference, a powerful Monte Carlo parameter estimation algorithm, we study how each of these networks estimates the parameters of GW signals emitted by compact binary systems, and compare the results with a baseline advanced LIGO-Virgo network. We find that, even in its simplest implementation, squeezing has a large positive impact: the sky error area of detected signals will shrink by $\sim 30\%$ on average, increasing the chances of finding an electromagnetic counterpart to the GW detection. Similarly, we find that the measurability of tidal deformability parameters for neutron stars in binaries increases by $\sim 30\%$, which could aid in determining the equation of state of neutron stars. The degradation in the measurement of the chirp mass, as a result of the higher low-frequency noise, is shown to be negligible when compared to systematic errors. Implementations of a quantum squeezer coupled with a filter cavity will yield a better overall network sensitivity. They will give less drastic improvements over the baseline network for events of fixed signal-to-noise ratio but greater improvements for identical events.

Figure 1

(Top) The PSDs for the baseline LIGO (black solid) and Virgo (blue dotted) detectors and for a LIGO detector with frequency-independent squeezing (red dashed). (Bottom) The PSDs for a squeezed LIGO detector with lossy (green dashed) and lossless (magenta dotted) filter cavities; the design Advanced LIGO curve (solid line) is given for reference.

Figure 2

A cumulative histogram of the injected distance for the BNS events for each network (where the SNR distribution is identical for all networks). Note that the distances of the squeezed events are very similar to those of the baseline events, which implies that the squeezed and baseline networks have very similar network sensitivities. On the other hand, the distances of events for the lossy and lossless networks are further than for the baseline network, which implies that these two networks have greater network sensitivity (with the lossless network being most sensitive).

Figure 3

Plots of the posterior density distributions of $\tilde{\mathrm{\Lambda }}$ as recovered by the baseline (blue, continuous) and squeezed (red, dashed) networks, for the BNS system with $\tilde{\mathrm{\Lambda }}=197.899$ and SNR 20 (top) and the system with $\tilde{\mathrm{\Lambda }}=820.610$ and SNR 30 (bottom).

Figure 4

The posterior density distributions for $\mathcal{M}$ obtained with both the IMRb and IMRc waveforms for a BBH event with $\mathcal{M}=11.43M_{\odot }$ (analyzed with the baseline network). The absolute separation of the distribution medians is 2.8 times larger than the standard deviation of the IMRb mode.