Search for gravitational waves from Scorpius X-1 with a hidden Markov model in O3 LIGO data

Results are presented for a semicoherent search for continuous gravitational waves from the low-mass x-ray binary Scorpius X-1, using a hidden Markov model (HMM) to allow for spin wandering. This search improves on previous HMM-based searches of Laser Interferometer Gravitational-Wave Observatory data by including the orbital period in the search template grid, and by analyzing data from the latest (third) observing run. In the frequency range searched, from 60 to 500 Hz, we find no evidence of gravitational radiation. This is the most sensitive search for Scorpius X-1 using a HMM to date. For the most sensitive subband, starting at 256.06 Hz, we report an upper limit on gravitational wave strain (at 95% confidence) of $h_0^{95\%}=6.16\times {10}^{-26}$, assuming the orbital inclination angle takes its electromagnetically restricted value $\iota =44°$. The upper limits on gravitational wave strain reported here are on average a factor of $\sim 3$ lower than in the second observing run HMM search. This is the first Scorpius X-1 HMM search with upper limits that reach below the indirect torque-balance limit for certain subbands, assuming $\iota =44°$.

Figure 1

Uncertainty ellipses corresponding to $1\sigma$ (blue line), $2\sigma$ (yellow line), and $3\sigma$ (green line), for $T_{\mathrm{asc}}$ and $P$. Top: the curves at the reference time $T_0$, when $T_{\mathrm{asc},\mathrm{ref}}$ is uncorrelated with $P_0$ or its uncertainties. Bottom: the effect of propagating $T_{\mathrm{asc},\mathrm{ref}}$ and $P_0$, with their respective uncertainties, to the start of O3 ($T_{\mathrm{O}3,0}=1238166483$ GPS time) (solid lines) or the end of O3 ($T_{\mathrm{O}3\mathrm{end}}=1269361423$ GPS time) (black dots). $T_{\mathrm{asc},0}$ represents the central value of $T_{\mathrm{asc},\mathrm{ref}}$ propagated to the start or end of O3; that is, $T_{\mathrm{asc},0}$ is different for the solid and dotted lines. In both panels we have $P_0=68023.86048\mathrm{s}$.

Figure 2

Flowchart of the search pipeline for a subband. The light red octagons are the start and end points, the orange rectangles are processes, the blue parallelograms are input or output data, and the green ovals stand for decision points. The full search repeats the steps in the flowchart for 725 subbands from 60 to 500 Hz.

Figure 3

Candidates plotted as a function of their terminating frequency bin $q^{*}(t_{N_T})$ (horizontal axis, units in hertz) and the orbital parameters $a_0$ (vertical axis in left panel, units in seconds), offset from the central time of ascension $T_{\mathrm{asc}}-T_{{\mathrm{asc}}_0}$ (vertical axis in central panel; units in seconds) and offset from the central period $P-P_0$ (vertical axis in right panel; units in seconds). The color scale indicates the $\max (\mathcal{L})$ obtained for the candidate. Candidates marked with purple squares are eliminated by the single IFO veto, while red circles mark the ones eliminated by the known lines veto. The candidate with no marking survives both the single IFO and known lines vetoes, but is eliminated by its absence when using noise-subtracted data (see Sec. 4a).

Figure 4

Frequentist effective wave strain upper limits at 95% confidence as a function of subband frequency, for three scenarios: circular polarization with $\iota =0$ (blue stars), $\iota \approx 44°$ based on the electromagnetic measurements (see Ref. [51]; orange dots), and a flat prior on $\cos \iota$ (green dots). Indirect torque-balance upper limits (see Sec. 5c) for two torque lever arms are also shown: the stellar radius (red solid line) and the Alfvén radius (dashed red line).