Skip to main content
Sandbox Physics
Matched filtering with GW150914

What arrival time and chirp mass do to the match

Sandbox Physics ·

A time shift moves every crest by the same amount. A mass change also stretches or squeezes the pattern, so finding the best shift cannot in general repair a mismatch in mass.

Open the GW150914 analysis

Compare two whitened waves

Use complex Fourier samples of the measured strain and of the template. The inner product divides by the detector noise power.

(a∣b)=4 Re∫fmin⁡fmax⁡a~(f)b~∗(f)Sn(f) df.(a\mid b)=4\,\mathrm{Re}\int_{f_{\min}}^{f_{\max}}\frac{\widetilde{a}(f)\widetilde{b}^{*}(f)}{S_n(f)}\,\mathrm{d}f.

A trial arrival multiplies the template spectrum by a phase that varies linearly with frequency. With many shifts to test, a single inverse FFT evaluates the whole set of inner products. Compare a cosine-like template and its quarter-cycle companion. Their two scores give the strength of the best overall phase at each time. Compact binary search methods in the LIGO–Virgo guide

One family of black holes

In an inspiral the dominant rate of frequency change depends on the chirp mass. If MM is the total mass and η\eta is the product of the component masses divided by the square of the total mass, then

M=M η3/5.\mathcal{M}=M\,\eta^{3/5}.

The GWOSC teaching template comes from SEOBNRv2 and fixes the mass ratio and both dimensionless aligned spins. Stretch time by the ratio of new to original detector-frame mass. A heavier choice places fewer oscillations into the same time window. After the fit the amplitude is free again, so the reported result is a conditional shape comparison. Frozen tutorial configuration

For a nearly circular orbit that changes slowly, Kepler’s law gives the orbital frequency and the quadrupole power gives the rate at which the orbit loses energy. The waves have twice the orbital frequency. Equating the orbital energy loss with that power gives the rate for the observed wave frequency.

dfdt=965 π8/3(GMc3)5/3 ⁣f11/3.\frac{\mathrm{d}f}{\mathrm{d}t}=\frac{96}{5}\,\pi^{8/3} \left(\frac{G\mathcal{M}}{c^3}\right)^{5/3}\!f^{11/3}.

Here time is measured in seconds and frequency in hertz. A doubling of the frequency raises its rate of change by about thirteen at fixed mass. This is an early-inspiral approximation. The full template also follows the last cycles and merger. The basic physics of GW150914, Eq. 7 and Appendix A

Worked comparison at the two observatories

The highest combined score in the frozen family uses total detector-frame mass 68 solar masses and chirp mass 29.0 solar masses. It combines the separately maximized Hanford and Livingston scores. At this mass Hanford’s best arrival is 7.08 milliseconds after Livingston’s. A common shift convention applies before subtracting those times.

Exercise. For equal component masses, what chirp mass corresponds to a total mass of 68 solar masses?

Answer. Equal masses give η=1/4\eta=1/4, hence about 29.6 solar masses. The value changes when the mass ratio changes. The comparison in the Lab holds that ratio fixed. Source-frame mass would need a separate redshift correction.

Sources

  1. Compact binary search methods in the LIGO–Virgo guide
  2. Frozen tutorial configuration
  3. The basic physics of GW150914, Eq. 7 and Appendix A

Keep reading

What this page links to directly

See this series on the content map →

Matched filtering with GW150914 · Reading path

Matched filtering with GW150914

  1. ArticleHow to find a chirp in detector noise
  2. ArticleWhy a strain record is whitened
  3. You are hereWhat arrival time and chirp mass do to the match
  4. ArticleRecompute the GW150914 match from the public strain
  5. ArticleRead a frequency-consistency test for glitches
1 more group

Other articles that use the same experiments

Each experiment from this article that another series also uses, followed by the articles there.

  1. LabBinary Neutron Star Merger
  2. ArticleA pulsar clock revealed an orbit losing energy
Something broken, confusing, or missing on this page?Tell us
What kind of feedback?

Sent with this page’s address, its Lab ID, the site language and your device type. Please leave out personal details. Feedback is kept for up to 12 months.Privacy policy