Play gravitational waves
The chirp
Two supermassive black holes in a tight orbit leak energy as gravitational waves, so the orbit shrinks and they go around faster, and that makes them leak energy faster still. For the default pair the first orbits here take 21 years and the last one takes under four days. The clock runs on a log scale, so every second on screen divides the time left by the same factor.
Loading the 3D view.
- GW frequency
- 3.0 nHz
- Orbital period
- 21 years
- Separation
- 0.033 pc
- Relative speed
- 0.032 c
Sound is off. Pitch follows the real gravitational-wave frequency, squeezed into 60 to 1000 Hz.
Who could hear it
characteristic strain vs frequencyWhat a telescope would see
too slowWhat you're seeing
The 3D view is the orbit. Every pair starts 10 million years before it merges. Sizes and speeds are squashed so it fits on a screen, but the order of events is real: the frequency follows Peters (1964) with the first post-Newtonian correction, and the readouts are in real units. The sheet underneath is a cartoon of curved space. Each hole sits over its own dip, and the ripples running outward are the gravitational waves, two crests per orbit, packed tighter as the orbit speeds up. The glow around a hole turns bluer and brighter while that hole is coming toward you. That's Doppler boosting, and you only notice it near the end, when the two holes pass each other at up to 0.41 times the speed of light.
The plot is strain against frequency, both on log axes, and the dot is the binary. The shaded columns are where pulsar timing arrays listen (1 to 100 nHz, with NANOGrav's 15-year curve from Agazie et al. 2023) and where LISA would (0.1 to 100 mHz, the Robson, Cornish & Liu 2019 curve for the mission in Amaro-Seoane et al. 2017). A signal has to sit above a curve to be heard. The merger frequency goes as one over the total mass, so merging inside the LISA band needs less than about 4 × 10⁷ M☉ in total. Merging inside the PTA band would need more than 4 × 10¹⁰ M☉, which the sliders don't even reach.
The sound is the same frequency sweep squeezed into 60 to 1000 Hz, and it gets louder as the real strain grows. The real waves are far too slow to hear. At 1 nHz a single wave takes 32 years to go past.
The bottom strip is what an optical survey might record: a sine at the orbital period from Doppler boosting, sitting on damped-random-walk noise, which is how quasars flicker anyway. I made both amplitudes up. A 10-year survey needs three full cycles and a few visits per cycle, so it can only catch periods of about 100 to 1,200 days, and random flickering is annoyingly good at faking exactly this kind of wiggle.
What this toy leaves out
Orbits are circular and the black holes don't spin. The frequency evolution stops at first post-Newtonian order, which is already unreliable in the last few orbits, and the merger flash is only a drawing. There's no gas and there are no stars around the pair. At wide separations those shrink the orbit faster than gravitational waves do, so the early numbers on the clock are too long. No redshift either.
Both sensitivity curves are approximate. The NANOGrav one is 11 points read off Agazie et al. (2023) by hand, and the LISA fit leaves out the foreground from binaries in our own galaxy. The height of the dot also flatters slow PTA pairs. It counts every cycle the binary spends near a frequency, but 15 years of timing only sees about one and a half cycles at 3 nHz, and the NANOGrav curve is meant for a background made by many binaries at once.
The Python code this is ported from
I searched 1,369 galaxies for these
References
- Agazie, G., et al. 2023, ApJL, 951, L8. The NANOGrav 15 yr data set: evidence for a gravitational-wave background.
- Amaro-Seoane, P., et al. 2017, arXiv:1702.00786. Laser Interferometer Space Antenna.
- Peters, P. C. 1964, Phys. Rev., 136, B1224. Gravitational radiation and the motion of two point masses.
- Robson, T., Cornish, N. J., & Liu, C. 2019, CQG, 36, 105011. The construction and use of LISA sensitivity curves.