In 2015, two detectors in the United States felt the universe ring. Two black holes, more than a billion light-years away, had spiraled together and merged, and the shudder it sent through spacetime stretched the Earth by a distance far smaller than the width of a single proton. That such a thing could be measured at all is remarkable. How it was measured is the part worth telling.
The detectors, known as LIGO, are enormous L-shaped instruments with arms four kilometers long, watching for the instant those arms change length by an almost unimaginably tiny amount. The difficulty is that nearly everything changes their length — a truck on a distant road, a faint earthquake, the thermal jitter of the atoms in the mirrors. The signal they wanted was buried beneath noise thousands of times larger.
You cannot simply look at the data and see the wave. It isn’t visible. It has to be pulled out, and the tool that pulls it out is an idea called matched filtering.
The idea is this: if you know roughly what a signal should look like, you can slide that expected shape along your noisy data and measure, at every instant, how well the two line up. Noise, being random, rarely matches a specific shape for long. A real signal does. The known shape acts like a key cut for one particular lock.
But the shape of a gravitational wave depends on the masses spiraling together, and those were unknown in advance. So the physicists computed not one expected shape but hundreds of thousands of them — a vast bank of templates, each the predicted waveform of a different pair of merging bodies, every one derived from Einstein’s equations. The data was compared against the whole library.
One matched: two black holes, each around thirty times the mass of the Sun, rising in pitch as they fell together into a single object — a chirp lasting a fifth of a second. The detection was as much an act of computation as of measurement, a century-old theory turned into a catalog of predictions and then searched for the one the universe had actually played.
It is easy to call this signal processing and move on. But notice what it demanded — that you know what you are looking for precisely enough to recognize it inside noise that drowns it. The instrument heard everything at once. The software decided what was real.
There’s a discipline in that we admire — knowing exactly what you’re building toward, precisely enough to find it in the noise. More about how we work →