A Still Planet in a Dance of Light ✛
To see an exoplanet, a planet around another star, astronomers use special instruments that block the star's blinding light. But even inside these instruments, some light remains, turning into twinkling speckles — false targets that look very much like planets. In a single image, it is nearly impossible to tell a planet from a speckle. However, astronomers have a clever trick: the telescope is equipped with a flexible mirror that can be bent ever so slightly. The speckles are coherent light (they can interfere like waves on water), so they start to dance with each movement of the mirror. A real planet shines incoherently, its light is like that from a light bulb, and remains completely still. Like a stone at the bottom of a stream, around which sun sparkles dance.
A group of researchers from the Paris Observatory built a Machine Learning algorithm (XGBoost)—a computer program that learns from its mistakes and becomes more accurate. For training, they created 2250 numerical simulation images of the starry sky using the Asterix package. They mixed in 0 to 2 planets into these images, and then, by slightly changing the mirror shape, obtained series of frames. All suspicious points were passed through a detection algorithm, and for each, a list of features was compiled — how brightness changes from frame to frame. The program analyzed 1200 such examples and learned to distinguish where a real planet is and where it is just ripples.
After training, the algorithm was tested on 750 new scenes. It made very few mistakes: false alarms were rare, and about 80% of planets were found — even those 500 000 times fainter than the star and completely invisible at first glance. The key advantage is that the method needs only 5–7 images to distinguish a stone from dancing reflections. This is especially valuable for the future space telescope JWST (named after Nancy Grace Roman) and ground-based instruments like SPHERE+, where every minute of observation is precious. This entire approach combines ideas from interferometry (the study of light interference), Digital Image Processing, and signal processing. Instead of a long struggle with noise, astronomers can immediately start studying the atmospheres of distant planets — a task that Michel Mayor, who discovered the first exoplanet around a normal star, dreamed of.
The word ‘speckle’ comes from an Old English root meaning ‘little spot.’ These devious little spots have long taunted astronomers by posing as planets at every opportunity. To outsmart them, EPICX borrows an idea from another era: Pair Wise Probing was first proposed in 2007, long before anyone dreamed of widespread machine learning in astronomy.
In Carl Sagan’s novel *Contact*, Ellie Arroway struggles to fish a signal out of the noise — much like EPICX pulls a planet from a chorus of stellar interference, only with light waves instead of radio waves, and giant ground- and space-based observatories instead of radio telescopes.