Black holes produce gravitational waves when they collide
Victor de Schwanberg / Science Photo Library
The efforts to understand the universe could obtain a boost of an AI developed by Google Deepmind. The algorithm, which can reduce unwanted noise up to 100 times, could allow the laser interferometer the gravitational wave observatory (LIGO) to identify a particular type of black hole that has so far escaped us.
Ligo is designed to detect gravitational waves produced when objects such as black holes are unleashed and collided. These waves cross the universe at the speed of light, but the fluctuations they cause in space -time are extremely small – 10,000 times smaller than the nucleus of an atom. Since his first observations 10 years ago, Ligo recorded these signals produced by nearly 100 collisions of black holes.
To do this, the experience consists of two observatories in the United States, each with two arms 4 kilometers long which are perpendicular to each other. The lasers are striped on each arm, thought out by precise mirrors at the end, then compared using an interferometer. The length of the arms is modified by a small quantity as gravitational waves reveal them, and this is carefully recorded to build an image of the origin of these signals.
The problem is that such demanding clarification is necessary for even the waves or the clouds of the ocean at a distance can affect the measurements. This noise can easily drown signals, making observations impossible. Dozens of major settings must be made to filter the worst of this noise, refining the orientation of mirrors and other equipment.
Rana Adhikari at the California Institute of Technology in Pasadena, who worked with Deepmind to develop the new technology of AI, says that trying to automate these adjustments can ironically create more noise. “This controls the noise striking for decades and decades – everything in this area has been blocked,” said Adhikari. “How do you hold the mirrors if still without inducing noise? If you do not control them, the mirrors are swinging everywhere, and if you control it too much, then it buzzes in a way.”
Laura Nuttall at the University of Portsmouth in the United Kingdom was one of the scientists who used to manually make these adjustments to the Ligo. “As you move something, something else goes, and something else goes and something else goes,” she said. “You would go to evannage forever.”
Deepmind’s new deep loop aims to reduce the noise level of the adjustment of mirrors to Ligo up to 100 times. The AI has been trained in a simulation before testing in the real world and is indeed responsible for reaching two objectives: reducing noise and minimizing the number of adjustments it makes. “Over time, by doing this several times – it’s like hundreds and thousands of events that work in simulation – the controller will find it in a way what works and what will not work and find a very good policy,” explains Jonas Buchli in Deepmind.
Alberto Vecchio at the University of Birmingham, in the United Kingdom, which has not been involved in research but which works on Ligo, says that AI is exciting, although there are still many obstacles to overcome.
First, the technology has only been managed for an hour in the real world on Ligo, so it must be shown that it can work for weeks, or even months at a time. Second, the technology has so far been applied to a single aspect of control, helping to stabilize mirrors, and there are hundreds, even thousands of aspects, to which it could be applied.
“This is clearly the first step, but I still think it’s very intriguing. And there is clearly a lot of room for enormous progress, ”explains Vecchio.
If similar improvements could be made at all levels, he thinks that we could spot so -called intermediate -sized black holes – for example those who have masses about 1000 times that of our sun – a class of objects without any confirmed observation. Improvements would tend to occur on low frequency gravitational waves, where the wavelength is more sensitive to noise and which are created by larger objects.
“We know black holes up to 100 solar masses. We know the black holes of our galaxy which are a million solar masses and above. What is between the two? ” said Vecchio. “People think that there will be black holes in all these different mass channels, but no one has had proof of non -controversial experimental observation.”
Nuttall says that the new approach could also provide a more detailed observation of the types of black hole that we have already seen. “It seems damn well,” she said. “I am super excited by that.”
Subjects:
