above Artificial intelligence algorithms and Python big data Get more dry goods
On the top right ··· Set to star *, Get resources the first time
Just for academic sharing , If there is any infringement , Contact deletion
Reproduced in : qubits
Simulate gravitational waves produced by black holes , That sounds terrible , It must take a lot of computing power ?
exactly , Discover the gravitational waves predicted by Einstein , Humans use it 100 year , And using supercomputing to simulate it exactly , People use 90 year !
1915 year , Einstein published general relativity , And then physicists predicted , A merger of two black holes produces Gravitational waves .
until 2005 year , Scientists got the first Black holes merge numerical solutions , And it's done off and on with supercomputing 2 Months .
But now , You just need one macOS or Linux System laptop , You can also calculate black hole mergers , It's the one with animation .
This is a Ph.D. program from Caltech Vijay Varma Developed a Python package , It is used to simulate how two black holes radiate gravitational waves during rotation , And the whole process of their merger .
The installation method is so simple that it can even be done through PyPI Direct installation :
pip install binaryBHexp
And then you just type in a bunch of parameters , You can simulate gravitational waves on your laptop . Even drag the animation with the mouse , comprehensive 360 Watch the black hole merging process :
You think that's all ? No no no , These are just “ By-product ” nothing more .
real “ The main course ” yes , This doctor uses AI Developed By far the most accurate Black hole merging model , And it greatly shortens the simulation time .
Now physicists are going to use this technique to simulate more complex black hole merging processes , help Gravitational wave interference observatory (LIGO) More gravitational waves can be found , Or to test general relativity , Or find its flaws .
With this work , The doctor has been in the top journals 《 Physical Review Letters 》 A number of papers have been published .
Since gravitational waves can be found experimentally , Why do we have to simulate it numerically ? Before we get to know the problem , We have to solve a problem first :
How do we observe gravitational waves ?
Einstein's general theory of relativity , The gravitational wave is “ Ripples of time and space ”, It's the disturbance of a mass object to space-time when it's moving .
however , Gravitational waves are so weak . Only Black hole merging Such incidents , To radiate the gravitational waves that we find .
Black hole merging is now recognized as The strongest gravitational wave source , Because the black hole itself is just a strong gravitational source , So in the merger process , They only radiate gravitational waves .
In order to observe gravitational waves , come from Caltech and MIT A group of physicists , We set up a laser interference gravitational wave observatory LIGO.
This place is dedicated to detecting gravitational waves , It seems that there are two right angle distribution “L” Arm shaped . When gravitational waves appear , The laser in each arm , Will measure the relative difference in arm length .
This process is very difficult , Because every day LIGO Will receive a lot of weak signals with a lot of noise .
Assistant professor from the University of Massachusetts Scott E. Field Explain this :
The difficulty , It's like trying to listen to music on your mobile phone in a noisy restaurant .
I only know the content of the song , In order to find it more easily in the background noise .
in other words , We have to find a way to simulate it numerically , And then we can detect it .
This is the significance of numerical simulation of gravitational waves .
After all , Only a few gravitational waves have been detected , Not even what it looks like Completely understand , Different quality 、 What kind of gravitational wave does rotation and revolution speed produce , We need to solve the very complex general relativity equation to simulate .
But in the numerical simulation of gravitational waves , Physicists are having trouble again ——
Using supercomputing to solve the general relativity equation , Can only Faster Simulate some of the gravitational wavelengths , It's the mass ratio Less than 10:1 The gravitational waves produced by the merging of two black holes in the universe .
For the merging of these black holes , From the University of Massachusetts Gaurav Khanna Express :
It's like simulating the interaction between a big ship and a small sailing boat , After all, the latter will hardly affect the route of the giant ship .
But for another part of the black hole , That is, the mass ratio is greater than 10:1 The gravitational waves produced by the merging of two black holes in the universe , The amount of computation required for simulation is too large .
2005 year , Physicists used supercomputers to simulate 2 Months , And then we get a numerical solution . For a mass ratio greater than 10:1 The situation of , It may take years for Supercomputing , This is obviously unrealistic .
So the mass ratio is greater than 10:1 The two black holes of the merger , Can't really detect their gravitational waves ?
There is actually another way —— simplified calculation .
These physicists from the University of Massachusetts , I hope to use machine learning to simplify the calculation process .
They even made a real Python tool kit , And from the research results , It has been successfully simulated that the mass ratio is 3:1 The merging process of black holes .
The calculated results are consistent with those of supercomputing simulation , The accuracy is no less than 1%.
This visualization tool that simulates black hole merging is called binaryBHexp.
The installation process is very simple , I've said that before . It's also very simple to use .
Because the merging of black holes only depends on the following physical quantities : quality 、 Angular momentum 、 Revolution speed .
Enter these values into the command :
binaryBHexp --q 2 --chiA 0.2 0.7 -0.1 --chiB 0.2 0.6 0.1
Parameters q It's the mass ratio of two black holes ,chiA and chiB Then there are the rotation and revolution speeds of the two black holes ( All have been normalized ).
Different parameters lead to different black hole merging phenomena .
For example, the following set of parameters , It shows that gravitational waves are huge “ Recoil ”, Its energy can accelerate the black hole to the speed of light 1/100, Throw it out of the galaxy :
binaryBHexp --q 1.34 --chiA 0.62 -0.27 0.34 --chiB -0.62 0.27 0.34
On the ground LIGO It's not enough for physicists .
On earth, , The length of the two interference arms used to measure gravitational waves is limited , If we build a probe into space , So the interference arm can be as long as 100 More than kilometers , Greatly improve the detection accuracy .
This is the European Space Agency ESA and NASA Imagine a space-based gravitational wave exploration program LISA, Due in 2035 launch .
In space , The improvement of accuracy can let us see black hole merging events with larger mass ratio , For example, the mass ratio exceeds 100 The situation of Wan .
Because there might be in the center of the galaxy 10 Hundreds of millions of solar mass black holes , When it sucks ordinary black holes into it , This kind of super mass ratio merging event will occur .
On the other side , Physicists are preparing for numerical computation .
Scott Field and Gaurav Khanna The professor expects to publish a larger mass ratio calculation model in this summer arxiv On , I don't know what surprise it will bring .
Project address :
https://vijayvarma392.github.io/binaryBHexp/
Reference link :
[1]https://www.quantamagazine.org/new-black-hole-math-closes-cosmic-blind-spot-20210512/
[2]https://phys.org/news/2019-01-physicists-supercomputers-ai-accurate-black.html
[3]https://phys.org/news/2020-03-method-black-holes-galaxies.html
[4]https://pypi.org/project/surfinBH/
[5]https://pypi.org/project/binaryBHexp/
------------------
Statement : This content comes from the Internet , The copyright belongs to the original author
Picture source network , It does not represent the position of the official account . If there is any infringement , Contact deletion
AI Doctor's personal wechat , There are still a few vacancies
How to draw a beautiful deep learning model ?
How to draw a beautiful neural network diagram ?
Read all kinds of convolutions in deep learning
Let's have a look and support