Modeling the greatest discovery in theoretical Physics with SciML | Chandrasekhar equation | NeurIPS
"Chandrasekhar limit: The greatest discovery in theoretical Physics"
In 1834 F.W. Bessel noticed something strange: Sirius, the brightest star in our night sky was moving a bit irregularly. 28 years later this led to the discovery of Sirius B in 1862 by Alvin G. Clarke, a star with a mass similar to our Sun but 400 times less luminous!
The mystery deepened when Walter Adams obtained Sirius B's spectrogram in 1915, revealing a temperature similar to Sirius A despite its low luminosity. This could only be explained by an incredibly small size. But this would mean that Sirius B's density would be 100,000 times that of water!
In 1930, Subrahmanyan Chandrasekhar worked on this problem. This brilliant Indian physicist combined quantum mechanics and relativity to explain the "electron degeneracy pressure" that balanced the self-crushing gravitational force. He calculated that no white dwarf could exist with a mass exceeding 1.4 times the sun's mass - the famous Chandrasekhar limit. This is one of the greatest discoveries of all time in theoretical Physics for which Chandrasekhar earned the Nobel Prize in Physics in 1983.
Our story begins with Chandrasekhar's White Dwarf equation which explains the relationship between density and the radial location of the white dwarf. We at Vizuara recently had a paper accepted to a prestigious NeurIPS workshop, applying scientific machine learning to solve Chandrasekhar's white dwarf equation.
We could model various cases of the equation with no-noise, high-nose, and intermediate-level noise. We showed that the Universal Differential Equation (UDE) framework from SciML can model Chandrasekhar's equation, even in the absence of 90% of the data. We also found a "forecast breaking point" after which the UDE framework makes poor predictions.
Shout out to Raymundo, Raj, and Rajat for putting this great work together.
I have released this simple video on Vizuara's YouTube channel explaining white dwarfs, Chandrasekhar limit, and our paper. I am sure you will enjoy this.
If you are interested in reading our paper check out on arXiv: https://arxiv.org/abs/2410.14998
If you are interested in Machine Learning or Scientific Machine learning feel free to reach out to us or check out our program: https://vizuara.ai/spit
