On ab initio-based, free and closed-form expressions for gravitational waves

Area of Science:

• Astrophysics
• Gravitational Wave Astronomy
• Computational Physics

Background:

• Numerical relativity simulations are crucial for understanding binary black hole mergers.
• Existing models often involve complex computations and approximations.
• Developing accurate, efficient analytical descriptions of gravitational waves remains a challenge.

Purpose of the Study:

• To introduce a novel AI-based approach for generating high-accuracy, closed-form expressions for gravitational waves.
• To enable rapid searching and validation of gravitational wave signals.
• To provide physically accurate models without underlying approximations.

Main Methods:

• Combining Artificial Intelligence (AI) with Reduced Order Modeling (ROM) techniques.
• Utilizing symbolic regression via genetic programming.
• Employing sparse representations in parameter and time domains with Reduced Basis and Empirical Interpolation Methods.

Main Results:

• Achieved minimum overlap of 99% with ground truth solutions for binary black hole collisions.
• Generated closed-form expressions that are concise and computationally efficient.
• Demonstrated a method free from underlying physical approximations.

Conclusions:

• The developed AI-driven approach offers a significant advancement in modeling gravitational waves.
• This method enhances the speed and accessibility of analyzing gravitational wave data.
• The findings contribute to both gravitational wave astrophysics and the field of artificial intelligence.