Emergent Schrödinger equation in an introspective machine learning architecture

Area of Science:

• Quantum mechanics
• Artificial intelligence
• Machine learning

Background:

• Investigating the emergence of physical laws within neural networks.
• Exploring AI's capability to learn from physical system data.

Purpose of the Study:

• To demonstrate an AI architecture that can automatically develop physical concepts.
• To show the discovery of the quantum wave function and Schrödinger equation using simulated data.

Main Methods:

• Developed an introspective learning architecture.
• Utilized a machine translator for potential-to-density mapping.
• Employed a knowledge distiller auto-encoder to extract hidden state information.

Main Results:

• The AI architecture successfully developed the concept of the quantum wave function.
• The Schrödinger equation was discovered from simulated quantum data.
• The system extracted essential information and update laws from hidden states.

Conclusions:

• Introspective learning architectures can enable AI to discover fundamental physics.
• This approach serves as a proof-of-principle for AI-driven physics discovery.
• The developed architecture has the potential to accelerate future physics research.