Data driven theory for knowledge discovery in the exact sciences with applications to thermonuclear fusion

Area of Science:

• Exact sciences, Physics, Computational science

Background:

• Traditional hypothesis-driven methods struggle with complex, non-linear systems and large datasets.
• Limitations of first-principle theories in handling uncertainty and data volume are increasingly apparent.

Purpose of the Study:

• To introduce and evaluate a novel data-driven theory formulation approach.
• To complement existing scientific methodologies by leveraging large datasets for model discovery.

Main Methods:

• Symbolic manipulation via genetic computing for data-driven model discovery.
• Exploration of vast datasets to identify optimal mathematical models.
• Application to complex physical systems, including thermonuclear plasmas.

Main Results:

• Demonstrated potential in formulating scaling laws and identifying key dimensionless variables.
• Successful application to complex experiments, including analysis of catastrophic instabilities in thermonuclear plasmas.
• Validation through extensive numerical testing across diverse scientific studies.

Conclusions:

• The data-driven approach effectively bridges the gap between experiments, data analysis, and theory.
• Genetic computing offers a robust tool for uncovering insights in complex, non-linear systems.
• This methodology is increasingly adopted across various scientific fields for advanced data interpretation.