Complex optimization for big computational and experimental neutron datasets.

Area of Science:

• Computational physics
• Materials science
• Data science

Background:

• Solving inverse optimization problems with large experimental datasets is computationally challenging.
• Accurate computational models are crucial for interpreting complex experimental results.

Purpose of the Study:

• To present a high performance computing framework for accurate solutions to inverse optimization problems.
• To demonstrate the application of image processing, mathematical regularization, and hierarchical modeling for big data optimization.
• To refine first principles calculations using experimental data for improved model accuracy.

Main Methods:

• Utilized high performance computing for inverse optimization.
• Employed image processing, mathematical regularization, and hierarchical modeling techniques.
• Integrated model and data information to establish confidence regions for algorithms.
• Applied the framework with the SIMPHONIES software package.

Main Results:

• Successfully solved complex optimization problems on big data.
• Increased solution accuracy by incorporating model and data information.
• Provided confidence regions for processing and regularization algorithms.
• Refined first principles calculations for neutron scattering data on silicon.

Conclusions:

• The developed framework effectively addresses inverse optimization challenges with big experimental data.
• The integration of computational techniques enhances the accuracy of scientific models.
• This approach offers a robust method for analyzing experimental results and improving theoretical descriptions.