Gaussian process tomography for soft x-ray spectroscopy at WEST without equilibrium information.

Area of Science:

• Plasma physics
• Tomography
• Bayesian inference

Background:

• Soft X-ray (SXR) tomography is essential for understanding plasma behavior in fusion devices.
• Existing tomography methods can be computationally intensive, limiting real-time applications.
• Gaussian process tomography (GPT) offers a novel Bayesian approach to SXR tomography.

Purpose of the Study:

• To validate the Gaussian process tomography (GPT) technique using synthetic data from the WEST tokamak.
• To apply GPT for reconstructing SXR emissivity profiles from Tore Supra experimental data.
• To compare the performance of GPT against the standard Fisher information minimization algorithm.

Main Methods:

• Modeling the soft X-ray (SXR) emissivity field as a Gaussian process.
• Utilizing Bayesian probability theory for tomographic reconstruction.
• Validation with synthetic WEST tokamak data and application to Tore Supra measurements.

Main Results:

• GPT provides robust and extremely fast SXR tomography reconstructions.
• The Bayesian formalism allows for quantification of uncertainty in inferred parameters.
• GPT demonstrates comparable or superior performance to standard methods.

Conclusions:

• GPT is a promising candidate for real-time SXR tomography in fusion research.
• The method's speed and robustness are advantageous for impurity transport and magnetohydrodynamic control.
• GPT offers a reliable tool for analyzing plasma emissivity profiles.