Principal Component Analysis Applied to Digital Pulse Shape Analysis for Isotope Discrimination.

Area of Science:

• Nuclear Physics
• Detector Technology

Background:

• Digital pulse shape analysis (DPSA) is crucial for nuclear reaction studies with modern digitizers.
• Accurate (A, Z) value determination of reaction products is essential for solid-state detector applications.

Purpose of the Study:

• To develop a computationally efficient method for discriminating isotopes with similar energy levels.
• To enable the edge-computing paradigm in future FPGA-based acquisition systems.
• To compare selectivity and computational efficiency against existing methods.

Main Methods:

• Utilized a dataset from the FAZIA Collaboration at GANIL.
• Applied Principal Component Analysis (PCA) for data preprocessing.
• Trained and tested linear and cubic Support Vector Machine (SVM) classification models.

Main Results:

• Achieved high identification capability for isotope pairs (e.g., 12,13C, 36,40Ar, 80,84Kr).
• The cubic SVM model demonstrated particularly high identification capability.
• The proposed method offers improved computational efficiency compared to prior techniques.

Conclusions:

• The developed DPSA approach effectively discriminates challenging isotope pairs.
• This method supports the integration of advanced analysis into edge-computing systems for nuclear physics research.