Time of Flight Detector Denoising and Reconstruction

A deep-learning based online denoising and reconstruction method for time of flight (TOF) detectors in PyTorch.

Examples

A randomly generated spectrogram with simulated detector image

Denoising and reconstruction of sample with two disabled TOF detectors

Quick Start

from evaluation import Evaluator

# Initialize evaluator
e: Evaluator = Evaluator({}, output_dir="outputs/", load_max=0)
# Create spectrogram with 3 peaks with seed 12 and simulate detector image
e.plot_spectrogram_detector_image(3, 12)

# Initialize evaluator with General model
model_dict = {
     "General model": "outputs/tof_reconstructor/hj69jsmh/checkpoints"}
e: Evaluator = Evaluator(model_dict, output_dir="outputs/", load_max=10)

# Disable TOF detectors on position #8 and #13
e.plot_reconstructing_tofs_comparison([7, 12], "General model")

Citation

If you find this useful in your research, please consider citing:

Reconstructing Time-of-Flight Detector Values of Angular Streaking Using Machine Learning.

@misc{meier2025reconstructingtimeofflightdetectorvalues,
  title={Reconstructing Time-of-Flight Detector Values of Angular Streaking Using Machine Learning}, 
  author={David Meier and Wolfram Helml and Thorsten Otto and Bernhard Sick and Jens Viefhaus and Gregor Hartmann},
  year={2025},
  eprint={2501.08966},
  archivePrefix={arXiv},
  primaryClass={physics.data-an},
  url={https://arxiv.org/abs/2501.08966}, 
}