Skip to content
DraconiticResearch
Search Sources API
226,742 studies31 sourcesUpdated 34 min ago
Peer-reviewedCC BY-NCNewton

Synthetic data training for machine-learning-assisted ARPES experiments

  1. 1Department of Physics & Astronomy, University of British Columbia, Vancouver, BC V6T 1Z1, Canada
  2. 2Quantum Matter Institute, University of British Columbia, Vancouver, BC V6T 1Z4, Canada

Abstract

Angle-resolved photoemission spectroscopy (ARPES) now generates increasingly rich, high-dimensional datasets, making sample assessment and data acquisition more demanding. Machine learning (ML) could accelerate these tasks, but its adoption is limited by the scarcity of large, labeled ARPES datasets. Here, we introduce aurelia , an open-source simulator that generates realistic synthetic ARPES spectra for ML training and benchmarking. As proof of principle, we train a convolutional neural network to evaluate spectral quality during sample alignment. Benchmarking against experimental Bi 2 Se 3 grid-scan data shows that the model trained on synthetic spectra accurately identifies the optimal measurement region and captures spectral quality variations with high sensitivity. Our results establish that synthetic ARPES spectra can train useful ML models for real experimental workflows, opening a practical route to ML-assisted ARPES experiments.

Related

  1. Training of Neural Network Machine Learning Potentials for Activation Barrier Calculations in Predicting Growth Modes of Deposited Metals
    Karl Rönnby, Cara-Lena Nies, Savannah May Timony +2
    ChemRxivPreprintLicence not statedGeneral chemistryChemRxiv
  2. Understanding the Relationship Between Synthetic Conditions and H ₂ Uptake Capacity in Oxidatively Activated Turbostratic Carbons through Machine Learning
    L Scott Blankenship, Filippo Federici Canova, David Gao +1
    ChemRxivPreprintCC BYFull textMaterials & polymersChemRxiv
  3. Machine Learning-Assisted Analysis of Chloride-Enhanced BiOCl/DSA Anode for Synergistic Denitrification and Tetracycline Degradation: Performance, Mechanisms, and Product Regulation
    Nanomaterials (Basel, Switzerland)Peer-reviewedCC BYFull textMaterials & polymersEurope PMC
  4. Growth driven phase transitions in Zinc Oxide nanoparticles through machine-learning assisted simulations
    Quentin Gromoff, Magali Benoit, Jacek Goniakowski +2
    NanoscalePeer-reviewedPublisher termsFull textMaterials & polymersHAL
  5. Hydrothermal-Assisted Extraction of Cinnamomum camphora Essential Oil Optimized by Machine Learning
    Yancai Qi, Xiangzhou Li, Jun Zhou +2
    Journal of Oleo SciencePeer-reviewedLicence not statedFull textGeneral chemistryJ-STAGE
BibTeX
@article{na2026synthetic,
  title = {{Synthetic data training for machine-learning-assisted ARPES experiments}},
  author = {Meng Xing Na and Chris Zhou and S. K. Y. Dufresne and Matteo Michiardi and A. Damascelli},
  journal = {Newton},
  year = {2026},
  month = oct,
  doi = {10.1016/j.newton.2026.100716},
  url = {https://doi.org/10.1016/j.newton.2026.100716},
  note = {CC BY-NC},
}
RIS
TY  - JOUR
TI  - Synthetic data training for machine-learning-assisted ARPES experiments
AU  - Meng Xing Na
AU  - Chris Zhou
AU  - S. K. Y. Dufresne
AU  - Matteo Michiardi
AU  - A. Damascelli
PY  - 2026
DA  - 2026/10/01
JO  - Newton
DO  - 10.1016/j.newton.2026.100716
UR  - https://doi.org/10.1016/j.newton.2026.100716
AB  - Angle-resolved photoemission spectroscopy (ARPES) now generates increasingly rich, high-dimensional datasets, making sample assessment and data acquisition more demanding. Machine learning (ML) could accelerate these tasks, but its adoption is limited by the scarcity of large, labeled ARPES datasets. Here, we introduce aurelia , an open-source simulator that generates realistic synthetic ARPES spectra for ML training and benchmarking. As proof of principle, we train a convolutional neural network to evaluate spectral quality during sample alignment. Benchmarking against experimental Bi 2 Se 3 grid-scan data shows that the model trained on synthetic spectra accurately identifies the optimal measurement region and captures spectral quality variations with high sensitivity. Our results establish that synthetic ARPES spectra can train useful ML models for real experimental workflows, opening a practical route to ML-assisted ARPES experiments.
SN  - 2950-6360
ER  -