Preprint “Automated Charge Transition Detection in Quantum Dot Charge Stability Diagrams”and SimCATS Datasets available
The preprint of our paper entitled “Automated Charge Transition Detection in Quantum Dot Charge Stability Diagrams” by Fabian Hader, Fabian Fuchs, Sarah Fleitmann, Karin Havemann, Benedikt Scherer, Jan Vogelbruch, Lotte Geck, and Stefan van Waasen is now available.
A scalable quantum computing platform requires the automation of the quantum dot tuning process. One crucial step is to trap the appropriate number of electrons in the quantum dots, which is typically accomplished by analyzing charge stability diagrams (CSDs), in which charge transitions manifest as edges. Hence, it is necessary to recognize these edges automatically and reliably. Machine learning methods for this purpose require large amounts of data for training and testing. Therefore, we introduced SimCATS (published in IEEE TQE, https://doi.org/10.1109/TQE.2024.3445967), which is a new approach for the realistic simulation of such data. This enables us to investigate possible edge detection methods, train them withsimulated data, and carry out quantitative and qualitative comparisons on simulated and experimentally measured data from a GaAs and a SiGe qubit sample.
In this study, we compared a large number of classical and machine learning algorithms from different categories. Our focus was on analyzing suitable approaches for future cryogenic hardware implementations. Therefore, we analyzed not only the charge transition detection capabilities but also the size, speed, and number of floating point operations of the machine learning networks. Furthermore, we demonstated the generalizability to experimental data of machine learning networks trained with only the simulated data. Finally, we identified promising approaches for cryogenic hardware implementations and point out future research directions.
Curious?
https://doi.org/10.36227/techrxiv.172963185.53119182/v1
SimCATS Datasets available
You may have come across our simulation framework, SimCATS (Paper: https://doi.org/10.1109/TQE.2024.3445967, Code: https://github.com/f-hader/SimCATS), or perhaps our recent post regarding the preprint of our paper on “Automated Charge Transition Detection in Quantum Dot Charge Stability Diagrams” (https://doi.org/10.36227/techrxiv.172963185.53119182/v1).
If your research also includes the analysis of charge stability diagrams (CSDs), then we have great news for you: Our Python package SimCATS-Datasets (https://github.com/f-hader/SimCATS-Datasets) provides an even more straightforward way to use our simulation for your research! The package simplifies the creation and loading of large SimCATS datasets. Its flexible implementation of a PyTorch dataset class allows users to directly use the simulated data for machine learning purposes.
We believe that shared datasets are essential to provide reliable benchmarks between different research groups. To this end, we have published the evaluation dataset used in our paper: https://doi.org/10.5281/zenodo.13903285