A5013728933- Advancements in Semiconductor Devices and Circuit Design
- Quantum and electron transport phenomena
- Semiconductor materials and devices
- Machine Learning in Materials Science
- Quantum Computing Algorithms and Architecture
- Diamond and Carbon-based Materials Research
- Neural Networks and Applications
- Integrated Circuits and Semiconductor Failure Analysis
- Neural Networks and Reservoir Computing
University of Oxford
2019-2024
Abstract Scalable quantum technologies such as computers will require very large numbers of devices to be characterised and tuned. As the number on chip increases, this task becomes ever more time-consuming, intractable a scale without efficient automation. We present measurements dot device performed by machine learning algorithm in real time. The selects most informative perform next combining information theory with probabilistic deep-generative model that can generate full-resolution...
Abstract Deep reinforcement learning is an emerging machine-learning approach that can teach a computer to learn from their actions and rewards similar the way humans experience. It offers many advantages in automating decision processes navigate large parameter spaces. This paper proposes efficient measurement of quantum devices based on deep learning. We focus double dot devices, demonstrating fully automatic identification specific transport features called bias triangles. Measurements...
Device variability is a bottleneck for the scalability of semiconductor quantum devices. Increasing device control comes at cost large parameter space that has to be explored in order find optimal operating conditions. We demonstrate statistical tuning algorithm navigates this entire space, using just few modelling assumptions, search specific electron transport features. focused on gate-defined dot devices, demonstrating fully automated two different devices double regimes an up...
The discrepancies between reality and simulation impede the optimization scalability of solid-state quantum devices. Disorder induced by unpredictable distribution material defects is one major contributions to gap. We bridge this gap using physics-aware machine learning, in particular, an approach combining a physical model, deep Gaussian random field, Bayesian inference. This enables us infer disorder potential nanoscale electronic device from electron-transport data. inference validated...
Abstract The potential of Si and SiGe-based devices for the scaling quantum circuits is tainted by device variability. Each needs to be tuned operation conditions each realisation requires a different tuning protocol. We demonstrate that it possible automate 4-gate FinFET, 5-gate GeSi nanowire 7-gate Ge/SiGe heterostructure double dot from scratch with same algorithm. achieve times 30, 10, 92 min, respectively. algorithm also provides insight into parameter space landscape these devices,...
Pauli spin blockade (PSB) can be employed as a great resource for qubit initialisation and readout even at elevated temperatures but it difficult to identify. We present machine learning algorithm capable of automatically identifying PSB using charge transport measurements. The scarcity data is circumvented by training the with simulated cross-device validation. demonstrate our approach on silicon field-effect transistor device report an accuracy 96% different test devices, giving evidence...
The discrepancies between reality and simulation impede the optimisation scalability of solid-state quantum devices. Disorder induced by unpredictable distribution material defects is one major contributions to gap. We bridge this gap using physics-aware machine learning, in particular, an approach combining a physical model, deep Gaussian random field, Bayesian inference. This has enabled us infer disorder potential nanoscale electronic device from electron transport data. inference...
The potential of Si and SiGe-based devices for the scaling quantum circuits is tainted by device variability. Each needs to be tuned operation conditions. We give a key step towards tackling this variability with an algorithm that, without modification, capable tuning 4-gate FinFET, 5-gate GeSi nanowire 7-gate SiGe heterostructure double dot from scratch. achieve times 30, 10, 92 minutes, respectively. also provides insight into parameter space landscape each these devices. These results...
An amendment to this paper has been published and can be accessed via a link at the top of paper.
Abstract Pauli spin blockade (PSB) can be employed as a great resource for qubit initialisation and readout even at elevated temperatures but it difficult to identify. We present machine learning algorithm capable of automatically identifying PSB using charge transport measurements. The scarcity data is circumvented by training the with simulated cross-device validation. demonstrate our approach on silicon field-effect transistor device report an accuracy 96% different test devices, giving...
The equal author contributions statement was not included in the final publication. This has now been included: These authors contributed equally: D. T. Lennon, H. Moon.
Pauli spin blockade (PSB) can be employed as a great resource for qubit initialisation and readout even at elevated temperatures but it difficult to identify. We present machine learning algorithm capable of automatically identifying PSB using charge transport measurements. The scarcity data is circumvented by training the with simulated cross-device validation. demonstrate our approach on silicon field-effect transistor device report an accuracy 96% different test devices, giving evidence...