In this video, we map the charge tuning problem on a neural network with feedback. We show how to classify small patches of charge stability diagram and embed this classification within a Python feedback loop that allows us to navigate the charge states.
- Classify the type of charge transition, not a charge state.
- Use the output of neural network as an input for shifting function that counts the charge transition.
What are the stages of the charge tuning we learned about?
a. Tuning to (0,0) followed by tuning to (m,n)
b. Tuning to (m-1,n-1) followed by tuning to (m-2, n-2)
c. Tuning charge state (m) followed by tuning charge state (n)
d. Tuning charge state (n) followed by tuning charge state (m)
Automated tuning of double quantum dots into specific charge states using neural networks: https://arxiv.org/abs/1912.02777
An example of automated tuning of a quantum dots device – from an unknown charge state into a pre-defined charge state, using a neural network.
Computer-automated tuning of semiconductor double quantum dots into the single-electron regime: https://arxiv.org/abs/1603.02274
Another example of automated tuning of the quantum dots device – from double quantum to single electron regime.
Automated tuning of inter-dot tunnel couplings in quantum dot arrays: https://arxiv.org/abs/1803.10352
Example of auto-tuning in a quantum dot device for the tunnel couplings via tuning the barrier gate voltages.
A Machine Learning Approach for Automated Fine-Tuning of Semiconductor Spin Qubits: https://arxiv.org/abs/1901.01972
On the auto-tuning of gate voltages in quantum dot devices using machine learning algorithms.
Efficiently measuring a quantum device using machine learning: https://arxiv.org/abs/1810.10042
On the automated tuning of the measurements on quantum dot devices using machine learning algorithms.
Machine Learning techniques for state recognition and auto-tuning in quantum dots: https://arxiv.org/abs/1712.04914
On the auto-tuning of states and charge configurations of single and double quantum dot arrays using deep and convolutional neural networks.