Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userMiniaturizing neural networks for charge state autotuning in quantum dots
280
0
0.0
(
0
)
Ask ChatGPT about the research
No Arabic abstract
A key challenge in scaling quantum computers is the calibration and control of multiple qubits. In solid-state quantum dots, the gate voltages required to stabilize quantized charges are unique for each individual qubit, resulting in a high-dimensional control parameter space that must be tuned automatically. Machine learning techniques are capable of processing high-dimensional data - provided that an appropriate training set is available - and have been successfully used for autotuning in the past. In this paper, we develop extremely small feed-forward neural networks that can be used to detect charge-state transitions in quantum dot stability diagrams. We demonstrate that these neural networks can be trained on synthetic data produced by computer simulations, and robustly transferred to the task of tuning an experimental device into a desired charge state. The neural networks required for this task are sufficiently small as to enable an implementation in existing memristor crossbar arrays in the near future. This opens up the possibility of miniaturizing powerful control elements on low-power hardware, a significant step towards on-chip autotuning in future quantum dot computers.
rate research
Read More
While quantum dots are at the forefront of quantum device technology, tuning multi-dot systems requires a lengthy experimental process as multiple parameters need to be accurately controlled. This process becomes increasingly time-consuming and difficult to perform manually as the devices become more complex and the number of tuning parameters grows. In this work, we present a crucial step towards automated tuning of quantum dot qubits. We introduce an algorithm driven by machine learning that uses a small number of coarse-grained measurements as its input and tunes the quantum dot system into a pre-selected charge state. We train and test our algorithm on a GaAs double quantum dot device and we consistently arrive at the desired state or its immediate neighborhood.
We investigate phonon-induced spin and charge relaxation mediated by spin-orbit and hyperfine interactions for a single electron confined within a double quantum dot. A simple toy model incorporating both direct decay to the ground state of the double dot and indirect decay via an intermediate excited state yields an electron spin relaxation rate that varies non-monotonically with the detuning between the dots. We confirm this model with experiments performed on a GaAs double dot, demonstrating that the relaxation rate exhibits the expected detuning dependence and can be electrically tuned over several orders of magnitude. Our analysis suggests that spin-orbit mediated relaxation via phonons serves as the dominant mechanism through which the double-dot electron spin-flip rate varies with detuning.
Semiconductor quantum dots provide a two-dimensional analogy for real atoms and show promise for the implementation of scalable quantum computers. Here, we investigate the charge configurations in a silicon metal-oxide-semiconductor double quantum dot tunnel coupled to a single reservoir of electrons. By operating the system in the few-electron regime, the stability diagram shows hysteretic tunnelling events that depend on the history of the dots charge occupancy. We present a model which accounts for the observed hysteretic behaviour by extending the established description for transport in double dots coupled to two reservoirs. We demonstrate that this type of device operates like a single-electron memory latch.
Spin qubits in silicon quantum dots offer a promising platform for a quantum computer as they have a long coherence time and scalability. The charge sensing technique plays an essential role in reading out the spin qubit as well as tuning the device parameters and therefore its performance in terms of measurement bandwidth and sensitivity is an important factor in spin qubit experiments. Here we demonstrate fast and sensitive charge sensing by a radio-frequency reflectometry of an undoped, accumulation-mode Si/SiGe double quantum dot. We show that the large parasitic capacitance in typical accumulation-mode gate geometries impedes reflectometry measurements. We present a gate geometry that significantly reduces the parasitic capacitance and enables fast single-shot readout. The technique allows us to distinguish between the singly- and doubly-occupied two-electron states under the Pauli spin blockade condition in an integration time of 0.8 {mu}s, the shortest value ever reported in silicon, by the signal-to-noise ratio of 6. These results provide a guideline for designing silicon spin qubit devices suitable for the fast and high-fidelity readout.
Automated tuning of gate-defined quantum dots is a requirement for large scale semiconductor based qubit initialisation. An essential step of these tuning procedures is charge state detection based on charge stability diagrams. Using supervised machine learning to perform this task requires a large dataset for models to train on. In order to avoid hand labelling experimental data, synthetic data has been explored as an alternative. While providing a significant increase in the size of the training dataset compared to using experimental data, using synthetic data means that classifiers are trained on data sourced from a different distribution than the experimental data that is part of the tuning process. Here we evaluate the prediction accuracy of a range of machine learning models trained on simulated and experimental data and their ability to generalise to experimental charge stability diagrams in two dimensional electron gas and nanowire devices. We find that classifiers perform best on either purely experimental or a combination of synthetic and experimental training data, and that adding common experimental noise signatures to the synthetic data does not dramatically improve the classification accuracy. These results suggest that experimental training data as well as realistic quantum dot simulations and noise models are essential in charge state detection using supervised machine learning.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
