guest ::
| Home > Publications database > Sensor Scan Simulation for Automated Tuning of Gate-Defined Semiconductor Quantum Dots |
| Home > Publications database > Sensor Scan Simulation for Automated Tuning of Gate-Defined Semiconductor Quantum Dots |
| Poster (Other) | FZJ-2026-01869 |
; ; ; ; ; ;
2026
This record in other databases:
Please use a persistent id in citations: doi:10.34734/FZJ-2026-01869
Abstract: Precise tuning of semiconductor quantum dots is essential for their operation as qubits. During this process, sensor scans measure the response of the sensor dots coupled to quantum dots, revealing optimal operating conditions. Because experimental data are time-intensive to obtain and label, progress in data-driven tuning methods has been limited. Simulated sensor scans provide a controllable and reproducible framework for developing, testing, and benchmarking a wide range of tuning algorithms under well-defined conditions.To address this need, we developed a simulation algorithm that generates realistic sensor scans. The algorithm was implemented as a modular extension of the SimCATS framework. Using empirical and phenomenological modeling, it reproduces the sensor’s response to gate voltages based on the corresponding lever arms, which depend on the gate layout. The sensor response is modeled through an equivalent circuit, where the two barriers and the sensor dot are represented as three resistors in series. The resulting simulated scans reproduce characteristic features such as Coulomb peaks. Furthermore, the simulation incorporates configurable distortions such as white noise, pink noise, random telegraph noise, and peak deformations to enhance realism. The algorithm will be released as an open-source Python package on GitHub and made easily accessible for researchers via PyPI.The simulation parameters can be defined manually by the user or automatically generated through built-in samplers that produce randomized parameter sets tailored to a specific setup. For instance, a dedicated sampler for GaAs/AlGaAs heterostructures is included. Such samplers enable the rapid generation of large, labeled, and diverse datasets for training and evaluating automated tuning algorithms. The algorithm’s ability to replicate experimental measurements was evaluated by generating datasets that mimic the sensor behavior observed in GaAs/AlGaAs heterostructures. Parameter-extraction methods were developed and applied to experimental data to obtain realistic simulation inputs. The resulting simulated datasets were compared with actual experimental measurements to validate the simulation’s accuracy. The simulated datasets were evaluated using quantitative and qualitative methods.By providing labeled and realistically distorted sensor scans, our work enables the development and validation of automated tuning algorithms. The simulated data are currently used to develop and test algorithms that automatically identify the Coulomb oscillation area in sensor scans. These algorithms can afterward be applied and tested on real experimental data. This demonstrates how the simulation contributes to the advancement of scalable quantum computing.
|
The record appears in these collections: |
PDF