The purpose of the workshop is to convene stakeholders from industry, academia, and the government interested in the research and development of semiconductor quantum computing technologies. Topics to be discussed include opportunities for research and development of tuning, characterization, and control methods for semiconductor quantum dot devices, the need for facilitating interaction and collaboration between the stakeholders to build a large open-access database of experimental and simulated data for benchmarking new machine learning algorithms, determining key performance metrics for the various aspects of the tuning, characterizing, and controlling of quantum dot devices, and identifying barriers to near-term and future applications of the auto-tuning methods. Furthermore, this workshop will provide a discussion space to consider methods of collaboration in a neutral setting and develop a roadmap for the future of tuning large-scale devices.

Workshop objectives

1. Assess the availability and accessibility of experimental datasets for semiconductor quantum dot device automation and identify gaps or barriers that limit their effective use in advancing research.
2. Explore emerging frontiers in automation, with emphasis on how datasets can be leveraged to benchmark new approaches and reveal unmet research needs.
3. Examine the role of simulating and modeling quantum dot systems, focusing on how simulated datasets can complement experimental data to accelerate automation and enable scalability.
4. Investigate approaches for building digital twins to close the loop between experimental data, modeling, and automation workflows, identifying the types of datasets and simulations required to make digital twins robust and broadly applicable.
5. Evaluate strategies for enabling edge-level hardware automation and discuss challenges in integration, performance, and deployment.
6. Describe what role NIST can play in developing infrastructure that supports the use of large-scale datasets for research on tuning quantum dot devices

9:00 am - 9:15 am : Welcome and opening remarks

9:15 am - 10:50 am : Scientific Session
Chair: Jesus Cifuentes

• 9:15 am - 9:50 am: Anthony Sigillito (University of Pennsylvania)
  Noise and Knobs: Challenges in Autotuning Quantum Gates

  Compared to other quantum technologies, quantum-dot-based spin qubits show great potential for scalability. They offer high readout visibility, competitive control fidelities, and a scalable semiconductor-based platform. Moreover, the many degrees of freedom in tuning up a lithographically defined quantum dot device enable varied qubit encodings and a rich set of primitive gates. These primitives may enable very efficient circuit-level optimization when running quantum algorithms. However, these many degrees of freedom come at the cost of increased susceptibility to noise and a control complexity that necessitates automated device tuning and calibration as processors scale to even a modest number of qubits. In this talk, I will discuss the steps required to tune up single- and two-qubit quantum gates in a silicon quantum processor. I will describe early efforts to automate the recalibration of single-qubit gates and identify key challenges in the automated tune-up of multi-qubit gates and larger qubit arrays.

• 9:50 am - 10:10 am: Pau Dietz Romero (Forschungszentrum Jülich)
  Applying Shuttling Optimal Control to Cryogenic Electronics

  Spin qubit shuttling is necessary for creating scalable quantum processors based on sparse qubit grid architectures. The SpinBus architecture [1] is one example of such a sparse grid that allows for the integration of tailored cryogenic electronics. Generating shuttling signals in close proximity to qubits mitigates the wiring bottleneck and optimizes the qubit-electronics interface in terms of cost, power, and signal distortion. However, it also increases the demand for co-simulation and pre-silicon verification tools and methodologies, as described in [2].

  In particular, the adoption of optimal control theory in cryogenic electronics necessitates cross-domain collaboration between quantum physicists and integrated circuit designers. Our digital twin, which represents a shuttling channel and integrated control circuit, can accelerate collaborative development. The digital twin co-simulates the integrated circuit using a behavioral or transistor-level model. It also uses a COMSOL model to calculate the electrostatic potential landscape induced by the electrical signals applied to the electrodes and a quantum solver to calculate the resulting spin shuttling fidelity. This interdisciplinary model encompasses most limitations associated with cryogenic electronics, such as a reduced number of waveform shapes, and reduced time and amplitude resolution.

  In the two-dimensional heterostructure of the SpinBus architecture, the shape and disorder of the quantum-well confinement, influenced by alloy diffusion, generate an unpredictable evolution of the electron's valley state during shuttling. In addition to this, the different g factors associated with the valley eigenstates lead to entanglement between valley and spin state, deteriorating the purity of the qubit spin and causing dephasing of the spin state. This work describes an investigation into applying a valley-aware shuttling strategy, as described in [3], onto an existing integrated circuit design of an ultra-low-power cryogenic shuttling signal generator. Due to limitations in measuring device-specific valley maps, we use artificial valley map datasets generated using a diffusion model from [4]. Using these simulated valley maps, we then apply an optimizer that tunes the pulse shapes in the digital twin depending on the valley map experienced by the electron.

  We advocate for this approach by demonstrating its feasibility through modeling and artificial datasets. Tuning involves adjusting the parameters of the electronic circuit offered by the cryogenic shuttling chip to achieve high shuttling fidelity by adjusting the pulses of the shuttling signal generator. We plan to run the modeled shuttling control on our real cryogenic chip integrated with a qubit to verify the effectiveness of the digital twin.

  [1] M. Künne, et al., Nat. Commun. 15, 4977 (2024).
  [2] P. Dietz Romero, et al., Proc. SMACD 2025, 1-4 (2025).
  [3] A. David, et al., Preprint arXiv:2409.07600v1 (2024).
  [4] Paquelet Wuetz, B., Losert, M. P., et al., Nat Commun 13, 7730 (2022)
  
  
• 10:10 am - 10:30 am: Shize Che (University of Pennsylvania)
  NeuroQD: A Learning-Based Simulation Framework For Quantum Dot Devices

  Spin qubits in semiconductor quantum dots are promising for scalable quantum computing, yet progress in control system development is limited by the lack of simulators that are both realistic and fast enough to interface with control systems in real-time. Physics-based solvers [1-5] offer high accuracy but are too slow for real-time use. Faster, more qualitative models [6-9] ignore gate layout and heterostructure details and struggle to capture device behavior from a cold-start.

  Here, we present NeuroQD, a learning-based simulation and emulation framework. NeuroQD interfaces with control software in real time and given a set of gate voltages, returns the device response with millisecond-scale latency. Moreover, NeuroQD can simulate device behavior in the grossly mistuned and cold-start regimes. Our method leverages an empirical property observed in physics-based workflows, enabling the training of a compact convolutional neural network (CNN) to model the 2DEG electrostatic potential from gate voltages with high accuracy and throughput. On GPU, NeuroQD delivers >1000× speedup while maintaining >96% agreement compared to our high-accuracy, physic-based baseline (COMSOL). Interfaced with the experimental control stack, the simulator reproduces key tuning features and yields device metrics and behavior consistent with measurements on real-world devices. Together, these results provide a practical path to safe, fast, and realistic in situ development, and testing of quantum control hardware and software.

  [1] COMSOL Multiphysics v6.2, COMSOL AB, Stockholm, Sweden (2024).
  [2] T. Ikegami, et al., J. Comput. Electron. 18(2), 534-542 (2019).
  [3] S. Birner, et al., IEEE Trans. Electron Devices 54(9), 2137-2142 (2007).
  [4] X. Gao, el al., J. Appl. Phys. 114(16), 164302 (2013).
  [5] QTCAD: A Computer-Aided Design Tool for Quantum-Technology Hardware, Nanoacademic Technologies Inc., Montreal, Canada (2025).
  [6] V. Gualtieri, et al., SciPost Phys. Codebases 46 (2025).
  [7] J. A. Krzywda, et al., SciPost Phys. Codebases 43 (2024).
  [8] B. van Straaten, et al., SciPost Phys. Codebases 35 (2024).
  [9] F. Hader, et al., IEEE Trans. Quantum Eng. 5, 1-14 (2024).
  
  
  
• 10:30 am - 10:50 am: Tara Murphy (Quantum Motion)
  RF-Squad: A radiofrequency simulator for quantum dot arrays

  Spins in semiconductor quantum dots represent a large-scale integration route for quantum computing hardware. Many efforts have been devoted to creating quantum dot array simulators, with an aim to better understand the complexity of these systems as they scale up. Typically based on the well-established Constant Interaction Model (CIM), they often fail to simulate realistic datasets or recreate important experimental details, such as the set up used to measure the device.

  We present RF-Squad, a physics-based quantum dot array simulator capable of realistically replicating the outputs of radiofrequency reflectometry measurements. Implemented in JAX, an accelerated linear algebra library, RF-Squad facilitates the simulation a double quantum dot in milliseconds at the CIM level.

  A key feature of RF-Squad is the inclusion of advanced physical phenomena, such as the use of tunnel couplings, the WKB approximation, Fock-Darwin states, different noise models and tunnel rates. Furthermore, the code has been optimized and structured in a layered approach, giving users the flexibility to balance realism and computational speed. In summary, RF-Squad enables efficient generation of large, realistic charge stability diagram datasets, and serves as a practical tool for researchers to gain a deeper understanding of experimental data. Finally, with its user-friendly design, RF-Squad is accessible to a broad range of users, making it easy to integrate across a wide range of experimental contexts.

10:50 am - 11:20 am : Coffee break

11:20 am - 12:40 pm : Scientific Session
Chair: Krishna Choudhary

• 11:20 am - 11:40 pm: Tyler J. Kovach (University of Wisconsin-Madison)
  Scalable Autotuning of High-Temperature Quantum Dot Spin Qubits

  As quantum dot (QD) spin qubits are beginning to deploy in small arrays [1], achieving robust and scalable autotuning-particularly at elevated temperatures-remains a formidable challenge. A major hurdle arises from trapped charges within the oxide layers, which induce random offset voltage shifts on gate electrodes, with magnitudes reaching approximately 650mV in state-of-the-art devices[1]. While partial mitigation techniques exist, they typically necessitate prior device characterization [2].

  In this talk, we introduce a streamlined, five-step physically intuitive framework for initializing and bootstrapping QD devices [3]. The most intricate step involves understanding the interaction between screening and finger gates during the independent formation of conduction channels. To dissect this relationship, we employ coupled Schrödinger-Poisson simulations integrated with current continuity analyses, enabling us to identify optimal operating regimes within the device's high-dimensional gate voltage space.

  These insights are synthesized into a unified autotuning protocol, allowing for fully autonomous calibration and characterization. We demonstrate this methodology experimentally at 1.3K using an autotuner-BATIS (Bootstrapping Autonomously Testing Initialization System)-to configure a quad-QD Si/SiGe heterostructure device.

  We will also discuss our development of an open-source software platform designed to facilitate the deployment and scaling of autotuning algorithms. This platform-agnostic approach addresses a critical bottleneck in quantum dot scalability, paving the way for the broader implementation of large QD arrays.

  [1] S. Neyens, et. al, Probing single electrons across 300-mm spin qubit wafers, Nature 629, 80 (2024).
  [2] M. Meyer, et. al, Single-electron occupation in quantum dot arrays at selectable plunger gate voltage, NanoLetters 23, 11593 (2023).
  [3] T. Kovach, et. al, BATIS: Bootstrapping, Autonomous Testing, and Initialization System for Quantum Dot Devices, arxiv: 2412.07676 (2024).
  
  
  
• 11:40 pm - 12:00 pm: Fabian Hader (Forschungszentrum Jülich GmbH)
  Towards Scalable Robust Charge Transition Detection for Quantum Dot Devices

  Reliable detection of charge transitions in charge stability diagrams (CSDs) is a key requirement for the full automation of quantum dot device control. Performing this task directly at the cryogenic stage reduces data transfer and supports scalability. To provide the large labeled datasets required for developing and evaluating detection methods, we introduced Sim-CATS [1], a simulator that generates realistic CSDs including sensor responses and distortions. We optimize both traditional and machine-learning-based detection methods using simulated data and benchmark them on simulated and experimental measurements from GaAs and SiGe qubit devices. We also investigate the potential of model compression and find its performance closely tied to task complexity, which can be alleviated by sensor dot compensation. In fact, we find that sensor compensation allows machinelearning approaches to be reduced in size by up to two orders of magnitude while maintaining, or even improving, detection quality. Together with high-quality measurements, this enables robust and scalable (ray-based) charge transition detection. Finally, we estimate the cryogenic power budget for applying this approach to large-scale systems with up to one million qubits.

  [1] F. Hader et al., "Simulation of Charge Stability Diagrams for Automated Tuning Solutions (SimCATS)", IEEE Transactions on Quantum Engineering, DOI: 10.1109/TQE.2024.3445967 (2024).
  
  
• 12:00 pm - 12:20 pm: Han Na We (Forschungszentrum Jülich GmbH and Aachen University)
  Efficient Quantum Dot Tuning with Ray-Based Algorithms

  As quantum processors scale to larger numbers of qubits, efficient and reliable tuning of quantum dots has become a critical challenge. To address this, we have developed automated algorithms that we integrated into our Tuning Toolkit Framework (TTF).Within TTF, ray-based tuners have demonstrated the ability to tune QuBus devices from the depleted state to the first electron transition in the single quantum dot (SQD) regime without human intervention, achieving success rates above 75% within 30 minutes. Compared to the tuners that required twodimensional data for inspection, ray-based tuners required substantially fewer measurements, reducing runtime, including measurement time and data analysis, by at least a factor of two. Additionally, bootstrapping tuners, which perform accumulation and sensor coarse tuning, have been benchmarked on the Cryogenic Wafer Prober (CWP), achieving success rates of approximately 90%. A dedicated sensor fine-tuner further stabilized sensors, transforming disordered regimes into clean Coulomb oscillations in under half an hour. Our findings highlight ray-based tuning as a resource-efficient and scalable approach to quantum dot control, enabling faster optimization and paving the way toward large-scale device operation.

12:20 pm - 1:20 pm : Lunch break

1:20 pm - 2:35 pm : Scientific Session
Chair: Fabio Ansaloni

• 1:20 pm - 1:40: Joel Pendleton (Conductor Quantum)
  Cloud API toolkit for tuning quantum dot devices

  Quantum dot automation has traditionally relied on bespoke heuristic tools and machine learning models tailored to individual device architectures and noise profiles, limiting generalizability. With the advent of digital twins and wafer-scale characterization, the availability of diverse training and test data is rapidly increasing. We present a suite of core analysis and characterization tools that operate across device architectures and are instantly accessible via the cloud. This is the first cloud-based API designed specifically for automatic quantum dot device tune-up. We benchmark our models against experimental data labeled by human experts and outperform previous characterization methods.

• 1:40 pm - 2:00 pm: Joost van der Heijden (Quantum Machines)
  Control hardware with automated calibration software for quantum dots

  Stand-alone tuning and stabilization protocols for spin qubits in semiconductor quantum dots, often based on physics-informed machine-learning or Hamiltonian estimating algorithms, have been successfully implemented, see for instance [1,2]. Connecting these algorithms together in a user-friendly and time-efficient manner is key to reach effective initialization and calibration protocols for the spin-based quantum processors of the future. Especially as scaling to large arrays introduces major calibration challenges: resonance frequencies drift with charge noise, two-qubit couplings vary with cross-capacitance, and fidelities degrade if parameters are not continuously optimized [3].

  A software package that can implement the completely automated, and adaptive tuning and calibration graphs is therefore essential. The QUAlibrate software package from Quantum Machines [4] provides modular pipelines for tuning and calibrating qubits, while keeping track of all essential qubit parameters. The already lab-proven algorithms can easily be connected together in a QUAlibrate workflow, allowing the efficient tune-up and calibration of quantum dot arrays. With interactive execution, live visualization through a video mode tool, and robust data management, it is straightforward to keep track of the performance of separate algorithms as well as the combination of them.

  Integration with DGX Quantum [5], developed by NVIDIA and Quantum Machines, extends these capabilities. GPU acceleration allows machine learning algorithms to be embedded into calibration loops, where neural estimators and reinforcement learning agents can be utilized to find the processors operation points and optimize noisy multi-parameter landscapes. Together, QUAlibrate and DGX Quantum provide a scalable, automated framework for the tune-up of spin qubit arrays, addressing one of the primary bottlenecks in advancing toward fault-tolerant, industrial-scale spin-based quantum processors. [1]

  [1] F. Berritta, et al., Nat. Commun. 15, 1676 (2024).
  [2] A. Zubchenko, et al., Phys. Rev. Appl. 23, 014072 (2025).
  [3] J. Yoneda, et al., Nat. Nanotechnol. 13(2), 102 (2018).
  [4] https://www.quantum-machines.co/products/qualibrate/.
  [5] https://www.quantum-machines.co/products/nvidia-dgx-quantum/.
  
  
  
• 2:00 - 2:35: Andrew E Oriani (HRL Laboratories)
  HRL and open-source software for spin-qubits: Future outlook

  To ensure the long-term success of spin-qubit development, accessible software is necessary to address the complications of device control, simulation, and analysis, while also promoting healthy exchange between various parts of the spin qubit design, fabrication, verification, and measurement processes. As a vertically integrated organization and leader in spin-qubit technologies HRL is uniquely positioned to provide software expertise in the creation of open-source, community focused tools to address these problems and promote a healthy domestic spin qubit ecosystem. In this high-level talk, we will discuss the areas which HRL would like to expand offerings and discuss some insights and challenges that must be overcome.

2:35 pm - 3:00 pm : Coffee break

3:00 pm - 4:50 am : Breakout discussions

• 3:00 - 4:10: Breakout discussion
Group 1: Automation frontiers: From capacitance to quantum gates [Chair: Florian Luthi (Intel)]
• Explore emerging frontiers in automation, with emphasis on how datasets can be leveraged to benchmark new approaches and reveal unmet research needs.
• Assess the availability and accessibility of experimental datasets for semiconductor quantum dot device automation and identify gaps or barriers that limit their effective use in advancing research.
Group 2: Closing the loop: Building digital twins and other agentic AI approaches [Chair: Gushu Li (UPenn)]
• Investigate approaches for building digital twins to close the loop between experimental data, modeling, and automation workflows, identifying the types of datasets and simulations required to make digital twins robust and broadly applicable.
• Examine the role of simulation and modeling of quantum dot systems, focusing on how simulated datasets can complement experimental data to accelerate automation and improve scalability.
Group 3: Moving to the edge: Automation in hardware [Stefan van Waasen (Forschungszentrum Juelich)]
• Evaluate strategies for enabling edge-level hardware automation and discuss challenges in integration, performance, and deployment.
• Examine the role of simulation and modeling of quantum dot systems in advancing the near-and on-chip calibration and control.
• 4:10 - 4:20: Summarization period / break
• 4:20 - 4:50: Summary from breakout session chairs

4:50 am - 5:00 am : Closing remarks

• 

  Justyna Zwolak
  UMD

• 

  Jake Taylor
  UMD

• 

  Mark Gyure
  UCLA

• Andrew Oriani, HRL Laboratories, LLC
• Judith Olson, HRL Laboratories, LLC
• Marc Dvorak, HRL Laboratories, LLC

• Quantum Machines
• Conductor Quantum