Intelligent autonomy for frictionless quantum device bring-up

The proliferation of quantum devices and quantum computing workloads has begun. With more devices accessible than ever before, and viable workloads accelerating toward quantum advantage, the need for abstraction and consistency is key for broader adoption of this transformational technology.

Quantum systems need to be operated consistently at high performance over extended periods, by quantum and non-quantum users alike. For the next phase of the quantum computing industry to take place, a new set of quantum infrastructure tools is required to better enable this technology for a continuously growing and more diverse user base. 

Within this context, the underlying quantum technology continues to change. Devices are getting bigger and more complex, and they're being accessed by a broader range of users. This means the old manual techniques to test devices and tune up performance won't work moving forward. The process is too complex and unwieldy, leading to long, wasteful process times with suboptimal results.

To overcome these dynamics, we need abstraction so any user can take advantage of the technology and we need consistency so you get the same answer, every time. But simple mechanical, linear automation isn’t enough when devices growing in size and complexity also embody the stochastic nature of the quantum realm. Autonomous solutions are needed.

At Q-CTRL, we enable anyone to achieve consistent, repeatable, and high-performance results through software so they can get the most out of cutting-edge quantum computing hardware. We have combined our deep quantum control expertise with our pioneering efforts to deploy new forms of numerical optimization and artificial intelligence (AI) to solve the most pressing challenges in the control of quantum systems at scale. The result? Boulder Opal Scale Up, which brings true intelligent autonomy to any quantum user in a simple package for tuning up quantum computers.

Understanding intelligent autonomy

Quantum device characterization and calibration are some of the most common and important tasks required to build and use quantum computers. Simply put, characterization is the process of defining a specific quantum device by quantifying key system parameters (like resonant frequencies and anharmonicity values). Calibration then uses that data, combined with waveform definitions and pulse shaping, to create the optimal quantum operators for controlling said device. Together, those processes are foundational for preparing a quantum computer for eventual use. 

For years, researchers and hardware providers have used our quantum control toolkit to address these tasks. Boulder Opal provides everything you need to model and optimize quantum systems in a simulation space, and then experiment on and tune those same designs in the real world. These tools have helped power the broader R&D effort of quantum computing development through manual characterization and calibration efforts. For example, by helping Chalmers achieve 8X faster quantum logic or Nord Quantique accelerate the path to useful quantum error correction with 15% longer qubit lifetimes.

We now need to move forward with more consistent solutions and abstracted operations to bring technologies from the lab into the hands of a broader audience. 

Boulder Opal Scale Up builds on this quantum control toolkit to create autonomous device tuning that delivers results for organizations that would rather not build, maintain, and debug everything anew. Due to the pace of innovation with bigger, more complex hardware, complete autonomous software is the only way to create the fast-paced, reliable experience that is necessary to keep up. That is why we designed Boulder Opal Scale Up to handle all these tasks for you. The experiment process is encoded into an autonomous routine that runs on devices and completes the desired characterization or calibration tasks. 

But don’t be fooled, this is not as easy as a simple, linear automated script. “Irregularity is the regular” in quantum computing, and in today’s manual approach, they lead to exceptions that cause calibration times to blow up and the resulting quantum performance to be diminished. Instead, Boulder Opal Scale Up is the only solution available to manage that complexity through intelligent autonomy. The software effectively applies universal analytics and AI concepts into a quantum application that eliminates the need for a human-in-the-loop, leading to fast, reliable, and higher-performant device tuning.

Why autonomy over automation?

Automation simply implies the act of self-direction. Automation follows a set of rules to complete a task in an ordinal way. For example, cruise control is a form of automation. If a car is below the set speed point, it accelerates. Even when adding in adaptive cruise control with lane assist, the automation maintains a simple set of linear rules – when to speed up, when to break, and when to turn. But these rules don’t tell the car where to go. They don’t get you, the passenger, from point A to Point B. For that, you need autonomous driving, which uses a more complex system of inputs and outputs to actually drive the car to the intended destination. That is what we need in quantum computing – an autonomous method of achieving high-performance results.  

Automation can be applied to run a series of experiments one after the other, which is the basic definition of a characterization or calibration process. However, each of these experiments has to deal with the fickleness of quantum uncertainty. Experiments require preset parameters to run, and we can certainly make educated assumptions about what those parameters should be. However, device variability, environmental noise, drift over time, and even human error, mean the parameters that work one time, might not work the next. Linear rules are not meant to operate outside of their narrowly defined scope. Going back to the driving analogy, adaptive cruise control will slow down if a car breaks in front of you, but it will not help you navigate unplanned highway construction. That would require driver intervention.  

Similarly today, quantum deviations are overcome by manual intervention and debugging, which are slow and subjective. Instead, new techniques are required to help quantum routines run on their own, quickly, without interruption, and resulting in peak hardware performance.

Akin to autonomous driving which is a collective system of techniques to drive a car, Boulder Opal Scale Up uses a network of analytics and AI systems to tune a quantum device. Two key examples of such tools are the robust error handling it uses to effectively navigate the experimental space, and the closed-loop optimizers it uses to converge on peak quantum performance. Together, these technologies realize the promise of intelligent autonomy, making it the fastest, most reliable, and highest-performing quantum calibration solution available. 

Reliable execution with robust exception handling

Robust exception handling is paramount to Boulder Opal Scale Up’s autonomy. The experimental steps to characterization and calibration are known - the trick is to sequence those routines and logically connect them in a robust way that overcomes any deviations from the expected results. Humans can do this manually quite easily.

For example, if a spectroscopy experiment does not have the right input range to identify the resonant frequency, a human can easily look at the resulting data plot and adjust the input range as needed. In software, that process is not as straightforward, requiring trend lines, gradient calculations, and experimental context to determine how to react. This is just one example of one experiment.

Every experiment has dozens of exceptions. Every routine requires dozens of experiments. And every device requires hundreds of routines executed on an increasingly growing number of qubits. The complexity of that dynamic is what the systematic approach of Boulder Opal Scale Up aims to solve by using a variety of physics, heuristics, AI/ML, and other strategies to overcome known exceptions types, such as:

• Preventing wire mapping errors through active system instantiations of system configurations
• Converging on device parameters across a wider range of out-of-spec possibilities
• Identifying controller electronics limitations impacting operational results

These exceptions, among others, are encoded into Boulder Opal Scale Up, which uses them in an intelligent way to logically run, rerun, and connect experiments to create a complete characterization and calibration routine. It is the software encoding of the human decision-making process that powers much of the quantum experimental process today. 

Tuning for performance with closed-loop optimization

Boulder Opal closed-loop optimizers are already a hardware-validated approach to performance tuning, completing optimization tasks 10x faster than traditional approaches. By scheduling these AI tools within an autonomous routine, Boulder Opal Scale Up efficiently and effectively delivers peak performance across an entire quantum device. The closed-loop optimizers use an iterative approach with an AI agent working to converge on the best-performing waveform pulse for each quantum operator. Then, by running on real hardware, they automatically factor in real-time quantum, environmental, and noise factors, resulting in the best possible performance for that moment in time. 

Boulder Opal Scale Up leverages these optimizers to tune every quantum gate and operator across the whole device. With the combination of intelligence, speed, and performance, this results in highly-tuned devices, every time.

Get real results on real hardware

We’re excited to announce that Boulder Opal Scale Up will soon be available for characterization and calibration routines on any QuantWare device. This means that with one line of code, or one API call, any QuantWare user can tune up their device to peak hardware performance. We are democratizing access to high-performing quantum devices, so users can make faster progress on the challenges that still need to be overcome. 

Join us on the continuous evolution toward intelligent autonomy

Boulder Opal Scale Up is the application of a variety of analytics and AI tools to solve a collection of quantum problems. Through sustained efforts with a variety of industry partners, Boulder Opal Scale Up will continually improve. It will become more robust by adding more exceptions it can handle; it will incorporate more parallelism for faster execution; it will integrate new experimental techniques as they emerge, continuously evolving to meet industry needs. This will be all while continuing to deliver faster and higher-performing quantum results. Boulder Opal Scale Up will be ever-improving and learning with the rest of the industry. 

With a flexible architecture, our solutions will expand to various qubit technology types, and support all leading control electronics. So rather than continuously building DIY routines, and manually running and debugging quantum control tasks, this software can leverage the latest in autonomous quantum control techniques, accelerating the collective path to a broadly usable quantum advantage. 

If you are interested in autonomous characterization or calibration routines on your own device architecture, reach out to us to schedule a Boulder Opal Scale Up demo. Come check out our demo on QuantWare devices at the APS Global Physics Summit in March17-20 at Booth #600.