A short background about quantum computing

Since its first proposal in the 1980s, quantum computation has developed into an interdisciplinary and active field of research. Today's classical computers, such as CPUs, GPUs, and TPUs, operate on the principles of classical physics, which is why they are often referred to as "classical" systems. The fundamental unit of information in these classical devices is the binary digit, or bit in short, which can take the values 0 or 1. At the lowest level, a classical computer encodes bits using two distinct electrical signals that represent the binary values 0 and 1, respectively.

Quantum computers, on the other hand, use quantum systems such as individual photons, ions, or electrons to encode information in quantum bits (qubits). For instance, a qubit can be defined by the horizontal and vertical polarization states of a photon or by two distinct energy levels of an ion. Following the physical laws of quantum mechanics, qubits can exist in a superposition of states, allowing them to represent 0, 1, or any combination of these two states "at the same time"*.

(*This expression is somewhat unfavorable, because a qubit in the superposition state does not imply that it is in multiple physical states; rather, its physical state may be expressed as linear combination of the two "basic" states, 0 and 1.)

This property of quantum superposition opens up an entirely new computing paradigm, enabling solutions to problems that are intractable for classical computers. However, the transition from the theoretical potential of quantum computing to practical, scalable implementations remains a major challenge, requiring research and development of hardware, software, and infrastructure.

The promise of quantum speed-up

One of the most exciting prospects for quantum computing is its potential to outperform classical computers in solving certain "computationally hard" problems - problems that are extremely difficult or time-consuming for classical computers. Algorithms such as Shor's algorithm for factoring large integers and Grover's algorithm for unstructured search are prime examples where quantum computers theoretically offer exponential and quadratic speedups, respectively, as the problem size increases. Potential applications range from optimization problems to machine learning and simulation of complex systems. These capabilities could have far-reaching implications for materials science, manufacturing and logistics, and more.

However, the reality of today's hardware tempers this promise. Current quantum systems operate in the noisy intermediate scale quantum (NISQ) era. The noisy and error-prone nature of this generation of current quantum computing devices limits the complexity of computations they can reliably perform. Achieving a quantum advantage, where quantum computers significantly outperform classical computers in practical tasks such as simulating molecular structures or solving optimization problems in manufacturing and logistics, remains a distant goal.

The NISQ era and cloud-based access

In the NISQ era, quantum hardware is scarce, specialized, and limited in its computational capabilities. While the state of the hardware is still in its infancy, building and maintaining a quantum computer requires enormous resources and specialized expertise, limiting access to a select few. To make this technology accessible, IBM took a pioneering step in 2016 by making quantum computing resources available through the cloud. Since then, many cloud providers, including Amazon, Google and Microsoft, have followed suit. Based on established cloud service models for classical software, this internet-based provision of quantum computing resources is often referred to as Quantum Computing as a Service (QCaaS).

Practical challenges in quantum cloud computing

Through QCaaS, users can access quantum hardware without the need to own the physical infrastructure. Despite the potential of QCaaS, the model also presents challenges. Quantum applications are hybrid in nature, involving both classical program execution and execution on a quantum device, typically as a subtask within a larger application workflow.

The integration of quantum systems into existing classical infrastructures and software systems is one of the main obstacles. A key issue is the lack of standardized interfaces and programming tools for the seamless integration of quantum hardware and software components. Another critical issue is the development of effective deployment methods for both quantum and classical components within a hybrid quantum-classical computational workflow. In contrast to classical software, which is typically hosted on servers ready for on-demand invocation, quantum programs are not designed in this manner. Instead, each time they are executed, they must be recompiled and transmitted to the quantum computer backend.

In addition, the diverse and heterogeneous landscape of quantum computing resources further complicates the practical implementation of QCaaS. The underlying quantum devices vary widely in physical characteristics such as error rates, supported gate sets, and computational models. Cloud providers offer different configurations, access models, and software ecosystems. This lack of uniformity presents additional challenges to the deployment of quantum applications.

Our work addresses these challenges and focuses on enhancing the execution of hybrid quantum applications in an as-a-service fashion.