Implementation in Quantum Clouds

This chapter will dig deeper into the options for executing our algorithms in quantum devices or at least solutions that will go beyond the capabilities of our classical devices. For the sake of simplicity, most of the algorithms you have seen so far used some kind of local simulation to mimic how the outcome would look when running on a real quantum computer.

When developing our approaches, we have the means to locally simulate the behavior of an ideal quantum computer while taking from the mathematical description each operation requires. But at the end of the day, the goal is to be able to send our work to a quantum device that will leverage the potential of actual quantum computing.

Given that owning a quantum computer is something very few privileged people will be able to do, we will highlight how cloud access has been an important way of leveraging those still experimental resources.

We will also illustrate several examples demonstrating...

Challenges of quantum implementations on cloud platforms

As we mentioned previously, most of the examples shown previously leverage the fact that quantum computing can be mimicked by our classical resources (using simulators). As an example, in Chapter 5, we used the following routine:

backend = Aer.get_backend('qasm_simulator')
result = execute(quantum_circuit, backend, shots=10).result()
counts  = result.get_counts(quantum_circuit)

We utilized a qasm_simulator, an implementation that can execute the operations defined in our quantum circuit and provide the expected outcome as dictated by the mathematical principles governing quantum computing.

We would like to take this very same circuit to a quantum computer, but it is not as easy as purchasing one on Amazon. We can purchase commercially available devices, but the price might be too high for most organizations.

D-Wave

In 2011, D-Wave Systems announced the world’s first commercially available...

Cost estimation

We briefly mentioned resource estimation, but we would like to highlight the importance of this ability when aiming for a sustainable adoption strategy.

Quantum computing is at an early stage, requiring continuous and complex maintenance tasks to provide the best possible service. This is something that classical computing resources have mastered for a long time. Due to the limited ecosystem of hardware providers and the status of the technology, specifically, when aiming for real hardware, we will see costs ramp up significantly, even for the simplest providers. That is why resource estimation is such a crucial step in any QC pipeline, particularly if the model’s training requires iterations:

Figure 7.26 – Cost for Quantinuum on Azure (below) and IonQ in AWS (above)

As an example, we could take the European Call Pricing from Chapter 4 and extract its underlying quantum circuit, which we already know is composed of a block...

Summary

The transition from on-premises to cloud-hosted has been a complicated journey for many organizations, including the switch to be made from owning the computing resources to pay-per-use modalities common today. Quantum computing made its initial foray directly into the cloud. Different platforms give access to services and providers with efficient costs. Many institutions have facilities to onboard into this quantum journey.

Quantum hardware is problem-focused and increases the complexity and decisions to be made as you must decide what option, out of the plethora of devices, is the most convenient for your problems. Estimators help companies evaluate the cost of using this new type of machine, providing them with an efficient way to estimate the budget required per study.

No technology has faced such a wide, almost free-of-charge offering to learn and adapt than quantum computing. Hardware providers enable their latest devices in those cloud services and help boost the...

Further reading

Many providers are already shaping their offerings so that new as-a-service paradigms will emerge in the following years.

Oxford Quantum Circuits has already embraced the concept of Quantum Computing as a Service (QCaaS) (https://www.techradar.com/news/quantum-computing-as-a-service-is-going-mainstream), whereas other companies like QCentroid are targeting a wider audience by offering off-the-shelf solutions tailored to industry-specific applications through their Quantum-as-a-Service (QaaS) platform (https://marketplace.qcentroid.xyz/).

When thinking about cloud-accessible resources, one of the most interesting cases is the one posed by the variational quantum algorithm, where a constant interchange between classical and quantum resources must be sorted out. Given the queue times we have seen, we must be aware that any remote training of the ansatz will face important delays per iteration if we attempt to train on an actual device remotely.

Given the existing...

References

Cross, A. W., Bishop, L. S., Smolin, J. A., & Gambetta, J. M. (2017). Open quantum assembly language. arXiv preprint arXiv:1707.03429.

Anis MS, Abraham H, AduOffei RA, Agliardi G, Aharoni M, Akhalwaya IY, Aleksandrowicz G, Alexander T, Amy M, Anagolum S. (2021). Qiskit: An open-source framework for quantum computing. Qiskit/qiskit.

Smith, R. S., Curtis, M. J., & Zeng, W. J. (2016). A practical quantum instruction set architecture. arXiv preprint arXiv:1608.03355.