Quantum computing is undeniably one of the most hyped technologies in recent years. With seemingly frequent advancements and milestone announcements, this field has attracted billions of investments into both established and nascent quantum computing vendors. The penultimate promise of quantum computers is that they will enable the computation of intractable problems that classic computers cannot easily solve. One such problem is the travelling salesman problem (TSP), where we want to compute the shortest cyclic route between different cities without going to the same city twice. While it sounds simple enough, the combinatorial space grows exponentially as we add more cities and routes. It would take a tremendous amount of classical computing power to solve this problem beyond a certain size. Given its characteristics to exploit the quantum states, a quantum computer promises to solve this problem exponentially faster. What makes the TSP problem a very attractive low-hanging fruit is that there are a lot of economically-relevant problems that share the same characteristics, particularly in logistics, chip/circuit design, fleet scheduling, etc.

In 2019, Google published a paper demonstrating quantum supremacy where they claimed that their quantum computer managed to solve a problem that would take classical computers 10,000 years in just 3 minutes. Although IBM hotly contests the claim, it arguably evoked the public and market imagination about how quantum computing has finally arrived. In 2020, D-Wave announced its Quantum Annealing computer with 5000+ qubits. IBM recently announced the first gate-model quantum processor that breaks the 100-qubit barrier with its 127-qubit Eagle processor. But despite these advancements, the question remains. Is it good enough to solve practical problems yet? A practical quantum processor would have to be both large-scale (in terms of the number of qubits) and coherent (in terms of error correction). Both come with their own set of significant challenges. However, as it stands today, we can already access these quantum technologies and apply them to some computational problems, albeit with some limitations.

In the following paragraphs, I will describe some of the readily-available quantum technologies in the market, their characteristics, limitations, and how we can peel off the hype and see what is practical today. Thereafter, I will describe how you can use these technologies for exploration or testing in your research with our Applied Quantum Cloud Credit program.

There are currently two types of commercially-available quantum computers: gate-model and quantum annealer. A gate-model quantum computer uses quantum logic gates as the basic operation units on qubits. Like logical operators, e.g., AND, OR, XOR in a classical computer, it uses logic gates as building blocks to form a quantum circuit that expresses the equation to solve. As the gates are fully configurable and can perform all operations performed by classical digital circuits, it is the most general-purpose quantum technology available today. IBM, IonQ, and Rigetti are examples of companies that build the gate-model type of quantum computer.

The second type is quantum annealer. It is a more limited version of quantum computing specializing in solving NP-complete optimization problems by exploiting quantum mechanics. Multiple qubits in a lattice are arranged and connected by couplers and exposed to external magnetic fields called bias. The coupler connects two qubits and allows entanglement to happen and influence each other’s state. The optimization problem, expressed in a quantum equation, will then be coded through the configuration of couplers and biases and form a specific quantum energy landscape. Once run, the system will naturally settle in the minimum energy state of the configured landscape, which should represent the optimum solution. Unlike the gate-model, the quantum annealer is more stable and less affected by noise. Thus, D-Wave managed to build a 5000+ qubits quantum annealer, while IBM Q only has 127 qubits of its gate-model quantum processor. Consequently, by virtue of qubits count, D-Wave has the advantage of solving some practical optimization use-cases with its current technology.

Coming back to whether these technologies are practical enough to solve some of your research problems, most likely the answer is no. Even when these problems can be demonstrated to work on a quantum computer, the classical computer might still do it faster, given its ubiquitous availability. This is not a slight against quantum computing itself but a comment on the limitation of current technology iteration. Given the amount of investment being poured into it, we may see something entirely different in 5-10 years.

In the meantime, what we can do today is to prepare and be ready when these technologies come in a mature enough state. We would recommend using the quantum annealer technology as a start as it has enough qubits to make any experiment and exploration meaningful and informational. In addition, it naturally fits combinatorial optimization problems, something that we often see in scientific research.

There are four steps that a researcher can take to explore quantum computing today:

1. Identify potential combinatorial optimization problems in your research area. Material discovery and RNA folding are two optimization problem examples that might benefit from quantum annealer.
2. Convert the optimization problem into QUBO (Quadratic unconstrained binary optimization) format.
3. Code and execute the QUBO-formatted problem in a quantum annealer system. As purchasing the system is probably not realistic, the next best thing is to use a cloud-based quantum annealer.
4. Benchmark the result against a known baseline. If this is a novel work in your field, you might as well publish a paper on it.

The Research Computing group can help you access cloud-based quantum technologies through the AWS Braket offering. You will have a selection of current quantum computers from D-Wave, IonQ, Rigetti, and OQC. With our recently announced AWS Cloud Credits for Research programme, you can apply for the Applied Quantum in Research track that gives you access to (up to) $20,000 worth of quantum computing experiment time in AWS Braket. We can also connect you to AWS experts and resources to deepen your expertise in this new field. All you need to do is apply and write a short proposal to be considered and evaluated. The programme does not require any commitment or obligation at all. At the end of 1 year, we only need you to share your experience and result.

If you are keen on starting your quantum journey with us, you can follow this link to check the details, and do contact us if you have follow-up questions through this email: nusit-hpc@nus.edu.sg.

Looking forward to hear from you. Happy experimenting!