Since we first described the state of the art in quantum computing in 2018, there have been significant advancements to align this progressive field of computing with today’s problems. In this article, we provide an update on the key developments in quantum computing to help solve complex problems. Even with these advances, quantum computing will continue to work alongside classical computing to get most jobs done.
How digital business connects
Connect digital infrastructure and services on demand at software speed via secure, software-defined interconnection. Scale hybrid deployments, achieve network agility, directly connect to partners and providers easily and securely.Read More
What is quantum computing?
In a classical computer, electronically operated switches called transistors are used to represent the familiar 0’s and 1’s that represent individual ‘bits’. From those basic ingredients, computer scientists have demonstrated that it’s possible to perform a wide variety of computational approaches to solve novel problems. While engineers have ensured that transistors have gotten smaller and more numerous, broadening the kind of problems that computer scientists can solve, the technology is not much different from 1930’s and 40’s pre-transistor devices based on valves or tubes.
In a quantum computer, transistors are replaced with devices that represent quantum bits or “qubits,” which are capable of representing both a 0 and a 1 at the same time. This by itself is not very useful, unless it’s part of a quantum processor capable of handling a larger number of qubits. In a classical computer, modern CPUs require billions of transistors to run modern operating systems and applications. In a quantum computer, every state is represented simultaneously, using as few as one hundred qubits to solve computational problems with a high degree of complexity. Such problems would take a modern classical CPU much longer than the expected lifetime of the universe to solve.
Take the famous traveling salesperson problem for example — depending upon the problem size, in classical computing one would quickly need a very large number of bits to represent all the cities in the search space. If the salesperson needed to visit more cities, one would require vastly more bits to represent the problem search space. However, in quantum computing, since an array of entangled qubits can simultaneously represent 2 exp (n) states all at once (where n represents the number of entangled qubits), more alternate solution options can be simultaneously represented more efficiently and processed in parallel much more quickly than in a classical computer.
In real-life, as shown in this diagram of a logistics company, one can represent all the cities it needs to deliver goods as a quantum computing problem to determine the optimum driving pattern to reduce fuel costs.
Logistics Company Delivery Map
Why quantum computing?
The promise of quantum computing is that it will help us tackle certain types of problems that today’s classical computers cannot solve in a reasonable amount of time. It is important to note that quantum computing is not a panacea for all types of computing problems but is good for most “needle in a haystack” types of search and optimization problems. For example, if you wanted to find one item in a list of 1 trillion and each item took 1 microsecond to check, it would take a classical computer one week versus 1 second for a quantum computer.
Real-life applications for quantum computing include:
- Figuring out the optimal route that a salesperson should take to cover multiple cities to save on time and fuel-costs.
- Simulating chemical reactions to create better batteries for electric cars.
- In the area of cryptography, for breaking or creating security codes.
- Finding Higgs events (one of the fields in particle physics theory) and the origin of the universe.
- Enabling financial companies to better analyze their data to determine fraud or balance their investment portfolios.
A lot of initial interest in quantum computing can also be attributed to the fact that quantum computers that are sufficiently powerful enough can be used to break certain types of crypto codes (e.g. Diffie-Helman, and Elliptic Curve Diffie-Helman key exchange). This can impact security for cell phones, bank accounts, email addresses and crypto wallets. As a result, organizations will eventually have to modify some of their existing solutions (similar to the Y2K problem) for the quantum computing era.
In many use cases, one needs to take a hybrid approach and use a combination of classical and quantum computing to solve a problem. This is very similar to artificial intelligence pipelines that use CPUs to clean the data and GPUs for doing AI model training operations, and subsequently use either CPUs, GPUs, field programmable gate arrays (FPGAs) or ASIC solutions for model inference operations.
The challenge for quantum computing
One of the major problems being faced by today’s quantum computers is that entangled qubits quickly become decoherent with respect to other qubits. Therefore, an algorithm needs to quickly get its work done before the qubits become decoherent.
Currently, most quantum computers can only keep a few tens of qubits coherent. A recent study has shown that cosmic rays can introduce a burst of decoherence errors that are hard to correct using standard error correction techniques. This contributes toward our inability to represent meaningful real-life problems on a quantum computer.
Furthermore, there is no uniformity in the underlying quantum computing hardware. Currently, various companies are pursuing different approaches toward building a quantum computer — for example, Quantum Annealer, Analog Quantum Computer and Universal Quantum Computer. This is very similar to how we had multiple transistor designs in the initial days of computing. As a result, it is only possible for certain problems to be efficiently mapped onto specific types of underlying quantum computing hardware. Research into solving the decoherence problem and designing general purpose quantum computers is ongoing, and we are still about five years away from solving meaningful problems on a quantum computer. In the meantime, we anticipate the deployment of both quantum and classical computers in a hybrid manner to provide computational efficiencies.
How quantum computers are being deployed
Quantum computers require customized hardware; today only the big hyperscalers and a few hardware companies are providing quantum computer emulators, as well as limited-sized quantum computers, as a cloud service. Quantum computers are presently being targeted for problems that are compute intensive and not latency sensitive. Also, today’s quantum computer architectures are not mature enough with respect to handling large data sizes. As a result, in many cases, a quantum computer is typically deployed along with a classical computer in a hybrid manner. Even though a quantum computer itself does not consume much power during computation, it requires specialized cryogenic refrigerators to maintain low superconducting temperatures.
Current predictions show that there will be fixed cooling costs with very little incremental power cost as the number of qubits increases up to a certain limit. However, one will need many of such cooling refrigerators to support computers that provide millions of qubits. The current cryogenic refrigerators consume around 24KW per unit for approximately 1000 qubits. This power budget does not include the power for the classical computer, storage and networking hardware surrounding a quantum computer.
Quantum software and networking stacks
Many software stacks are being proposed for quantum computing that consist of virtualizing the underlying physical quantum computing hardware and building a virtual layer of logical qubits. Furthermore, the software stacks provide compilers that convert higher level programming language constructs into lower-level assembly commands that operate on the logical qubits. Software stack providers are also developing specific application-level templates that are domain specific (e.g., optimization problems or specific machine learning problems) and that map onto the quantum computing programming model. The goal of the software stack is to hide complexity without compromising the overall performance or maneuverability of the underlying quantum computing hardware.
With respect to a native quantum computing networking stack, its development is still in the early stages. Currently, quantum computing data and results need to be converted into a form that can be understood by classical networking equipment and then re-converted back into a quantum computing-understandable format. Currently, a lot of research is being done into the area of native quantum computing networks, where qubit entanglement can be achieved across long distances, however, these are not ready for commercial deployment.
Getting quantum computing ready for prime time
Many large computer companies are investing billions of dollars into building quantum computers. Similarly, many academic institutions are also investing a lot of money and brainpower into this area. The current generation of quantum computers need to be managed by expert staff due to their specialized hardware and cooling requirements. As a result, in the near future, quantum computing functionality will be mostly offered as a cloud service.
However, there will be customers who will want to deploy their data and host the classical computing and storage portion of their overall architecture in their own private data centers or cages in a colocation facility for privacy and control reasons, and these customers would like to leverage quantum computing functionality from a service provider. As shown in the figure below, a highly interconnected colocation data center platform, such as Equinix, allows both enterprises as well as quantum computing service providers to have multiple hybrid deployment options.
Hybrid Quantum Computer Deployment
In conclusion, we are in the initial stages of the era of hybrid quantum computers where classical computers offload certain types of processing to quantum computers with limited number of qubits. We believe this will be the norm for quantum computing for the next five years before non-hybrid native quantum computers containing thousands of qubits start solving real-life problems.
Learn more about Equinix Fabric software-defined interconnection.