Monday, February 26, 2024

Quantum Computing Coming Faster Than You Think

By Janani G. Vikram

- Advertisement -

From cybersecurity to drug discovery, there are several challenges that quantum computing can tackle better than classical systems. With improving hardware, full-stack solutions, better algorithms, simulators, and cloud offerings, the technology is at present in ‘democratisation’ phase. Perhaps it is time to make your quantum plans too

“Given a list of cities and the distances between each pair of cities, what is the shortest possible route that visits each city exactly once and returns to the origin city?” First formulated in 1930, this mathematical problem, known as the ‘travelling salesman problem,’ is one of the hardest to solve, yet one of the most essential optimisation issues for any logistics provider.

“Quantum computing has the potential to blow classical computing out of the water for certain types of complicated problems. For example, constrained optimisation challenges like the famous travelling salesman problem can have trillions of equations and variables. The most complex of these problems will simply be unfeasible for classical computers to tackle, but quantum computers will be able to solve them in seconds. And these aren’t just laboratory problems; constrained optimisation challenges heavily impact industries like supply chain and retail, cybersecurity, life sciences and pharmaceutical, and government and applied research,” says Robert Liscouski, CEO of Quantum Computing Inc. (QCI).

- Advertisement -
IBM Quantum Lab at the Thomas J. Watson Research Center (Credit Connie Zhou for IBM)
IBM Quantum Lab at the Thomas J. Watson Research Center (Credit Connie Zhou for IBM)

Another great example involves autonomous vehicles. QCI recently solved a 3854-variable sensor placement problem for BMW using only quantum computing—and they did it in six minutes. This is a clear example of quantum computing’s ability to solve actionable, real-world problems that would take classical computers a much longer time to solve.

“To clarify, sensor placement is extremely challenging because of all the variables that must be considered, including wind resistance, weight balancing, chassis design, potential obstructions, and more. Our machine found a way to provide 96% vehicle coverage using only 15 sensors. So, when you ask about potential, we’re talking about saving autonomous vehicle designers a tremendous amount of money. This is compared to the solutions provided by other quantum and classical computers that suggested using well over 100 sensors,” explains Liscouski.

Mercedes-Benz, which hopes for all its vehicles to be carbon-neutral by 2039, is working with IBM Quantum to simulate the chemical reactions in batteries more accurately. Basically, they are trying to understand what happens at a molecular level inside the battery, while it is working. This is a task that is practically impossible for classical supercomputers, as it involves a tremendous number of electron interactions, each influencing the other in complex ways. No wonder, Dr Jeannette (Jamie) Garcia, Senior Manager of Quantum Algorithms, Applications and Theory at IBM, remarks that, “It is chaos in there!”

“Most industries have their own big, intractable challenges. And quantum computers offer potential solutions. For instance, quantum computing can shed light on processes of molecular and chemical interactions,” says L. Venkata Subramaniam, Senior Manager – AI Science and IBM Quantum Ambassador, IBM Research India. Mitsubishi Chemical, Keio University and IBM are working together to model and study the complex mechanism for lithium superoxide rearrangement, a key chemical step in lithium-oxygen batteries. Using a quantum computing setup enables them to more accurately measure the reactant energy used to produce electricity in a Li-air battery.

So, is quantum computing only for researchers and designers? No, every large business house can use it to optimise its running. “Quantum computing is enabling ExxonMobil to solve several computationally-challenging problems such as optimising natural gas supply and the discovery of new materials for efficient carbon capture,” says Subramaniam.

Efficient shipping of liquid natural gas (LNG) is extremely critical, because people could run out of power without it! The global LNG industry involves thousands of voyages across the world. Optimising these, at the most basic level, would involve millions of discrete decisions. If you take into consideration uncertainties like bad weather and market fluctuation, it could run into even trillions.

According to Dr Vijay Swarup, Director of Technology at ExxonMobil, this means that the number of combinations they would need to consider would be larger than the number of atoms in the entire universe. ExxonMobil is working on modelling maritime inventory routing on IBM’s quantum setup.

Quantum computer mixing chamber (Credit: IBM Research)
Quantum computer mixing chamber (Credit: IBM Research)

Think quantum

“We tend to think of problems and solutions in classical ways, for which the existing classical computational ecosystem is fast and powerful enough. Most of the standard/classical problems can be adequately addressed within the existing infrastructure. However, there exists a certain class of problems that quantum computers can solve in more effective ways,” explains Rahul Mahajan, CTO, Nagarro.

“Drug discovery, security, and search are a few domains where we foresee some interesting applications using quantum computers. Given a sufficiently large quantum computer, there exist a few algorithms like Grover’s and Shor’s, which can be deployed on quantum machines to execute the solution faster. For example, a class of RSA algorithms use factorisation based encryption, leveraging quantum parallelism in qubits registers, and simultaneous evaluation paths can be traversed,” he says.

Horizons of quantum growth
Horizon 1, also called now or near term, includes transactional use cases, such as credit scoring, vehicle routing, chemical design, chemistry, and drug/protein structure prediction.
Horizon 2, also called near term, follows with oil processing and shipping, refining processes, drilling, livestock, disruption management and supply chain issues, investment risk analysis, clinical trial acceleration, and optimisation of manufacturing and fabrication.
Horizon 3, also called future looking, consists of seismic imaging, consumer recommendation with financial analysis, disease risk prediction, and structural design for buildings.
(Credits: Paul Nashawaty, Senior Analyst at Enterprise Strategy Group, a division of TechTarget)

“Quantum computing makes different types of seemingly intractable optimisation problems, like routing of delivery vehicles, more manageable. Another class of problems comes from chemical reactions. These are by nature quantum mechanical. The best way to understand them is via simulations on a quantum computer. With enough qubits, we could tackle problems in electric battery design, find new molecules, and aid drug discovery. The societal benefits that would follow are immense,” says Professor Anil Prabhakar, Department of Electrical Engineering, IIT Madras.

In the last few years, there have been significant developments in this space—mergers, alliances, stock launches and public-private partnerships; better algorithms, full-stack solutions, hardware upgrades and concrete tech roadmaps; and an expanding portfolio of applications. We see people rethinking quantum deployments, often opting for quantum simulators, and hybrid systems with quantum computers working hand-in-hand with CPUs and GPUs. We see the emergence of cloud ecosystems that mask the complexity, enabling even non-quantum experts in different fields to use quantum computing to solve stubborn problems.

We do not have to wait for quantum computing to replace classical systems, to finally agree that quantum is here—because that will probably never happen. We need to understand that classical and quantum do not actually compete—the purpose of a quantum computer is not simply to solve classical computing problems faster than a supercomputer. It is fundamentally different, and ideally suited to crunch complex problems that require calculating a large number of possible combinations.

The problems best left to quantum computers are known as bounded-error quantum polynomials (BQP), that is, problems that can be solved by a quantum computer in polynomial time, such as factoring, searching an unordered database, and solving systems of linear equations. Problems like this abound in fields like optimisation, materials science, cryptography, artificial intelligence, machine learning, sensing, telecommunications, and finance. Many processes in nature also involve many-body quantum interactions. This makes quantum computing an essential tech to simulate natural processes like protein folding, molecular formations, photosynthesis, and superconductivity, and use this understanding to develop better nature-inspired technologies.

“Today I cannot see that there will be something like a quantum PC,” quips Professor Rainer Dumke, Principal Investigator at the Centre for Quantum Technologies, Professor at Nanyang Technological University, Singapore, and Lead for hardware at Singapore’s National Quantum Computing Hub (NQCH). “We will witness in the future that more and more data centres will have a quantum computer on the premises. To give a meaningful timeframe for this is very challenging, especially at this early stage. However, we see already that supercomputers and data centres around the globe are adopting quantum computers of the current generation.”

At the NQCH, they are working to establish a Singapore-built quantum computer, which can be accessed by users remotely. They also hope to develop an advanced quantum processor test platform, along with various components like chips, fast electronics, and cryoelectronics for the next generation of superconducting quantum processors.

Overcoming the hardware challenges

In an actual quantum computer, physical systems, such as the spin of an electron or the orientation of a photon, are used to store qubits. These, as you can imagine, are pretty fragile systems, and are easily affected by various factors, such as disturbances in the earth’s magnetic field, radiation from devices in the vicinity, heat, neighbouring qubits, and so on. These factors, which affect the information in a qubit, are known as ‘noise,’ and since today’s quantum computers are prone to their effects, they are known as noisy intermediate-scale quantum (NISQ) devices.

Unfortunately, even reading or accessing a qubit amounts to noise! Noise can cause the information in the qubits to change, or even fade away! This is known as decoherence—that is, the information in the qubit is no longer coherent. Noise and resulting decoherence are two of the biggest hurdles in the way of quantum computers.

Quantum computer makers try to protect the qubits from noise by using cooled silos, controlled pulses of energy, etc. So, if you look at a quantum processor, such as the one from IBM, the wafer is just about the size of the one in your laptop, but it is housed in a hardware system the size of a car, made up of cooling systems to keep the superconducting processor at its ultra-cold operational temperature.

The task of mitigating the effect of noise is known as quantum error correction. Mostly, a lot of spare physical qubits are used to store every logical qubit, so that in case one loses its coherence you can verify with another. Researchers also use noise simulations to try and understand the nature and effect of noise, and develop noise-tolerant algorithms, which will give fairly good results despite the noise. Finnish company Algorithmiq is working to develop noise-resilient quantum algorithms for drug discovery. Their strategy involves using standard computers to ‘un-noise’ quantum computers!

“Techniques for error mitigation and error correction are likely to be the way forward on NISQ computers. There is currently a trade-off between the number of qubits we can use and the depth of the quantum circuit. The product of the two is referred to as the quantum volume. Researchers are finding ways to increase the quantum volume, and demonstrate the advantage of quantum algorithms. One such approach is hybrid quantum computing, where the computational task is shared between classical and quantum processors,” says Prof. Prabhakar.

“The research on quantum fault tolerance is ongoing, and the initial research was done by Peter Shor. The latest innovation is about generating sufficient redundancy to maintain information integrity. These redundancy methods have negatively introduced logarithmic downsizing on actual logical qubits available for compute, and as of today, we are barely in the range of a few dozen of logical qubits,” explains Mahajan.

Subramaniam shares that, “At IBM, we are developing bottom-up approaches to the problem of noisy qubits and incorporating error mitigation techniques to realise this technology’s true potential. Superconducting qubit based systems have made tremendous strides in device performance, from improved coherences to lowered single- and two-qubit gate errors, and high-fidelity mid-circuit measurements and qubit resets. In the future, we plan to bring together advances in device performance, software and error correction to offer users a frictionless experience and the ability to solve tough computational problems.”

Last year, IBM unveiled Eagle, IBM’s first quantum processor developed and deployed to contain more than 100 operational and connected qubits. “It is the first IBM quantum processor whose scale makes it impossible for a classical computer to reliably simulate,” says Subramaniam.

IBM has the 433-qubit Osprey processor slated for release later this year, and the 1121-qubit Condor processor slated for release next year. In 2024, IBM plans to introduce longer-range quantum communication between chips and create clusters of quantum processors using a long-range coupler for connecting qubit chips through a cryogenic cable of around a metre long. This will be demonstrated by linking together at least three 462-qubit processors, each called Flamingo, into a 1386-qubit system.

IBM’s next quantum processor, Kookaburra, is also slated for release in 2025. It will be a 1386-qubit multichip processor with quantum communication link support for quantum parallelisation. IBM also plans to connect three Kookaburra chips into a 4158-qubit system connected by quantum communication.

Quantum Communication Lab (Credit: IIT Madras)
Quantum Communication Lab (Credit: IIT Madras)

Beyond superconducting qubits

Prof. Prabhakar remarks that most people tend to think only of superconducting qubits from companies such as IBM or Google, when considering quantum computers. “There are other hardware platforms also gaining traction, like ion-traps, cold atoms, and photonic cluster states. These are very different approaches and I expect that they will yield application-specific quantum processing units (QPUs) in the near term,” he says.

QCI has developed one such system, which Liscouski describes to us. “Natural quantum states interact freely, influencing and impacting each other as they evolve and change. This natural interaction (or ‘noise’) significantly impacts the accuracy and scale of first generation NISQ computers. These vendors painstakingly create hyper-cooled, vacuum environments to house their computers, but even this doesn’t prevent them from making significant errors, losing information, and suffering from limited scale. Our photonics-based Entropy Quantum Computer (EQC), on the other hand, harnesses the fundamentals of quantum physics to overcome these limitations. It operates on open quantum systems, carefully coupling to an engineered environment, so that the quantum state is collapsed to represent a problem’s desirable solution. In layman’s terms, where other quantum computers try to avoid noise, our system uses the noise,” he says.

As a result, QCI’s EQC does not require any specialised or costly infrastructure, and needs minimal calibration. It is stable and maintains coherence at room temperature, in any environment. So, there is no restriction to where and how it can be deployed.

Xanadu’s Borealis is also a photon based quantum computer, where a quantum light source, with adjustable brightness, emits trains of up to 288 squeezed-state qubits.

Honeywell Quantum Solutions uses trapped ions to achieve their quantum goals. According to their documentation, their systems ‘trap’ charged ytterbium atoms (ions) with electromagnetic fields so they can be manipulated and encoded with information using microwave signals and lasers. Their latest offering, the System Model H1, promises high-quality quantum operations (fidelities) and longer coherence times than other quantum computing technologies.

Honeywell has collaborated with Cambridge Quantum, which makes AI based quantum software and algorithms, to form Quantinuum, to offer an integrated end-to-end quantum platform that is entirely platform agnostic. Their first product, Quantum Origin, is the world’s first cryptographic key generation platform based on verifiable quantum randomness. They are also developing solutions for chemistry, natural language processing, machine learning, and more.

From here and there
  • Intel, together with collaborators, recently created the first silicon qubits at scale at its D1 manufacturing factory in Hillsboro, Oregon. The result is a process that can fabricate more than 10,000 arrays with several silicon-spin qubits on a single wafer with greater than 95% yield. The qubit count and yield are much higher than the typical university and laboratory processes used today.
    • Scientists at the Harvard-MIT Center for Ultracold Atoms have demonstrated a programmable quantum simulator capable of operating with 256 qubits. The system uses arrays of highly-focused laser beams, to trap individual qubits and drag them into desirable arrangements to perform meaningful calculations. This will help to build large-scale, practical quantum computers with great applicability, in the near future.

IonQ, the first quantum computing startup to go public, uses individual atoms as their qubits. According to company literature, “At IonQ, we take a different approach, and use a naturally occurring quantum system: individual atoms. These atoms are the heart of our quantum processing units. We trap them in 3D space, and then use lasers to do everything from initial preparation to final readout.”

Despite all the advancements, Prof. Dumke feels that, “Hardware is and will be always the bottleneck. This is true for any technology, and for classical computers as well as for quantum computers. The reason why we do not have a universal quantum computer yet is that the hardware is not mature enough. I can foresee that specific hardware solutions made from noisy processors will be tailored to solve special use-cases in the near future.”

Making quantum more accessible

More than pure quantum tech advancements, what is more interesting today are the different approaches being taken to make quantum computing accessible to more people. Two of the most significant efforts in this direction involve offering hybrid solutions, which combine quantum computing with classical computing setups; and full-stack cloud solutions that mask the complexity and let more people tap the benefits of quantum.

Liscouski says, “Our goal from the beginning has been to make quantum computing as available as possible so that business users with no quantum expertise can benefit from it. That’s why we invented Qatalyst, a ready-to-run quantum optimisation and machine learning software that masks the massive complexity and effort needed to solve even basic problems on quantum computers.”

Through QUBT U, they also provide the Qatalyst software to students, allowing them access to the world of quantum computing without having to be quantum programmers! “We believe that workforce development is a key component of our responsibility to our society,” says Liscouski.

IBM’s plans to build a 4,000+ qubit processor by 2023 goes hand-in-hand with significant milestones to build an intelligent quantum software orchestration platform that will abstract away the noise and complexity of quantum machines, and allow large and complicated problems to be easily broken apart and solved across a network of quantum and classical systems. “There is an inherent strong coupling of quantum and classical computers. Both paradigms of computing—classical and quantum—will work seamlessly together to solve different parts of complex computing problems that are best suited for their respective strengths and capabilities,” says Subramaniam.

IBM is integrating quantum computing with high-performance computing and hybrid cloud technologies in serverless implementations that remove the complexity of infrastructure management, putting the focus on coding only.

“It took 60 years to abstract software in classical computing to the point where users could input a simple line of code into a templated program in order to build an app or website. Quantum computing is going through a similar process—within the decade. IBM released the Qiskit Optimization Module 2020 as a first step in making frictionless quantum computing a reality by 2025.

“In the future, a program will be handed off to the cloud and vast quantum and classical resources will be employed, and within the blink of an eye the solution will be returned perfectly optimised. The Qiskit Optimization Module enables easy, efficient modelling of optimisation problems for developers and optimisation experts without quantum expertise. It uses classical optimisation best practices and masks complex quantum programming,” says Subramaniam.

IBM has installed on-site quantum systems in Germany and Japan, with plans for installations in Korea, Canada, and in the USA at the Cleveland Clinic. They also have over 20 systems in their Poughkeepsie and Yorktown locations in the USA, which can be accessed over the cloud.

In 2016, IBM became the first company to put quantum computers on the cloud. This was a turning point. “Since then, we have actively built up an active community of more than 400,000 users and 190+ organisations in the IBM Quantum Network that are running more than 4 billion circuits every day and exploring practical applications to realise the wide-ranging benefits of this technology for business and society,” says Subramaniam. From India, IIT Madras became the first institution to join the IBM Quantum Network, in September 2022.

“Companies such as IBM and Microsoft have built many sophisticated simulators and software stacks that are interoperable with different QPUs. There are also startups like D-Wave and Xanadu, which have got their own software platforms. Amazon Web Services (AWS) is attempting to become QPU-agnostic through its Braket portal. Most recently, Nvidia has stepped into this space. So, there are a lot of healthy alternatives,” says Prof. Prabhakar.

Microsoft Azure Quantum is a hardware-agnostic platform, which brings together a range of quantum computing and optimisation solutions into a single cloud service. You can write code and run it on the quantum hardware of your choice. QCI, IonQ, and Quantinuum are some of those who have partnered with Azure. NASA’s Jet Propulsion Lab used Azure Quantum to develop an optimisation solution that helped reduce scheduling times from hours to minutes. Ford and Microsoft teamed up to use quantum methods to figure out how to reduce traffic congestion in Seattle.

Google’s Cirq is a Python software library for writing, manipulating, and optimising quantum circuits, and then running them on quantum computers and quantum simulators. Cirq provides useful abstractions for dealing with NISQ computers, where details of the hardware are vital to achieving useful results.

Amazon Braket provides customers access to quantum computing technologies from multiple quantum hardware providers, including D-Wave, IonQ, and Xanadu. It also offers a consistent set of development tools for you to build and test quantum computing projects.

With simulators available, who needs quantum computers?
In 2019, Google researchers claimed they had achieved quantum supremacy when their quantum computer Sycamore performed an abstruse calculation, which would take a supercomputer 10,000 years to solve, in 200 seconds. Recently, a group of scientists in China created waves by completing the same computation in a few hours with ordinary processors. They recast the problem as a 3D mathematical array called a tensor network, and ran the simulation by multiplying all the tensors.
Quantum simulators are software programs that allow you to use a classical computer to run quantum circuits as if they were being run on a quantum computer. There are several simulators like Intel-QS and NVIDIA cuQuantum. And the question often arises as to why we need quantum computers when these simulators are there. While simulators can probably solve some problems, solving actual quantum problems using them would require an unmanageably high amount of computing resources. That said, simulators can be helpful in developing quantum computing applications and algorithms, as they can help debug code and test near-term systems.

Prof. Prabhakar adds that most of the software is built around an open source effort. “These are still early days, and a collaborative approach is likely to yield higher dividends. However, there is a danger of having too many alternatives for a beginner to comprehend. It helps that most platforms rely on Python libraries, making them amenable to a quick introduction,” he says.

Applications galore

Although quantum computing is quite a few years away from becoming mainstream, it is time for companies to start exploring what it can do for them. Liscouski lists out a few key industrial applications, for us to understand the potential of quantum computing to transform businesses:

Supply chain and retail:

  • Quantum computing enables users to improve multi-level replenishment throughout the entire supply chain from the manufacturer to the distribution centres, right down to local stores.
  • Users can accelerate warehouse performance with highly optimised pick order and path routing, while accommodating the use of more autonomous robots (as well as the growing data complexity that comes with them).
  • Quantum computing can perform inter-modal optimisations (for example, across long-haul and local surface freight) that would be very challenging for classical computers.

Cybersecurity:

  • Cybersecurity is a perfect application for quantum computing because of the complexity of unstructured data that must be repetitively and efficiently analysed to prevent threats and respond quickly to intrusions.
  • Quantum computing enables the real-time investigation of network intrusion to disrupt and prevent data exfiltration incidents.
  • It also has the ability to detect patterns of divergent network traffic.

Quantum sensing:

  • Quantum light detection and ranging (lidar) will extend the capabilities of existing lidar to provide enhanced data acquisition and analysis.
  • Quantum sensing is also applicable in the medical imaging field; imagine being able to see below the skin without high power such as X-ray.

Life sciences and pharmaceutical:

  • Quantum computers can assist in a wide range of application areas where understanding interrelatedness of common attributes is key to discovery. For example, identifying patient attributes and subpopulations in clinical trials.
  • Quantum computing can help discover heterogeneous networks of proteins and genes implicated in human biological function and disease states or in viral or microbial life cycles.
  • It can also analyse networks to compare molecular substructures and analyse the larger molecules necessary in pharmaceutical research.

“The early adopters are financial institutions seeking an advantage in wealth management, logistic companies attempting to reduce costs and improve efficiencies, and pharma companies that need to tackle the quantum chemistry of their new molecules. We also expect telecom and data security companies to be watchful of any breakthroughs in cryptography, and hence invest in quantum communications and post quantum cryptography,” says Prof. Prabhakar.

“IIT Madras has been an early mover in quantum science and technology. We had India’s first quantum key distribution demonstration as early as 2008. Today, we have about 20 faculty and their students and staff looking at different aspects of quantum science and technology. This has developed into a healthy interdisciplinary ecosystem, and is formally known as the Centre for Quantum Information, Communication and Computing. We continue to drive both basic and translational research in collaboration with start-ups and our partners in an Industry-Academia Consortium model, allowing members to leverage common knowledge while pushing specific applications of interest to them,” he adds.Experts

A good quantum of hope

“Hardware with 100+ logical qubits is now available. Hardware with a few 1000s of qubits is not that far, and we may see such machines in the next 2-3 years. On the software part, we will continue to see improvements in the quantum programming platforms,” says Mahajan.

“With the current noisy systems, we are only able to solve some artificial toy models without practical relevance. However, I am sure that in the next generation of systems we will be able to really realise quantum processors that make an impact and can outperform their classical counterparts on real world problems,” remarks Prof. Dumke.

According to a recent NASSCOM report, India too is gearing up for the quantum era, with about 10-15 government agencies, 20-30 service providers, 15-20 startups and 40-50 academic institutions active in this domain. It says that the industry would add about $310 billion to the Indian economy by 2030 and sectors such as manufacturing, high-tech, banking, and defence will likely lead the charge of adopting quantum technologies for critical and large-scale use cases.

Achyuta Ghosh, Research Head at NASSCOM, said in an Economic Times interview, that companies like Tata Consultancy Services, HCL Technologies, Infosys, Tech Mahindra, Zensar, Mphasis, and Coforge were creating use cases for quantum technologies and proof-of-concept for clients. There are also some interesting startups in this space such as QNu Labs, Taqbit Labs, QRDLab, and QpiAI Tech.

There are around a 100 quantum projects initiated in India, of which 92% are funded by the government. In August this year, the Ministry of Defence announced that the Indian Army had started the process of procurement of Quantum Key Distribution (QKD) technology developed by a Bengaluru based cybersecurity company by issuing a commercial request for proposal.

In a global survey of 750 companies conducted by KPMG early this year, a quarter of the business decision-makers said they already had quantum computing projects in place, one-third had either an internal team or external advisers looking into how they can use the technology, and 31% of organisations said they were discussing how they will leverage the technology in the future. KPMG reported that only 9% of the respondents said they were not thinking at all about how they can take advantage of quantum. We will leave you thinking about that!

Janani G. Vikram is a freelance writer based in Chennai, who loves to write on emerging technologies and Indian culture. She believes in relishing every moment of life, as happy memories are the best savings for the future


Janani G. Vikram is a freelance writer based in Chennai, who loves to write on emerging technologies and Indian culture. She believes in relishing every moment of life, as happy memories are the best savings for the future

SHARE YOUR THOUGHTS & COMMENTS

Electronics News

Truly Innovative Tech

MOst Popular Videos

Electronics Components

Calculators