Quantum computing has promised much, and absorbed enormous resources, without delivering a great deal in practical terms yet. However, the rationale for investing in research remains impeccable.
In theory, a quantum computer
could work much faster than any classical computer, enabling it to tackle currently impossible problems. In practice too, advances in quantum computing have equated to better understanding of many scientific phenomena.
The engineering problems are formidable. Superposition and entanglement are at the heart of quantum computing. At the quantum level, a particle can be in two states at the same time (or rather, it has probabilities of being in two states — this is “superposition”). Particles can be entangled, which means that a change in the state of one particle instantly causes corresponding change in the state of its entangled twin, even at a distance.
These two fundamental qualities offer the potential to make quantum computers work faster. Classical computers like your PC, or smartphone, work on binary principles with circuits switched on or off using “logic gates”. A quantum can be switched on and switched off, both at the same time, due to superposition.
Hence, while one classical bit can have either the value or a quantum bit, or qubit can have both values at the same time. A 2-qubit circuit can be in four possible states of superposition at a given moment, whereas a 2-bit circuit can be in only one of those four states. So a large qubit processor can process exponentially more information than equivalent classical circuits.
As the Boston Consulting Group (BCG) says, “Modeling the structure of penicillin, which has 41 atoms at ground state, requires a classical computer with 1086 bits, which is more transistors than there are atoms in the observable universe. Such a machine is a physical impossibility. But for quantum computers, this requires a processor with 286 logical qubits.”
However, every time quantum output is measured, it breaks down (“decoheres”) into a single measured state. To hold an entangled qubit circuit together requires a lot of work, since even small external disturbances causes coherence to break. Also quantum computers generate huge amounts of data, and a lot of error correction is required.
Currently, qubit chips are sealed in vacuum boxes and cooled to nearly absolute zero temperatures, which is only possible in tightly controlled lab environments. They even have to be shielded from vibrations since footsteps can destroy coherence.
Computer scientists use “logical” qubits as the basic “currency”. A logical Qubit can run two superposed states. It could consist of several physical qubits, or superconducting circuits that model the state of an atom. Indeed, a classical computer may simulate a quantum machine by imposing a structure of logical qubits on top of classical circuits. This is often done to test quantum algorithms.
There have been steady advances. Google’s Bristlecone is a 72-physical Qubit processor, currently the world’s largest. The University of Science and Technology of China has demonstrated 18 Qubit entanglement. A research team from the University of New South Wales (Sydney) has dramatically improved processing speed in a 2 Qubit gate, by embedding phosphorus atoms in silicon.
To achieve “quantum supremacy” that is beat classical computers at solving problems, more breakthroughs are necessary. It’s generally assumed that a stable 50 logical Qubit processing power is the minimum required to achieve supremacy. To calculate large molecules or run an AI Machine-learning algorithm could take much more, perhaps a 1,000 logical qubit machine. Breaking standard RSA 2048 encryption could take a 4,000-qubit machine.
Another issue is noise. A large qubit register could, in theory, produce more data than the Large Hadron Collider and a lot of that data would consist of errors, which would need corrections. Programming quantum algorithms is also more complex than algos for conventional machines. Even loading large amounts of data onto a quantum machine is a tricky problem.
The BCG recently came up with a set of projections. It believes that productivity gains from quantum computing will surpass $450 billion annually. But “it will be a slow build for the next few years: we anticipate value for end users to reach a relatively modest $2 billion to $5 billion by 2024”. Value will then increase rapidly as the technology matures. The benefits will be uneven and disproportionately favour early adopters.
Quantum computing could revolutionise sectors such as drug design, network optimisation (ride hire companies, airlines, railways), supply chain management, portfolio management, weather and climate change predictions, and encryption/ decryption methods. Research has led to new understanding about various exotic materials and the nature of quantum entanglement. But commercial viability will be a while coming.