Quantum computing harnesses the power of three critical quantum principles: superposition, entanglement and interference Today’s computers are ubiquitous and extremely powerful, accomplishing a wide range of tasks in science, education, economics and everyday life. They are accessible to anyone who can afford a laptop or a mobile phone. Despite the phenomenal progress in power processing spurred by advances in microelectronics, however, the computer structure has basically remained unchanged since the Hungarian physicist and mathematician John von Neumann proposed the eponymous architecture based on stored programs and inspired by British mathematician Alan Turing, who laid the logico-mathematical foundations of computation and modern computer science. A typical digital computer system has four basic elements: input/output ports, main memory, control unit and arithmetic-logic unit (ALU). The limitations of that conventional computing structure in tackling the most difficult tasks, current supercomputers notwithstanding, are pushing researchers toward quantum computer development. Together with artificial intelligence, quantum computing represents one of the main technical and scientific challenges of the near future. The first mechanical calculating machines were invented by Blaise Pascal of France and Gottfried Wilhelm Leibniz of Germany in the 17th century, but American scientist John V. Atanasoff is credited with building the first electronic digital computer, which he constructed between 1939 and 1942 with the assistance of a graduate student. In 1946, J. Presper Eckert and John W. Mauchly, both of the University of Pennsylvania, built the ENIAC (electronic numerical integrator and computer), which was derived from Atanasoff’s machine; both computers used vacuum tubes in place of relays as active logic blocks, a characteristic that enabled a significant increase in processing speed. The digital computer roadmap is dotted with other important innovations, from the transistor to the integrated circuit and, finally, to microprocessors and VLSI circuits in the 1980s. These improvements have supported the empirical Moore’s Law, which predicts a doubling of a chip’s processing capability (or transistor density) roughly every 18 months. In 1959, American physicist and Nobel laureate Richard Feynman, known for proposing a new formulation of quantum electrodynamics (QED) that describes light-matter interaction, argued that as electronic components approach microscopic dimensions, strange effects predicted by quantum physics appear. Feynman believed those effects could be exploited in the design of more powerful computers. Arcane phenomena occurring only at the scale of atoms or particles are the basis of quantum computing hardware. Quantum computing harnesses the power of three critical quantum principles: superposition, entanglement and interference. These concepts play a pivotal role in the capabilities of quantum computers, which differ greatly from conventional computers. Superposition: In the world of quantum mechanics, objects such as particles do not necessarily possess clearly defined states, as demonstrated by the famous double-slit experiment. In this configuration, a single photon of light passing through a screen with two small slits will produce an interference pattern on a photosensitive screen akin to that generated by light waves; this can be visualized as a superposition of all available paths. If a detector is used to determine which of the two slits the photon has crossed, the interference pattern vanishes. The interpretation of this strange outcome is that a quantum system “exists” in all possible states before a measurement, introducing a fatal perturbation, collapses the system into one state; this is called decoherence. Reproducing this phenomenon in a computer promises to expand computational power exponentially. A traditional digital computer employs binary digits, or bits, that can be in one of two states, represented as 0 and 1; thus, for example, a 4-bit computer register can hold any one of 16 (24) possible numbers. In contrast, a quantum bit (qubit) exists in a wavelike superposition of 0 and 1 values; thus, for example, a 4-qubit computer register can handle 16 different numbers simultaneously. In theory, a quantum computer can therefore operate on a much larger number of values in parallel, so that a 30-qubit quantum computer would be comparable to a digital computer capable of performing 10 trillion (10 x 1012) floating-point operations per second (TFLOPS), a speed matched by a very fast digital supercomputer. Today’s fastest digital computer is the Frontier. Installed at the Department of Energy’s (DOE’s) Oak Ridge National Laboratory in Tennessee, it reached a speed of 1.1 exaFLOPS (1018 FLOPS). Supercomputers built on classical architecture are very complex and heavy machines, requiring an incredibly high number of parallel components and processors. The Frontier contains 9,408 CPUs, 37,632 GPUs and 8,730,112 cores, all linked together by 145 kilometers of cables. The supercomputer stretches over an area of 372 square meters (4,004 sq ft) and consumes 21 MW of power, reaching 40 MW at peak. Entanglement: Whereas superposition is the ability of a qubit to exist in multiple states simultaneously, i.e., in a state of 0, 1 or any combination of both, entanglement is the quantum phenomenon in which two or more qubits become correlated. In other words, the state of one qubit cannot be described independently of the state of its companion(s). This interdependence allows for instantaneous information sharing between entangled qubits, however distant they may be. Einstein dubbed this phenomenon “spooky action at a distance” to underscore his aversion to the nondeterministic and non-local nature of quantum mechanics. Entanglement is the backbone of many quantum algorithms, leading to faster and more efficient problem-solving. Interference occurs when two or more quantum states are combined to create a new state, resulting in either constructive or destructive interference. Constructive interference amplifies the probability of obtaining the correct output, while destructive interference reduces the probability of incorrect outputs. By manipulating interference patterns, quantum computers can quickly parse potential solutions, converging on the correct answer much faster than classical computers. But how can a qubit be built? Consider a single electron and its angular momentum, namely spin. Spin, being quantized, can be either up or down. By defining 0 as the spin-up state and 1 as the spin-down state, an electron can be used as qubit. It is useful here to use the bra-ket notation introduced by theoretical mathematician and physicist Paul Dirac. Quantum states can be represented by “kets” that are basically column vectors, so the two states can be written as |0> and |1>. The spin therefore plays the same role that a transistor does for a bit in standard Boolean logic. The superposition principle states that, unlike a classical bit, a qubit can be represented by a superposition of both 0 and 1 simultaneously. In mathematical notation, if |ψ> identifies the state of a qubit, this can be expressed as: |ψ> = W0 |0> + W1 |1> where W0 and W1 are two numbers representing the relative weight of |0> and |1> in the superposition. More formally, such numbers are complex probability amplitudes of the qubit and determine the probability of getting a 0 or a 1 when measuring the state of the qubit. Of course, they must obey the normalization condition: |W0|2 + |W1|2 = 1. When W0 = 1 e W1 = 0, the qubit is in its |0> state, corresponding to the off state of a transistor. If W0 = 0 and W1 = 1, the qubit state corresponds to the on state of the transistor. For any other values of W0 and W1, it is as if the transistor, in classical terms, is neither “off” nor “on” but simultaneously both “on” and “off” — much as the cat can be both dead and alive in the famous thought experiment conceived by Edwin Schrödinger, one of the founders of quantum mechanics. Superposition lets one qubit perform two calculations at once, and if two qubits are linked through entanglement, they can help perform 22 or four calculations simultaneously: |ψ> = W00 |00> + W01 |01> + W10 |10> + W11 |11> Three qubits can handle 23 or eight calculations in parallel, and so on. In principle, a quantum computer with 300 qubits could perform more calculations instantaneously than there are atoms in our visible universe. A quantum computer with this many qubits already exists. It is IBM’s 433-qubit Osprey, which hosts the most powerful quantum processor to date and is accessible as an exploratory technical demonstration on IBM Cloud. After codification of data in qubits, it is necessary to modify and manipulate the states of the qubits. In a digital computer, this is done by means of basic operations performed by logic gates such as AND, NAND and NOR. The corresponding operations in quantum computers are implemented by quantum gates, which can be classified depending on the number of qubits involved. As opposed to classical gates, quantum gates can create and manipulate entanglement and superposition, which are essential for the increased computational power of quantum computers. Some of the quantum gates performing operations on qubits through a set of quantum logic operations are Pauli-X, Pauli-Y, Pauli-Z, Hadamard and CNOT (controlled NOT). Pauli-X, for instance, is the quantum analog to the classical NOT gate. The Hadamard gate transforms a single qubit and into a perfectly balanced superposition of the |0> and |1> states, such that a measurement of a single qubit that has been “transformed” by this gate will produce either |0> or |1> with equal probability: W1 = W2 = 1/√2. In fact, (1/ √2)2 + (1/ √2)2 = 1.
Quantum processing unit
The core component of quantum computing hardware, a quantum processing unit (QPU), executes quantum algorithms by processing qubits through a series of quantum gates. Whereas conventional processors like CPUs, GPUs and DPUs (data processing units extensively used in data centers) exploit principles of classical physics, QPUs handle qubits, enabling quantum computers to perform complex calculations exponentially faster than their classical counterparts. QPUs can vary in their underlying technology, such as nuclear magnetic resonance, trapped ions, superconducting qubits and photonic chips, with each approach offering unique advantages and challenges. Because of the different implementations and architectures, it is not straightforward to compare QPUs by just looking at the number of qubits they handle. Quantum computing is emerging as potentially one of the most transformative technologies in the world, but the constraints are stringent. A quantum computer must maintain coherence between its qubits (or quantum entanglement) long enough to run a complete algorithm. Because of nearly inevitable interactions with the environment, decoherence may happen; therefore, robust methods of detecting and correcting errors need to be worked out. Finally, because the act of measuring a quantum system disturbs its state, reliable methods of extracting information must be devised. It is certain that we will enjoy another revolution in computational science. Many of today’s intractable problems could be addressed with the new machines. At the same time, the impressive capabilities of quantum computing have started a fierce competition among companies to achieve “quantum supremacy.”