TMCnet News
PROMISE OF quantum information(New Scientist Via Acquire Media NewsEdge) PROMISE OF quantum information Quantum economy Information encoded in polarised laser light can be transmitted entirely securely In recent decades, we have done very well in cramming ever more transistors into classical computer chips. But the density of heat generated by constantly resetting these physical on-off switches now represents a fundamental barrier to further miniaturisation. Quantum computation can sidestep that barrier. That's because by using the right manipulations you can flick between quantum states such as the polarisations of photons without expending any heat. This is no blank cheque for low-power computing, however. Reading and writing information to a quantum memory entails making measurements akin to flipping a classical switch, and will still generate some heat. Processing information in quantum states, rather than in the electrical currents of conventional computer chips, exploits strange quantum effects such as superposition and entanglement to offer the prospect of peerlessly powerful, economical and secure number crunching. That is the well-developed theory, at least. The challenge is to make it a reality Quantum security Quantum information can be used to make communication totally secure. The trick is that before exchanging an encrypted message by classical means, a sender and receiver (conventionally called Alice and Bob) first share an encryption key imprinted in quantum states. The first implementation of this "quantum key distribution", the BB84 protocol (see "History of an idea", opposite), proved the principle using an encryption key imprinted on photon polarisation states in superpositions. An attempt by an eavesdropper (Eve) to intercept the key collapses the superpositions. Provided the system has been designed carefully enough, if no such effect is seen, Alice and Bob can be sure only they have the key, which can be used to decrypt the classically exchanged message. Entanglement-based quantum cryptography, devised by Artur Ekert of the University of Oxford in 1991, relies further on "monogamy of entanglement". In this implementation, Alice and Bob's shared key is made of maximally entangled qubits with the greatest degree of correlation allowed by quantum theory. Eve cannot then entangle any of her own qubits with the key's qubits, and so cannot become party to the information they contain. history of AN IDEA The decade or so after physicist Richard Feynman first floated the idea of a quantum computer saw the theory of quantum information bloom. 1981 Feynman argues that modelling the correlations and interactions of particles in complex quantum physics problems can only be tackled by a universal quantum simulator that exploits those same properties. 1982 The no cloning theorem threatens hopes for quantum computing. It states that you cannot copy quantum bits, so there is no way to back up information. The plus side is that this makes intercepting data difficult – a boon for secure transmission of quantum information. 1984 Charles Bennett of IBM and Gilles Brassard of the University of Montreal in Canada develop BB84, the first recipe for secure encoding and transfer of information in quantum states (see "Quantum security", opposite). 1985 David Deutsch at the University of Oxford shows how a universal quantum computer might, in theory, emulate classical logic gates and perform all the functions of quantum logic. 1992 Superdense coding theory shows how a sender and receiver can communicate two classical bits of information by sharing only one entangled pair of quantum states. 1993 In fact you do not need to transmit quantum states at all to exploit their power, as quantum teleportation protocols prove: it is sufficient to possess entangled quantum states and communicate using classical bits. 1994 Shor's algorithm indicates how a quantum computer might factorise numbers faster than any classical computer. 1995 US physicist Benjamin Schumacher coins the term qubit for a quantum bit. 1996 Grover's algorithm gives a recipe by which quantum computers can outperform classical computers in an extremely common task: finding an entry in an unsorted database (see "Number crunching", page vi). 1996 Quantum error correction theory finally overcomes the no-cloning problem. Quantum information cannot be copied – but it can be spread over many qubits. With the problem of copying quantum bits finally solved, the main theoretical tools for quantum information processing were in place – but it remained to make something with them in practice (see pages iv and v). Quantum power So what makes quantum computers so different? Conventional, classical computers process information using the presence or absence of electrical charge or current. These classical bits have two positions, on (1) and off (0). Semiconductor switches – transistors – flip these bits, making logic gates such as AND, OR and NOT. Combining these gates, we can compute anything that is in principle computable. In quantum computation, the switching is between quantum states. Quantum objects can generally be in many states at once: an atom may simultaneously be in different locations or energy states, a photon in more than one state of polarisation, and so on. In general, quantum theory allows us to distinguish two of these states at any one time. In essence, a quantum bit, or qubit, is in a "superposition" that stores 0 and 1 at the same time. That already suggests an enhanced computational capacity, but the real basis of a quantum computer's power is that the states of many qubits can be "entangled" with one another. This creates a superposition of all the possible combinations of the single-qubit states. Different operations performed on different parts of the superposition at the same time effectively make a massively powerful parallel processor. The power increase is exponential: n qubits have the information processing capacity of 2nclassical bits (see diagram, left). A 400-qubit quantum computer would be like a classical computer with 10120 bits – far more than the number of particles estimated to exist in the universe. Physicist Richard Feynman was the first to recognise the potential of quantum computing Quantum limits Computer scientists divide problems into "easy" problems, where the computational resources needed to find a solution scale as some power of the number of variables involved, and "hard" problems, where the resources needed increase in a much steeper exponential curve. Hard problems rapidly get beyond the reach of classical computers. The exponentially scaling power of a quantum computer could bring more firepower to bear, if not making the problems easy, then at least less hard. A quantum speed-up is not a given, however: we first need a specific algorithm that tells us how achieve it. For important, hard tasks such as factoring large numbers or searching a database, we already have recipes such as Shor's and Grover's algorithms (see "Number crunching", page vi). But for a mundane task such as listing all the entries in a database, the time or processing power required to solve the problem will always scale with the number of entries, and there will be no appreciable quantum speed-up. All this presumes a quantum computer big enough to make a difference. As we shall see on the following pages, however, making one is easier said than done. "Quantum computers could make all problems if not easy, then at least less hard" Building a quantum computer QUBIT: Superconducting states QUBIT: Cold atoms Collections of many hundreds of atoms might make for good qubits when trapped, cooled and arranged using lasers in a two-dimensional array known as an optical lattice. The energy states of these atoms can encode information that can be manipulated using further lasers, as with trapped ions (see opposite page). We've mastered the basic techniques, but making a true quantum computer from cold atoms awaits establishing reliable entanglement among these aloof bodies. At temperatures of a few kelvin, electrons in some superconducting materials form entangled pairs, known as Cooper pairs, which flow without resistance and are peculiarly resilient against decoherence. Superconducting quantum interference devices (SQUIDs) already exploit this effect to make incredibly sensitive measurements of electromagnetic fields. But electron movements and magnetic field states within a SQUID can also be manipulated using external fields to form the bits of a quantum logic device. SQUID qubits offer good initialisation and decoherence times, typically 10 or so times greater than the time taken to switch a gate. With large numbers of qubits, however, heating due to the external fields used for manipulation becomes an issue, and the largest verified system remains at just three qubits. The company D-Wave Systems in Burnaby, British Columbia, Canada, has claimed since 2011 to have a 128-qubit computer in operation (pictured, right); but it remains controversial whether this device is fully quantum, and if it can implement all quantum logic operations. There are many ways of making the "qubits" for a quantum computer to crunch, from polarising light to cooling atoms to taming the collective motions of electrons. But any qubit must fulfil some stringent criteria, particularly in proving robust, or "coherent", in the face of buffeting from its surrounding classical environment. No single sort of qubit has yet ticked all the boxes QUBIT: Nuclear spins Nuclear spin states manipulated using magnetic fields were among the first qubits explored. In 1998, the first implementation of Grover's algorithm (see "Number crunching", page vi) used two nuclear magnetic resonance qubits to seek out one of four elements in a database. The great advantage of spin states is that they make qubits at room temperature, albeit with a very low initialisation accuracy of about one in a million. But the disrupting effects of thermal noise on entanglement means that nuclear-spin computers are limited to about 20 qubits before their signal becomes washed out. A variant on the spin theme exploits nitrogen impurities in an otherwise perfect diamond (carbon) lattice. These introduce electrons whose spin can be manipulated electrically, magnetically or with light – but scaling up to anything more than a couple of spins has proved difficult. WHAT MAKES for A GOOD QUBIT? "To out-gun a classical computer, we must entangle hundreds of qubits. So far we have managed a handful" In 1997, David DiVincenzo of IBM wrote down some desirable conditions that remain a rough, though not exhaustive, checklist for what any practical quantum computer must achieve. Scalability To out-gun a classical computer, a quantum computer must entangle and manipulate hundreds of qubits. Quantum computers built so far have just a handful. Scaling up is a big hurdle: the larger the system, the more prone it is to "decohere" in the face of environmental noise, losing its essential quantumness. Initialisation We must be able to reliably set all the qubits to the same state (to zero, say) at the beginning of a computation. Coherence The time before decoherence kicks in must be a lot longer than the time to switch a quantum logic gate – preferably, several tens of times. In most practical implementations so far this requires an operating temperature near absolute zero to limit the effects of environmental interference. Accuracy The results of manipulations must be reproduced accurately by the qubit, even when many manipulations are applied in sequence. Stable memory There must be a reliable way to set a qubit's state, keep it in that state, and reset it later. QUBIT: Atom-light hybrids Cavity electrodynamics is a quantum computing approach that aims to combine stable cold atoms with agile photons. Light is trapped inside a micrometre-scale cavity and atoms sent flying through, with logical operations performed through the interactions of the two. Initialisation is highly efficient, and the decoherence time allows 10 or so gate operations to be performed – although scaling the technology up awaits reliable ways of entangling trapped cold atoms. Serge Haroche of the Collège de France in Paris, one of the pioneers of this approach, shared the 2012 Nobel prize in physics with trapped-ion researcher David Wineland (pictured, right). QUBIT: Photons QUBIT: Trapped ions The position, polarisation or just number of photons in a given space can be used to encode a qubit. Though initialising their states is easy, photons are slippery: they are easily lost and do not interact very much with each other. That makes them good for communicating quantum information, but to store that information we need to imprint photon states on something longer-lived, such as an atomic energy state. If we can nail that, quantum computing with photons is a promising concept, not least because the processing can be done at room temperature. In 2012, a team at the University of Vienna, Austria, used four entangled photons to perform the first blind quantum computation. Here a user sends quantum-encoded information to a remote computer that does not itself "see" what it is crunching. This may be a future paradigm – totally secure quantum cloud computing. Trapping ions is perhaps the most advanced method of making a quantum computer's qubits. Positively charged ions are caught in electromagnetic fields and cooled to a nanokelvin or so to reduce their vibrations and limit decoherence. Information is then encoded in the ions' energy levels and manipulated using laser light. That brings excellent initialisation success (99.99 per cent), accuracy (over 99 per cent) and stable memory storage (years). In 1995 David Wineland and his colleagues at the US National Institute of Standards and Technology in Boulder, Colorado, used trapped ions to create the first quantum logic gate – a controlled NOT (C-NOT) gate for disentangling entangled ions. In 2011, physicists from the University of Innsbruck, Austria, developed a 6-qubit trapped-ion computer that fulfilled the specifications for a universal quantum simulator that Richard Feynman had set out in 1981. Decoherence and scalability remain interrelated problems, however. With a few entangled qubits the decoherence time is 1000 times the gate-switching time, but this rapidly reduces as qubits are added. Using lasers to trap ultracooled atomic ions is a well-developed way to make a small-scale quantum computer Quantum computing pioneers David Wineland (left) and Serge Haroche shared the 2012 Nobel prize in physics QUBIT: Topological states This promising basis for a quantum computer has yet to get off the theoretical drawing board, because it depends on the existence of particles confined to two dimensions called anyons. These "topological" particles are peculiarly impervious to environmental noise, in principle making them excellent qubits. Particles such as Majorana fermions that fulfil some of the requirements of anyons have been fabricated in certain solids, but whether they are useful for practical quantum computing is still debatable. KILLER quantum APPS Quantum SIMULATION Richard Feynman's original motivation for thinking about quantum computers in 1981 was that they should be more effective than classical computers at simulating quantum systems – including themselves. This sounds a little underwhelming, but many of science's thorniest practical problems, such as what makes superconductors superconduct or magnets magnetic, are difficult or impossible to solve with classical computers. Quantum information theorists have already developed intricate algorithms for approximating complex, many-bodied quantum systems, anticipating the arrival of quantum computers powerful enough to deal with them. The beauty is that such simulators would not be limited to existing physics: we could also use them to glean insights into phenomena not yet seen. Quantum simulations might tell us, say, where best to look in nature for Majorana particles, for example in complex many-bodied superconductor states. Since these particles, thought to be their own antiparticles, have properties that could make them ideally suited to making robust quantum memories (see "Qubit: Topological states", page iv), this raises the intriguing possibility of using quantum computers to suggest more powerful quantum computers. The theory is in place, and we have no shortage of ideas as to how we can physically implement a quantum computer. But what might we use them for if we did? There are many suggestions – some practical, some highly fanciful Number crunching ULTRASECURE ENCODING The promise of quantum computers rests largely on two algorithms. One, developed in 1994 by Peter Shor, then of Bell Laboratories, provides a way for a quantum computer to speedily find the prime factors of large numbers. Classical computers effectively have to try to divide the given number by all possible prime factors (2, 3, 5, 7, 11 and so on) in turn, whereas quantum computers can do these divisions simultaneously. Conventional encryption methods rely on the fact that classical computers cannot factorise efficiently. If Shor's algorithm were ever implemented on a large scale, encrypted information such as the PIN for your bank card would be vulnerable to hacking – and quantum cryptography would be the only viable defence (see "Ultrasecure encoding", opposite). There is no need to worry just yet: demonstrations so far, for example using a 7-qubit nuclear-spin quantum computer, have been limited to demonstrating that the prime factors of 15 are 5 and 3. In the longer term, an algorithm devised by physicist Lov Grover in 1996, also at Bell Labs, may become a quantum computer's greatest selling point. This provides a recipe by which a quantum computer can radically speed up how we access and search large bodies of data. Take the example of a database listing the contents of a library. Searching this database for a particular book with a classical computer takes a time that scales with the number of books, n ; Grover's algorithm shows that for a quantum computer it scales with vn. For a library of a million books, this amounts to 1000 times faster. Implementing such an algorithm has ubiquitous appeal: almost all computationally hard problems – for instance that of the travelling salesman who has to find the shortest route between a number of different cities – ultimately reduce to a search for the optimal solution.There's a way to go yet. The biggest Grover search yet performed, with 3 qubits, allows for a search of just 8 database elements. One quantum information technology is already up and running. Various small-scale quantum cryptographic systems for secure information transfer, typically using polarised photons as their qubits, have been implemented by labs and companies such as Toshiba, Hewlett Packard, IBM and Mitsubishi. In October 2007, a quantum cryptography system developed by Nicolas Gisin and his colleagues at the University of Geneva in Switzerland was used to transmit votes securely between the city's central polling station and the counting office during the Swiss national elections. A similar trial system developed by the researchers' company, ID Quantique, was used to transmit data securely during the 2010 Football World Cup in South Africa. The distance through which quantum states can be transmitted through fibre-optic cables is limited to tens of kilometres owing to random diffusion. One promising way to get around this is akin to error correction protocols devised for quantum computers: to spread information over more than one qubit (see "History of an idea", page iii). But this might pose a security risk by giving more information for an eavesdropper to hack. Transmission via air is an alternative. The world record in faithfully teleporting a qubit of information, held by Anton Zeilinger of the University of Vienna, Austria, and his colleagues, is over a distance of 143 kilometres between the Canary Islands of La Palma and Tenerife. This indicates that delicate quantum states can be transmitted significant distances through air without being disturbed – and suggests that a worldwide secure quantum network using satellites is a distinct possibility. metrology Making precise measurements is a potentially highly significant application of quantum computers. When we record sensitive measurements of physical quantities, such intervals in time or distances in space, the effects of classical noise mean that the best statistical accuracy we can achieve increases with the square-root of the number of bits used to make the recording. Quantum uncertainty, meanwhile, is determined by the Heisenberg uncertainty principle and improves much more rapidly, simply with the number of measurements made. By encoding distances and time intervals using quantum information – probing them using polarised laser photons, for example – much greater accuracies can be achieved. This principle is already being applied in giant "interferometers" that use long laser beams in a bid to detect the elusive gravitational waves predicted by Einstein's relativity, such as the LIGO detector in Livingston, Louisiana (pictured, left). In these cases we can think of gravity as noise that disturbs qubits – the qubits being the position and momentum of laser photons. By measuring this disturbance, we can estimate the waves' strength. "A quantum computer could search the database of a million-book library 1000 times faster than a classical computer" Classical encryption methods depend on the difficulty of finding the prime factors of large numbers Vlatko Vedral Vlatko Vedral is a professor of information theory at the Oxford Martin School of the University of Oxford and at the National University of Singapore. He has published over 200 papers on quantum physics, and is currently focused on bio-inspired quantum-information technologies ARE WE nearly there yet? No overview of quantum computing would be complete without an attempt to answer the $64,000 (or possibly much more) question: are we likely to see working quantum computers in our homes, offices and hands any time soon? That depends largely on finding a medium that can encode and process a number of qubits beyond the 10 or 20 that current technologies can handle. But getting up to the few hundred qubits needed to outperform classical computers is largely a technological issue. Within a couple of decades, given improvements in cooling and trapping, as well as coupling with light, existing technologies of trapped ions and cold atoms may well be made stable enough in large enough quantities to achieve meaningful quantum computation. The first large-scale quantum computers are likely to be just that: large-scale. They will probably require lasers for qubit manipulation and need supercooling, so are unlikely to appear in our homes. But if the future of much computing is in centralised clouds, perhaps this need not be a problem. When it comes to anything smaller, the elephant in the room is entanglement, which is a fragile good at the best of times and becomes harder and harder to maintain as the quantum system grows. It would aid the progress of quantum computing if our assumption that entangled states are an essential, central feature turned out to be wrong. This intriguing possibility was raised in 1998 with the development of "single-qubit" algorithms. These can solve a large class of problems, including Shor's factorisation algorithm, without the need for many entangled qubits. That would be a remarkable trick if it could be pulled off in practice – although Grover's all-important database-search algorithm might still not be implementable in this way. Some people believe that the fragility of quantum systems will never allow us to implement quantum computation in the sort of large, noisy, warm and wet environments in which we humans work. But we can draw hope from recent evidence that living systems – such as photosynthesis in bacteria and retinal systems for magnetic navigation in birds – might be employing some simple quantum information processing to improve their own efficiency. If we can learn such secrets, a quantum computer on every desktop and in the palm of every hand no longer seems so fanciful an idea. recommended READING Quantum Theory: Concepts and Methods by Asher Peres (Springer, 1993) The Fabric of Reality by David Deutsch (Penguin, 1998) Quantum Computation and Quantum Information by Michael A. Nielsen and Isaac L. Chuang (Cambridge University Press, 2000) Introduction to Quantum Information Science by Vlatko Vedral (Oxford University Press, 2006) Programming the Universe by Seth Lloyd (Alfred A. Knopf, 2006) Decoding Reality by Vlatko Vedral (Oxford University Press, 2010) (c) 2013 Reed Business Information - UK. All Rights Reserved. |