Quantum Computing and Communications

Ten, maybe fifteen years ago, the term ‘Information Technology’ was not one that could be used without explanation. Most people, certainly in the developed world, knew what computers were, and what they could be used for. But that was only in terms of ‘end user’ applications, like word processing, performing mathematical calculations or creating a database. The fact that a computer simply stores and processes information is not a concept that appears to stick easily in the mind.

Compare the Babbage differential engine with a modern computer. The technological gap is obvious and yet there is essentially no difference between what the two machines can do - one simply does things faster than the other. The history of computer technology has involved a sequence of changes from one type of physical realization to another. Moving from gears to relays to valves to transistors to integrated circuits has gradually increased the computing speed, but the capability of the machines has not changed.

Currently, research is being undertaken on a number of possibilities for the implementation of the logic gates and networks discussed in the previous section, with the ultimate goal of implementing the remarkable algorithms that have been mentioned. However, all the technologies mentioned below will also have implications for quantum information processing in general. Further research may uncover other technologies which may provide an easier route for the implementation of quantum information processing.

In discussing the technology that can be developed for QCC, a useful starting point is to examine the ‘quantum engineering’ solutions that have been proposed in the field of optical telecommunications. In this field, the information is not encoded in terms of qubits that can take both values of 0 and 1 simultaneously, but rather as ‘classical’ bits, each having a definite value of 0 or 1. The aim of quantum engineering then is to preserve the classical value assigned to the bit in spite of the uncertainty that quantum mechanics inevitably introduces at every stage of the life of the optical signal: its generation, its transmission, its amplification, its distribution or its detection.

It is worth remarking at the outset of this chapter that, in all likelihood, most of the applications of quantum information processing are not yet even dreamed of. When few-qubit quantum devices become technologically feasible - which is an absolute pre-requisite for the entire field of quantum computation - possibilities other than those considered here will undoubtedly appear.

Demonstrations of principle, such as a quantum gate with a single spin or ion or the realisation of entanglement in various physical situations, constitute first steps towards quantum information processing that require up- or down-scaling to achieve effective processing of quantum information. Decoherence makes increasingly difficult the up-scaling of quantum information systems to a few spins or ions. This process destroys the correlations among the individual spins or ions, losing quantum information into the environment and introducing errors in any QIP routine.

Although I am a solid state experimentalist, I have spent most of last year dispensing with experiments and thinking exclusively about how I could create devices that could have an impact on quantum computation. I have even stopped worrying about particular devices and have started thinking about the general principles that might be useful for guiding us in thinking about which particular systems and devices could have an impact in this area.

I am aiming to give the perspective of an outsider. I’m not really into quantum computing - in some ways, I’m a critic, although I am a friendly critic. To begin with the conclusion, I would like to point out that, in this field, people are always making comparisons with computing history, stating that quantum computing has reached this point, or that point. Currently, I believe, quantum computing is where Charles Babbage was: he had a relatively prophetic, far fetched and complete vision of what you could do in a mechanical way for handling data. He did not, however, have the remotest chance of actually doing it in his time. The real breakthrough came in 1890 or thereabouts when Herman Hollerith put together the right technology with the right application - tabulating US census data with electrical sensing of holes in punched card. That’s the combination that started us on the whole road towards automated data processing.

One of my titles is ‘Director of Basic Research’ for Hewlett Packard Laboratories: it’s my privilege to conjecture what long-term research will benefit Hewlett Packard. I hire people, obtain resources for them and set rather loose boundaries for their research areas. I am a big fan of quantum information; I think it’s going to be very important scientifically and technologically, but I don’t really know what specific area to invest in now. I haven’t hired anyone to work directly on quantum information in my lab, even though HP as a company has a roughly $3.1 billion R&D portfolio devoted primarily to information technologies.

Let’s start off by looking at what we require to implement quantum computing. Any quantum computer needs four things:

I plan to explain how to do simple arithmetic operations, and then I want to demonstrate a simple example where we can easily see why quantum computing is more efficient than its classical counterpart. Just to remind you about looking at gates, Figure 12.1 is a simple network that accomplishes addition. Now this cannot be done on a quantum computer in this way, and there’s an easy way to see why that is true. Take the first gate as an example: we don’t even need to know what the gate does, what we need to see is that there are two inputs and one output. Therefore quantum information is lost on its way through the gate, and so this cannot be done reversibly. Since quantum computation is governed by unitary transformations, you need to do this reversibly. You simply would not be able to implement the first gate in Figure 12.1 on a quantum computer. The rule of thumb for doing this reversibly is that one has as many qubits coming into the gate as there are coming out of the gate. Let’s look at a very simple example: addends in this case.

At Los Alamos we have rather a different perspective - in some ways - to those working in a strictly university environment. We have to do things that are more application oriented in the long run, and quantum cryptography is an application that, at least from the point of view of physics, is feasible today.

The quantum computing system heralds the first fundamental change in the nature of computing since the age of modern computing dawned with the Turing machine The natural parallelism of superposition enables, in principle, computations to be carried out that are not now, or ever will be, practical with classical systems. For example, the application to factoring large numbers would have a revolutionary impact on cryptography. But in practice other applications, many as yet unrecognized, are likely to have an even bigger impact; for example, the ability to extract a single entry, fast, from a large database of unordered information would have a major impact on fields like feature (e.g. fingerprint) recognition, language translation and speech recognition by corpora matching, and the many-body problem (e.g. air traffic control) to name but a few. However, industrial economics dictates that a new approach is unlikely to obtain a market hold unless and until the benefit is very significant. In early applications the quantum computer might be used as a type of co-processor to the established classical hardware. But these will not take off in commercial terms until the number of qubits that can be processed provides a system that has indisputable benefits over the classical approach. This implies that there would be no mass market for normal computational purposes until the systems can handle large decimal numbers. (The actual number of qubits needed will probably be determined as much by the error correction requirements as by the numerical size of the task).

The subject of QIP may be said to have had its roots in Europe in the work of the great European physicists during the first 20 years of this century. Then when the subject became of interest again in the 1970s and early 1980s US work became of importance, notably that of Charles Bennett and Richard Feynman. But Europe was always well represented both in the fundamental thinking and in fundamental experiments. Indeed, in experiments Europe was leading in interferometry with individual quanta (electrons, neutrons, photons). These techniques are now important in QIP. In 1985, David Deutsch in Oxford moved the subject forward, and thereafter the theoretical advances have developed on both sides of the Atlantic. Quantum gate development has proceeded on both sides of the Atlantic. The USA has led in the work on algorithms (notably at Bell Labs), but the crucial work on Error Correction has been shared between work in Europe and USA. Experimental work has developed in the 1990s in parallel in USA and Europe, with significant teams in the USA, for example at Los Alamos and MIT, but with centres of experimental and theoretical work developing in Europe at, for example, Geneva, Innsbruck, Oxford and Paris. The strong optical physics community in Europe has helped to form a strong foundation for both experimental and theoretical work. The ion-trap quantum computer idea was first proposed in Innsbruck. NMR techniques, with the practical demonstration of quantum algorithms, seem to be being pioneered primarily by the two centres in MIT and Oxford.

In addition to identifying what I consider to be some worthwhile technical directions for the field, I thought it would be valuable to talk a little bit about my impressions of the quantum computing and quantum communications programme underway in the United States. In particular, I would like to focus on what works well and what does not and to see if there are any gaps that could be filled by a new, coordinated, European effort.

In the mid-1930s two influential but seemingly unrelated papers were published. In 1935, Einstein, Podolsky and Rosen proposed the famous EPR paradox that has come to symbolize the mysteries of quantum mechanics. Two years later, Alan Turing introduced the universal Turing machine and laid the foundations of the computer industry. Although quantum physics is essential to understanding the operation of transistors and other solid state devices in computers, computation itself has remained a resolutely classical process. Surely the uncertainty associated with quantum theory is seemingly not compatible with the reliability expected from computers. In 1982, Richard Feynman suggested that individual quantum systems could be used for computations. In 1985, David Deutsch from the University of Oxford described the universal quantum computer and showed that quantum theory can allow computers to do more rather than less. An important new observation is that information is not independent of the physical laws which govern the system used to store and process it (Landauer). On the atomic scale matter obeys the laws of quantum mechanics, which are quite different from the ones of classical physics that determine the characteristics of conventional computers. Therefore quantum computers will have qualitatively new properties and capabilities. During the past ten years scientific groups all over the world have worked to establish the theoretical foundations and to investigate different experimental realizations of quantum computing and quantum communications.

The following were suggested as part of the Pathfinder project’s goal of a preliminary analysis of this field. However, the field is constantly evolving, and the subdivision of the various research areas may prove to be inadequate or even incorrect in the not too distant future. Thus these definitions may soon seem incomplete: they are included here only as a guide to the subject areas within this somewhat amorphous field.