Quantum computing
Much more than quantum behaviour of the components is needed for successful quantum computing, and although the designs discussed here do use single atoms, the scale length is an order of magnitude greater than the 250 nm of present day integrated circuits.
Abstract
In 1965 Gordon Moore, one of the founders of the Intel Corporation, noticed that semiconductor technology was advancing so fast that the density of devices in an integrated circuit was doubling every year. Moore was interested in the commercial applications of integrated circuits, and used his observation to predict that within 10 years electronic devices would no longer be considered as high value commodities, a prediction fully justi® ed by history. He later identi ® ed another more interesting consequence of his lawÐ the doubling of the density of devices in integrated circuits every year. This is not a paradox, but a consequence of Japanese business practice. Once the large Japanese electronics ® rms knew what was expected they planned their development to achieve it, and their feedback into the industry ensured that only companies which can maintain this level of development stay in business. As a consequence, Moore’s law is still valid, although the eŒort required to maintain it is growing, and the doubling time is now about 2 years. The present scale length for devices is now about 250 nm. If the rate of growth can be maintained for the next 35 years, the scale length will be under 1 nm, so small that individua l devices will require quantum mechanics for their correct description. It is not yet known what this will mean for the development of computers, but one possibility which must be taken seriously is that quantum computation will emerge. It is certainly true that many researchers in this ® eld cite Moore’s law to justify the urgency of their work, and Quantum Computing is no exception. However, as is clear from the text, much more than quantum behaviour of the components is needed for successful quantum computing, and although the designs discussed here do use single atoms, the scale length is an order of magnitude greater than the 250 nm of present day integrated circuits. In order to understand what is needed we must look at how quantum computing diŒers from classical computing. The fundamental unit of a classical computer is a logical gate. Typically this gives a binary output (`0’ or 1̀’ ) which depends upon some logical function of its two binary inputs. For example, the exclusive OR (XOR) gate gives an output of 1̀’ if its two inputs are diŒerent, otherwise it gives an output of `0’ . By appropriately connecting sets of gates, using the outputs of some as inputs to others, sequences of bit patterns may be transformed to give new sequences. At one level this is all that computers do. Computer hardware designers ensure that the bit patterns of zeros and ones can be represented by electrical pulses, that structures etched into silicon wafers respond to these pulses as the gates are supposed to, and that the structures are connected so that the correct combination of gates results. Of course the bit patterns are interpreted as numbers, and the combinations of gates are the hardware for numerical processing. Exactly how numbers are encoded as bit patterns, which types of gate are best to use, what basic transformations are useful, how information should be transferred around the computer and how the inevitable errors are corrected are all subjects which can pro® tably be discussed at this basic level. However, one may move inwards and discuss the physics of the individua l gate operation. This already requires quantum mechanics for its detailed understanding Ð indeed since solidity is a quantum phenomenon, a proper understanding of any solid state device cannot be had without quantum mechanicsÐ but as gates become smaller, and their electronics require fewer and fewer electrons, pessimists fear that the inevitable uncertainty due to quantum mechanics will become more bothersome. Optimists hope to exploit it. As well as moving inwards to a deeper understanding of the workings of a computer at a physical level, we can move outwards and try to understand its workings at an algorithmic level, that is to understand what kind of calculations it could do. Perhaps surprisingly, the computer’ s capability does not depend upon the detailed workings Dr P. T. Greenland is at Imperial College, London. Contemporary Physics, 2001, volume 42, number 4, pages 239± 241