| Number of | In text | In summary | Ratio |
|---|---|---|---|
| Sentences | 167 | 7 | 4.191617 |
| Paragraph | 22 | 6 | 27.272728 |
The bits in a " quantum computer " would not , therefore , be ones or zeros . Dr Monroe 's group trapped a single ion ( an atom with missing electrons ) of beryllium , by walling it in with electric and magnetic fields and cooling it to within an ace of absolute zero ( - 273C ) . Why should DARPA , better known for researching missile defences and commissioning the precursor of the Internet , dabble in quantum theory ? Qubits , not being true bits , are not truly digital . One of the big advantages of recording and processing information digitally is that it is difficult to make a mistake ( electronic " ones " are not that easily turned into " zeros " ) , and surprisingly easy to design ways to recognise and correct any mistake that is made .
A quantum computer would rely on the surreal behaviour of the very small to work miracles with information . There is new hope that it might someday be more than fantasyTHE computer has become so widespread , its applications so boundless , that it is easy to forget that it is just a machine , constrained like everything else by the laws of physics . But a machine is all it is , and a pretty simple one at that . The underlying logic of the programs - - the binary choices that control the way a computer counts and calculates - - would work just as well ( if a lot more slowly ) in a device that shuttled ball - bearings instead of electrons .
What , though , if a computer could be built around a different kind of physics , the quantum physics that governs the weird and uncertain behaviour of atoms and sub - atomic particles ? A group of physicists and mathematicians that is working on this question believes that quantum computing could change computing just as thoroughly as quantum physics changed the Newtonian certainties of classical physics earlier in the 20th century .
Naturally , building a useful quantum computer would be difficult . Only the very smallest objects behave in a detectably quantum way . This means that the components of such a computer would be very tiny and very delicate . But so great is the theoretical appeal of the machine that America 's Defence Advanced Research Projects Agency DARPA has just created an Institute for Quantum Information and Computing , and given it $5m to investigate the possibilities .
Bit players What makes quantum computing so attractive ? For some jobs , a quantum computer would be able to put even the best of today 's supercomputers in the shade . It would do this by employing a trick that ordinary computers cannot truly master - - parallel processing . Although supercomputers are fast , they can , like all other digital computers , do only one thing at a time . Even the so called " parallel " computers now available are just collections of machines that each do only one thing at a time . But a quantum computer would be truly parallel . It would be able to do many different things simultaneously in the same piece of equipment .
To see why , remember that every digital computer , no matter how wondrous its applications , consists merely of a lot of switches that are linked together and are either on or off . These switches can be thought of as the digits of numbers , in which case their ons and offs are the ones and zeros of binary counting . This is why , by a fortunate contraction of the words " binary digit " , the currency of data processing is known as the " bit " .
The switches inside a computer can also be interpreted as the trues and falses of a form of logic known as Boolean algebra . Indeed , they are organised into " logic gates " that do calculations using this algebra . A digital computer is mainly a machine for altering bits by running them through logic gates . The gate known as AND , for example , compares two bits . If both have a value of " one " , it creates an output of one ; otherwise it creates a zero . With the next tick of its internal clock , the computer then passes this output bit to another gate for further processing . Each tick , then , corresponds to one step of the calculation .
In the quantum world , however , the rules are different . A quantum computer 's switches would not have to be either on or off . Like any quantum object which can come in several distinct states , a quantum switch can carry on indefinitely as a combination of all those states . It is not merely stuck halfway between on and off ; it is actually on and off simultaneously , as if it existed in two parallel worlds .
The bits in a " quantum computer " would not , therefore , be ones or zeros .
They would be quantum combinations of one and zero . Such vacillating pieces of information are known as " qubits " ( pronounced as " cubits " , though they are usually rather smaller than the ancient Hebrew unit of length ) .It is this simultaneous existence in many states that would give quantum computing its power . An ordinary computer having ( say ) ten bits could exist in only one of 1,024 states ( all the ten - digit sequences that can be made from ones and zeros ) available to it , at each instant . A quantum computer with ten qubits could exist in all 1,024 states at the same time . It could therefore work on 1,024 calculations at once .
But qubits are fiendishly hard to make . For a quantum switch to maintain its multiple existence depends on one big proviso . Nothing can happen that might disturb it . No light must shine on it , no stray air molecules may bump into it , no inquiring probe can intrude on it . If it is disturbed , it stops vacillating and chooses a definite state to be in . This choice depends on chance and on the way its multiple personalities have interacted until then . If it happens prematurely , the computation will be ruined .
This proviso has made technical progress agonisingly slow . Paul Benioff , at Argonne National Laboratory , in Illinois , first applied quantum theory to computers in 1981. David Deutsch , at Oxford University , proposed the possibility of quantum parallel computers in 1985. Yet both had to wait until last year for the first two qubits to come on - line .
One group of quantum mechanics who created them was led by Chris Monroe , who works at the National Institute of Standards and Technology , in Boulder , Colorado . Dr Monroe 's group trapped a single ion ( an atom with missing electrons ) of beryllium , by walling it in with electric and magnetic fields and cooling it to within an ace of absolute zero ( - 273C ) . Such an ion has two energy levels - - providing the first qubit . And it can vibrate in two ways within the trap - - the second qubit . These qubits are linked together via a pulse of laser light , which can switch the ion from one energy level to the other . Whether it does so or not depends on which of the two vibrational patterns the ion is in .
The result is a logic gate called XOR . This has two input bits , and reverses the value of the second if the first is " one " . That is an important step . Though using many types of gate makes designing ordinary computers easier , it is possible to make do with just three: AND , NOT and COPY . A quantum computer is even simpler . It requires only two sorts of gate , and XOR is one of them. ( The other would be able to change a zero into an arbitrary combination of zero and one . This has no equivalent in conventional computing. )
Qubits can also be made out of light , because light can be polarised in two perpendicular directions . Jeff Kimble , a physicist at the California Institute of Technology ( Caltech ) , in Pasadena , and the head of the DARPA quantum - computing institute , is working towards a light - based device . He and his team have designed a piece of apparatus that allows photons ( the particles of which light is composed ) to interact while they fly through a stream of caesium atoms . This interaction could in principle form the basis of logic gates like XOR .
Trapping ions , though , may ultimately prove to be the best way to make real quantum computers . Such devices would require thousands of ions , vibrating in synchrony , and are certainly still in the realms of science fiction . But ion traps housing up to six ions have already been built during other atomic - physics projects . Groups at the Los Alamos National Laboratory , in New Mexico , and the University of Innsbruck , in Austria , are trying to improve upon them to connect a handful of qubits together .
Qubits v true bits Why bother ? Why should DARPA , better known for researching missile defences and commissioning the precursor of the Internet , dabble in quantum theory ? Perhaps to make better calculators . There are already three mathematical challenges known in which quantum computers would beat the pants off the ordinary kind .
The first is up DARPA 's alley: cracking secret codes . The security of most " public key " coding schemes , in which people can exchange secrets without advance planning , relies on inscrutable large numbers . It is hard to factorise - - to find numbers that multiply together to produce - - some 200 digit numbers . But it is these factors that unlock the code . Since the programs for finding them are excruciatingly tedious , even for a fleet of supercomputers , such codes are pretty safe .
A quantum computer , on the other hand , would make short work of this task . In 1994, Peter Shor , a mathematician at AT&T Labs , in Murray Hill , New Jersey , designed a quantum circuit that could achieve it in many fewer steps . His quantum software arranges the multiple personalities of the qubits in a way that enables them to attack the problem separately , and then conspire together so that when they are , in the end , forced to behave like normal bits , they will be likely to produce the correct answer .
Like much of quantum theory , this can be properly explained only in mathematics , not words . But a sense of what it entails can be gained by recognising that the equations of quantum theory are not unlike those of wave motion , as Seth Lloyd , of the Massachusetts Institute of Technology , has suggested . Each parallel path of a qubit is analogous to a sound wave of a particular pitch . When different tones play together , they interfere , reinforcing or cancelling one another depending on their strengths and timing . The parallel lives of qubits do the same . And the strongest tones are the ones likeliest to be chosen by the qubits when they are measured .
In this analogy , an ordinary computation can be thought of as a melody of single notes plinked out on a piano . A quantum computation , with its complex array of chords and rhythms , sounds more like the Berlin Philharmonic . Dr Shor 's trick was to arrange the concert hall so that the music was completely dominated by the proper overtones and beats - - the rich timbre of the French horns , say - - that correspond to the correct solution .
Lov Grover , a computer scientist at Bell Labs ( which recently split from AT&T Labs ) , has written quantum symphonies to solve two other problems . One , published in May , is a way to search unsorted databases . If you were to sift through 10,000 pieces of paper scattered about your desk in search of one important memo , you would have to be prepared to scan 5,000 of them to have an even chance of finding it . By assigning each parallel computation to pursue a different choice , a quantum computer could do the same in only 40 runs .
Dr Grover 's other discovery , announced this month , is that quantum computers would be talented statisticians . They would excel at finding single numbers which depend collectively on lots of data: the median age of a large population , for example . If a quantum computer and an ordinary computer with the same internal clock rate raced to do it , the quantum computer would win hands down .
Back to the future There is , however , a snag . Qubits , not being true bits , are not truly digital . One of the big advantages of recording and processing information digitally is that it is difficult to make a mistake ( electronic " ones " are not that easily turned into " zeros " ) , and surprisingly easy to design ways to recognise and correct any mistake that is made .
This ease of correction was shown in the 1940s, at a time when the first digital computers were being constructed . Claude Shannon , who also worked at AT&T, set the ball rolling by investigating ways of encoding bits of information so that they would resist errors during transmission down telephone lines . In the 1950s, John von Neumann , one of the mathematicians who helped to develop the theory of these early computers , went on to show how a reliable computer could be built out of unreliable parts , by using Shannon 's ideas and combining them with extra ( so - called redundant ) circuitry .
The simplest error - correcting code is triple - redundant . Each bit is copied into two more bits: 0 becomes 000, and 1 becomes 111. If one bit accidentally flips , the other two indicate this and can be used to fix it . But Rolf Landauer , a physicist at the IBM Watson Research Centre , in Yorktown Heights , New York , has insisted that this cannot work with qubits , which cannot be measured , let alone copied . Yet error - correction would be vastly more important in a quantum computer than an ordinary computer , since the disturbance of even one qubit could ruin the coherence of the whole computation . If quantum computing is to become more than a dream , it needs a new way to correct errors as well as new hardware .
It was therefore a welcome surprise when Dr Shor announced a year ago that he had invented a quantum error - correcting code . Instead of three qubits , Dr Shor 's encoding formula required nine . A few months later other researchers at IBM , Los Alamos and Oxford came up with codes that work with five of them .
The idea begins with the qubit that needs to be protected from errors , and four other qubits , set to zero . A sequence of logic gates changes and tangles all five bits into a larger quantum combination . Just as one qubit is a mix of two different states , two qubits are a mix of four , three are a mix of eight , and so on . Five entangled qubits span 32 possibilities - - each of the five - digit sequences that can be made of ones and zeros .
Five qubits are enough , though , to preserve the quantum ambiguity of what is going on , even if one of them goes bad . One qubit being unmasked by , for example , a collision with a stray air molecule , does not reveal enough information for an observer to work anything out about the qubit originally used to make the mixture . In these circumstances , the calculation can be corrected , and proceed unhindered . The original mixing , in other words , creates quantum cannon fodder for such interactions as would otherwise spoil a combination . Some losses become tolerable without the game being given away .
There are still problems to be overcome , of course . It is necessary to design logic gates that will work on entangled bundles of five qubits at a time . And errors can still creep in during the encoding process itself . Dr Shor 's latest achievements , to be announced at the Symposium on the Foundations of Computer Science , being held in Burlington , Vermont , in October , address both of these concerns . First , he has designed logic gates for protected qubits by combining lots of two - qubit logic gates . Second , he has shown , theoretically , that the improvements in components needed to make quantum computing reliable should not , once such components actually exist , be too onerous .
His second result is analogous to the ideas in one of von Neumann 's achievements , the " Synthesis of Reliable Organisms from Unreliable Parts " . Early sceptics of computing thought that allowing computers to work , say , ten times longer on a problem would require components that were ten times as reliable . Von Neumann showed that , with the shrewd use of error correcting codes and redundant logic gates , the components did not need to be any more reliable . All that was necessary was that the speed of the calculation be slowed down to allow for error - correction . Dr Shor 's result is not quite as powerful as this . He has shown that to allow a quantum computation to proceed for ten times as long , the reliability of components does have to be improved a little . But the improvement required is a small fixed increment , not a factor of ten .
That quantum error - correction is possible at all has led to a new wave of optimism about quantum computing . The circuitry for error - correction is too complex to be built in a laboratory anytime soon . But the group from Innsbruck , which is led by Peter Zoller and Rainer Blatt , has just published another partial error - correcting code based on the principle of quantum cannon fodder . This one is still simpler than Dr Shor 's and should be particularly effective in correcting the sort of errors that might be expected in an ion - trap quantum computer .
Three times five , silly The Innsbruck group hopes to try out its scheme with a three - bit computer by 1998. Meanwhile , in Los Alamos , Richard Hughes and his team are planning to implement Dr Shor 's algorithm for factorisation . With luck they will be able to factorise the number 15 within five years .
That this is considered ambitious hints at the difficulty of this latest merger of physics and computing . Indeed , many physicists doubt whether quantum computing will ever amount to anything useful . Dr Landauer is one of them . His work on the ordinary ( not quantum ) physics of information led to a new appreciation of information 's inextricable relationship with physics . But he does not expect IBM to be mass - producing quantum computers anytime soon .
In this , he is assuredly correct . And even among the quantum computer 's enthusiasts , most researchers are motivated less by a desire to produce a practical quantum computer than by a wish to learn how to wield more control over quantum effects themselves . Until now , nobody has been able to direct quantum states with the flexibility or precision required for quantum computers . Doing so should help to illuminate some of quantum theory 's more curious features .
In addition , if quantum computing really could be made to work , physicists would have a powerful new analytical tool . One use for it would be to explore the many ramifications of quantum theory ( such as the " Hubbard model " of how electrons hop around in a crystal , or theories of fundamental particles ) that are thought to be correct but cannot be tested precisely . The sort of equations needed to check these ideas are too much for even the most nimble non - quantum supercomputer to handle .
In 1982, the late Richard Feynman , who was also at Caltech , suggested that a quantum computer would help to solve these sorts of problems . To simulate an ordinary object , only a few numbers are needed . For a quantum object every possible state must be tracked - - an enormous set of numbers . How better to keep up with the calculation than by using something that is itself in a multitude of states ?
Whether such uses will come to fruition remains to be seen . But , as Dr Landauer points out , in one sense quantum computing has already been useful . It has provided a salutary reminder that computing does not take place in an abstract mathematical world . It uses earth , air , fire and water - - or single atoms of those things . It is subject to all the limitations , and all the possibilities , written in the laws of physics . Some mathematicians and computer scientists spend their days searching for the quickest possible algorithm to solve particular problems . To their surprise they must now learn quantum theory , in order to cover all the possibilities allowed by nature . After a long absence , they are being forced to return from the virtual reality of information science and deal with the real world , in all its indeterminate queerness .