Quantum computing: 15 minutes to understand everything

This arti­cle is part of our spe­cial issue « Quan­tum: the sec­ond rev­o­lu­tion unfolds ».

Read it here

Quan­tum com­put­ers work with quan­tum bits, or qubits, which, unlike stan­dard com­put­er bits that can have a val­ue of either 0 or 1, can be in a super­po­si­tion of both 0 and 1. This char­ac­ter­is­tic means that quan­tum com­put­ers could be much faster than con­ven­tion­al com­put­ers for cer­tain tasks. They could also be used to solve cer­tain prob­lems that a con­ven­tion­al com­put­er cannot.

“A quan­tum com­put­er will manip­u­late many qubits in a mas­sive­ly super­posed state: 0000 plus 1111, for exam­ple, » explains Landry Bretheau. « In this ‘entan­gled’ state, sev­er­al cal­cu­la­tions can be car­ried out in par­al­lel. A con­crete exam­ple: imag­ine that the cal­cu­la­tion, the prob­lem, is to get out of a maze. How would you go about that? A human being or a com­put­er pro­gramme will try out dif­fer­ent pos­si­ble tra­jec­to­ries one by one. Each time, a dead end is reached, they will then retrace their steps. In this way, they will test all the paths until they emerge from the maze. Since a quan­tum sys­tem can be in a super­po­si­tion of states, that is, it can be in sev­er­al places at the same time, it can there­fore try to explore all the dif­fer­ent paths in par­al­lel and escape from the maze more quickly. »

Qubits can be built from dif­fer­ent plat­forms or mate­r­i­al build­ing blocks, such as super­con­duct­ing qubits, ele­men­tary par­ti­cles or trapped ions. Oth­er meth­ods in the pipeline are pho­ton­ic quan­tum proces­sors that use light. « We use the term quan­tum com­put­er, but it would be more accu­rate to talk about quan­tum proces­sor, because the entire cal­cu­la­tion can­not be imple­ment­ed on a quan­tum com­put­er, only a frac­tion of it, » explains Loïc Hen­ri­et. « We will always need a clas­si­cal proces­sor to coor­di­nate all the steps of the calculation. »

At present, a quan­tum proces­sor is still in the exper­i­men­tal stage: for one, it takes up a lot of space – for exam­ple, the one Löic Hen­ri­et’s team is work­ing on sits in a large box mea­sur­ing 3 metres by 2 metres by 2 metres. A very high vac­u­um, of 10^-11 mbar, is also required to place the qubits in well-defined posi­tions. This cor­re­sponds to around the same pres­sure as can be found on the sur­face of the Moon.

The sophis­ti­cat­ed optics need­ed to con­trol the qubits con­sist of lasers, lens­es and mir­rors. To coor­di­nate the oper­a­tion of each of these dif­fer­ent pieces of equip­ment (which make up the hard­ware) and to syn­chro­nise them, on-board soft­ware is need­ed. This soft­ware is the quan­tum proces­sor’s oper­at­ing system.

There are many areas in which a quan­tum com­put­er could prove more use­ful than a con­ven­tion­al com­put­er, either in terms of com­put­ing time or the qual­i­ty of the results obtained. The best-known exam­ple is Shor’s algo­rithm, which enables large num­bers to be effi­cient­ly fac­torised into prime fac­tors, for appli­ca­tions in cryp­tog­ra­phy and com­put­er secu­ri­ty, for exam­ple. Quan­tum com­put­ers will also be very good at mak­ing use of spe­cial algo­rithms to solve com­plex opti­mi­sa­tion prob­lems, such as those linked to sched­ul­ing, rout­ing and logis­tics. These prob­lems involve find­ing the opti­mum solu­tion from a large num­ber of pos­si­bil­i­ties – the most famous being the « trav­el­ling sales­man » prob­lem, which involves find­ing the short­est pos­si­ble route between sev­er­al cities. Deliv­ery and logis­tics com­pa­nies will cer­tain­ly want to adopt quan­tum tech­nol­o­gy for this type of application.

Prob­lems linked to the reac­tiv­i­ty of mol­e­cules will also ben­e­fit. « There is a lot of research going on in this area, » explains Löic Hen­ri­et. « With a quan­tum proces­sor, we will be able to car­ry out much more effi­cient cal­cu­la­tions to deter­mine the reac­tiv­i­ty of cer­tain pro­teins, for exam­ple, which will have enor­mous appli­ca­tions for the phar­ma­ceu­ti­cal indus­try and the syn­the­sis of new drugs. We will also be able to cal­cu­late the prop­er­ties of new mate­ri­als of inter­est for many tech­no­log­i­cal fields. »

Machine learn­ing and arti­fi­cial intel­li­gence are also impor­tant appli­ca­tion areas since quan­tum com­put­ers should be able to improve machine learn­ing algo­rithms – poten­tial­ly dra­mat­i­cal­ly – by pro­vid­ing faster and more effi­cient opti­mi­sa­tion rou­tines or by explor­ing new mod­els and archi­tec­tures. This could be a mas­sive new mar­ket, but it will depend on build­ing prac­ti­cal quan­tum com­put­ers on a large scale and devel­op­ing algo­rithms and appli­ca­tions that can take advan­tage of their unique capabilities.

For a quan­tum com­put­er to work, it must be « uni­ver­sal », that is, it must be capa­ble of cor­rect­ing the errors caused by the imper­fect nature of cur­rent hard­ware, which pre­vents a cal­cu­la­tion from being com­put­ed com­plete­ly. The main cause of these errors is the deco­her­ence of the qubits them­selves, which destroys the quan­tum char­ac­ter of the qubits and returns them to the state of clas­si­cal bits. Deco­her­ence is caused by the inter­ac­tion of the qubits with their envi­ron­ment. The real chal­lenge is there­fore to iso­late the sys­tem effec­tive­ly. To do this, the qubits gen­er­al­ly have to oper­ate at a tem­per­a­ture close to 0 K and be shield­ed from each oth­er and the envi­ron­ment. In addi­tion to this, quan­tum error cor­rec­tion (QEC) tech­niques can be used to achieve « fault-tol­er­ant quan­tum com­put­ing ». These tech­niques involve using a large num­ber of qubits to cre­ate a « log­i­cal qubit » that is much less prone to errors. 

Accord­ing to experts, a real quan­tum « advan­tage » or « suprema­cy » will only be achieved when quan­tum com­put­ers oper­ate with a mil­lion qubits. And since the cur­rent record is still less than 100 qubits, there is still a long way to go.

While, in the­o­ry, there is noth­ing pre­vent­ing the cre­ation of large-scale quan­tum com­put­ers, a num­ber of major engi­neer­ing prob­lems need to be resolved first. « Com­pa­nies see quan­tum com­put­ing as a strate­gic invest­ment and don’t want to miss out, » explains Löic Hen­ri­et. « It’s no longer a ques­tion of if, but rather a ques­tion of when the quan­tum proces­sor will become an inte­gral part of IT solutions.

We are cur­rent­ly at the cusp of a tech­no­log­i­cal rev­o­lu­tion and, in France, we have the resources to be at the heart of this, both at aca­d­e­m­ic lev­el and at the lev­el of com­pa­nies and start-ups. Of course, end-cus­tomers and busi­ness­es also need to be on board.

Over the last five years or so, we’ve seen a real increase in interest

That said, a quan­tum com­put­er, uni­ver­sal or not, will not replace your per­son­al com­put­er or smart­phone any time soon, and the first cus­tomers will cer­tain­ly be gov­ern­ments and major cor­po­ra­tions rather than the gen­er­al pub­lic, adds Landry Bretheau. « Sci­en­tists them­selves will also be the first users, which is why the world of quan­tum com­put­ing is of inter­est to so many dis­ci­plines: chem­istry, mate­ri­als sci­ence, biol­o­gy and physics. Each of these dis­ci­plines will devel­op its own algo­rithms to solve very spe­cif­ic problems.

« Over the last five years or so, we’ve seen a real increase in inter­est, » he says. “Some are call­ing it a ‘quan­tum boom’, with the cre­ation of numer­ous start-ups and major fund-rais­ing campaigns.”

Psi­Quan­tum and IonQ, which have raised €600 mil­lion and €400 mil­lion respec­tive­ly, are two out­stand­ing exam­ples. « In France, we have the Quan­tum Plan, announced by Pres­i­dent Macron in ear­ly 2021, and the start-up with the most wind in its sails from a finan­cial point of view in France at the moment is Pasqal, which has just raised €100 million.

Although we prob­a­bly won’t suc­ceed in build­ing a ful­ly oper­a­tional and fault-tol­er­ant com­put­er in the next ten years, we can be sure that we will make unthought-of dis­cov­er­ies along the way that will be poten­tial­ly use­ful and per­haps even change the face of soci­ety, just as clas­si­cal com­put­ing has done over the last 50 years and, more recent­ly, the Internet.

« It’s a great time to be work­ing in this field, » says Landry Bretheau. « There’s a lot of excite­ment and the sec­tor is evolv­ing very quickly.”

Isabelle Dumé