"Long before it's in the papers"
January 27, 2015


Physicists claim first true random number generation

Sept. 13, 2010
Courtesy of the Max Planck Society,
the University of Maryland
and World Science staff

Seem­ingly ran­dom events seen in daily life, such as the results of dice throws, ac­tu­ally have def­i­nite causes that de­ter­mine them ex­actly. So they are not really ran­dom. This is true for ob­jects that are at least large enough to see with ordi­nary mi­cro­scopes; such ob­jects fol­low the laws of what is known as “clas­si­cal” phys­ics.

But in the realm of quan­tum phys­ics—the rules that gov­ern atoms or smaller ob­jects—it’s not unusual to find events that, by all meas­ures, are truly ran­dom. In other words, they can­not be pre­dicted no matter how much you know about what led to up to them. 

Max Planck re­search­ers re­ported that they used a strong la­ser (com­ing from the left), a beam split­ter, two de­tec­tors and some elec­tron­ic com­po­nents in their set­up. The de­tec­tors were used to meas­ure the ran­dom­ly var­y­ing in­ten­si­ty of the quan­tum noise. The sta­tis­ti­cal spread of the meas­ured val­ues fol­lows a bell-shaped or "Gauss­ian" curve (bot­tom). In­di­vid­u­al val­ues were as­signed to sec­tions of the bell-shaped curve that cor­re­spond to a num­ber. (Cour­te­sy MPI for the Phys­ics of Light )

This year, scientists say they have for the first time built devices that ex­ploit quan­tum phys­ics to gen­er­ate real ran­dom num­bers. Such numbers can’t be obtained even with ordinary com­pu­ters, which can si­mu­late ran­dom­ness but not really pro­duce it. 

True ran­dom num­bers can be use­ful to se­curely en­crypt da­ta and to sim­u­late eco­nom­ic pro­cesses and cli­mate changes, among oth­er things.

One set of new findings was re­ported in the Aug. 29 on­line is­sue of the re­search jour­nal Na­ture Pho­ton­ics. The re­searchers ex­ploited the fact that mea­sure­ments based on quan­tum phys­ics can only pro­duce a spe­cif­ic re­sult with a cer­tain prob­a­bil­ity, that is, ran­domly.

The phe­nom­e­non we com­monly re­fer to as chance is merely a re­sult of a lack of knowl­edge. If we knew the loca­t­ion, speed and oth­er char­ac­ter­is­tics of all of the par­t­i­cles in the uni­verse with ab­so­lute cer­tainty, ac­cord­ing to clas­si­cal phys­ics we could pre­dict eve­ry­thing, in­clud­ing di­ce throws and lot­tery re­sults. 

By the same to­ken, computer-gen­er­ated ran­dom num­bers “sim­u­late ran­domness, but with the help of suit­a­ble tests and a suf­fi­cient vol­ume of da­ta, a pat­tern can usu­ally be iden­ti­fied,” said re­searcher Christoph Mar­quardt of the Max Planck In­sti­tute for the Phys­ics of Light in Er­lan­gen, Ger­ma­ny.

True ran­domness only ex­ists in quan­tum phys­ics. A quan­tum par­t­i­cle will re­main in one place or anoth­er and move at one speed or anoth­er with a cer­tain de­gree of prob­a­bil­ity. “We ex­ploit[ed] this ran­domness of quan­tum-mechanical pro­cesses to gen­er­ate ran­dom num­bers,” said Mar­quardt. 

He and a group of col­leagues used vac­u­um fluctua­t­ions, a sort of back­ground stat­ic that per­me­ates emp­ty space. 

Such fluctua­t­ions are anoth­er char­ac­ter­is­tic of the quan­tum world: there is no true emp­ti­ness. Even in an “emp­ty” space de­void of vis­i­ble light, pack­ets of en­er­gy equiv­a­lent to half of a pho­ton, or light par­t­i­cle, can be formed. These leave tracks de­tect­a­ble in soph­is­t­icated mea­sure­ments. This ran­dom “noise,” called vac­u­um fluctua­t­ions, arises only when the phys­i­cists look for it, that is, when they car­ry out a meas­ure­ment.

To meas­ure the noise, Mar­quardt and col­leagues split a strong la­ser beam in­to equal parts us­ing a de­vice called a beam split­ter. This de­vice had two in­put ports to col­lect in­com­ing light, and two out­put ports to re­lease out­go­ing light. The re­search­ers cov­ered the sec­ond in­put port to block light from en­ter­ing. The vac­u­um fluctua­t­ions were still there, how­ev­er, and they in­flu­enced the two out­put beams. 

When the sci­en­tists meas­ured the two out­put beams and sub­tracted the re­sults from each oth­er, they were not left with noth­ing. What re­mained, they said, was the quan­tum noise, whose pre­cise val­ues de­pended on chance.

“True ran­dom num­bers are dif­fi­cult to gen­er­ate but they are needed for a lot of ap­plica­t­ions,” said Gerd Leuchs, di­rec­tor of the Max Planck In­sti­tute. Se­cur­ity tech­nol­o­gy, in par­tic­u­lar, needs ran­dom com­bina­t­ions of num­bers to en­code bank da­ta for trans­fer. Ran­dom num­bers can al­so be used to sim­u­late com­plex pro­cesses whose out­come de­pends on prob­a­bil­i­ties.

There are oth­er quan­tum pro­cesses be­sides vac­u­um fluctua­t­ions that can pro­duce true ran­domness, the phys­i­cists said. But their set­up made it eas­i­er to sep­a­rate these fluctua­t­ions from “clas­si­cal” noise, or eve­ry­day types of seem­ingly ran­dom pro­cesses. These would pol­lute the mea­sure­ments by in­tro­duc­ing some­thing that’s not really ran­dom. “Clas­si­cal” noise can re­sult from, say, the slight wob­bling of a meas­ure­ment in­stru­ment.

Al­so, “the vac­u­um fluctua­t­ions pro­vide un­ique ran­dom num­bers” that can’t be cop­ied by a “da­ta spy,” said Mar­quardt. “We do not need ei­ther a par­tic­u­larly good la­ser or par­tic­u­larly ex­pen­sive de­tec­tors for the set-up,” added Chris­tian Ga­bri­el of the in­sti­tute.

Mar­quardt’s group isn’t the first to have claimed true ran­dom num­ber gen­er­a­tion. Ear­li­er this year, re­search­ers with the Joint Quan­tum In­sti­tute, a part­ner­ship of the Uni­ver­si­ty of Mar­y­land and the U.S. Na­tion­al In­sti­tute of Stan­dards and Tech­nol­o­gy, claimed to have used the phe­nom­e­non of quan­tum en­tan­gle­ment to com­mu­ni­cate ran­dom num­bers. In quan­tum en­tan­gle­ment, two par­t­i­cles set apart some dis­tance from each oth­er are found to have ex­act­ly the same prop­er­ties, some of which are ran­dom, at a giv­en point in time. This phe­nom­enon can in prin­ci­ple be used to se­curely send the ran­dom in­for­mation be­tween two, ar­bit­rarily dis­tant points.

“The ran­dom bit gen­er­a­tion rate is ex­treme­ly slow” by this meth­od, said the in­sti­tute's Chris Mon­roe last April, “but we ex­pect speedups by or­ders of mag­ni­tude in com­ing years as we more ef­fi­cient­ly en­tan­gle the atoms.” The find­ings ap­peared in the April 15 is­sue of the re­search jour­nal Na­ture.

* * *

Send us a comment on this story, or send it to a friend


Sign up for

On Home Page         


  • St­ar found to have lit­tle plan­ets over twice as old as our own

  • “Kind­ness curricu­lum” may bo­ost suc­cess in pre­schoolers


  • Smart­er mice with a “hum­anized” gene?

  • Was black­mail essen­tial for marr­iage to evolve?

  • Plu­to has even cold­er “twin” of sim­ilar size, studies find

  • Could simple an­ger have taught people to coop­erate?


  • F­rog said to de­scribe its home through song

  • Even r­ats will lend a help­ing paw: study

  • D­rug may undo aging-assoc­iated brain changes in ani­mals

Every seemingly random event actually has a definite cause. That’s true in the world of “classical” physics, the rules that govern objects as they present themselves at the visible, everyday level. But in the realm of quantum physics—the rules that govern the tiniest sub-microscopic objects—processes that by all measures appear truly random are routine. Researchers have now built a device that exploits these processes to generate what they say are truly random numbers. By contrast, most computer-generated random numbers aren’t really random, but follow a definite system meant to simulate randomness. With the help of quantum physics, the new machine generates random numbers that can never be predicted, according to the scientists, who report their work in the Aug. 29 online issue of the research journal Nature Photonics. The researchers exploit the fact that measurements based on quantum physics can only produce a specific result with a certain probability, that is, randomly. True random numbers are needed to securely encrypt data and to simulate economic processes and climate changes, among other things. The phenomenon we commonly refer to as chance is merely a result of a lack of knowledge. If we knew the location, speed and other characteristics of all of the particles in the universe with absolute certainty, according to classical physics we could predict everything, including dice throws and lottery results. By the same token, computer-generated random numbers “simulate randomness, but with the help of suitable tests and a sufficient volume of data, a pattern can usually be identified,” said researcher Christoph Marquardt of the Max Planck Institute for the Physics of Light in Erlangen, Germany. True randomness only exists in quantum physics. A quantum particle will remain in one place or another and move at one speed or another with a certain degree of probability. “We exploit[ed] this randomness of quantum-mechanical processes to generate random numbers,” said Marquardt. He and a group of colleagues used vacuum fluctuations, a sort of background static that permeates empty space, as “quantum dice.” Such fluctuations are another characteristic of the quantum world: there is no true emptiness. Even in an “empty” space devoid of visible light, packets of energy equivalent to half of a photon, or light particle, can be formed. These leave tracks detectable in sophisticated measurements. This random “noise,” called vacuum fluctuations, arises only when the physicists look for it, that is, when they carry out a measurement. To measure the noise, Marquardt and colleagues split a strong laser beam into equal parts using a device called a beam splitter. This device had two input ports to collect incoming light, and two output ports to release outgoing light. The researchers covered the second input port to block light from entering. The vacuum fluctuations were still there, however, and they influenced the two output beams. When the scientists measured the two output beams and subtracted the results from each other, they were not left with nothing. What remained, they said, was the quantum noise, whose precise values depended on chance. “True random numbers are difficult to generate but they are needed for a lot of applications,” said Gerd Leuchs, director of the Max Planck Institute. Security technology, in particular, needs random combinations of numbers to encode bank data for transfer. Random numbers can also be used to simulate complex processes whose outcome depends on probabilities. There are other quantum processes besides vacuum fluctuations that can produce true randomness, the physicists said. But their setup made it easier to separate these fluctuations from “classical” noise, or everyday types of seemingly random processes. These would pollute the measurements by introducing something that’s not really random. “Classical” noise can result from, say, the slight wobbling of a measurement instrument. “When we want to measure the quantum noise of a laser beam, we also observe classical noise that originates, for example, from a shaking mirror,” said the institute’s Christoffer Wittmann, who also worked on the experiment. In principle, the vibration of the mirror can be calculated as a classical physical process and therefore destroys the random game of chance. Also, “the vacuum fluctuations provide unique random numbers” that can’t be copied by a “data spy,” said Marquardt. “We do not need either a particularly good laser or particularly expensive detectors for the set-up,” added Christian Gabriel of the institute.