Vous êtes sur la page 1sur 4


for trapping as 6 decreases is reflective of the Our results strongly support the conclusions of
incorporation of periodic components into the Grossmann and Thomae.
sequence of numbers generated. This research was supported by the Office of
To summarize the motivation and principal con- Basic Energy Sciences of the U. S. Department
clusion of this Letter, we restate' that for values of Energy.
of b where numerically generated sequences ~P
Pear to be chaotic, it has not been settled wheth-
er those sequences "are truly chaotic, or wheth-
er, in fact, they are really periodic, but with Permanent address: Miles Laboratories, Elkhart,
exceedingly large periods and very long tran- Ind. 46652.
sients required to settle down. On the one hand, ( )Present address: Department of Chemistry,
Grossman and Thomae" have suggested that Stanford University, Stanford, Cal. 94305.
(only) the parameter value b =1 generates pure
'E. Ott, Bev. Mod. Phys. 53, 655 (1981).
'C. A. Walsh and J. J. Kozak, Phys. Rev. Lett. 47,
chaos [see the discussion following Eq. (31) of
1500 (1981).
Ref. 5 and the correlations plotted in their Fig. 3E. W. Montroll, Proc. Symp. Appl. Math. Am. Math.
9]. On the other hand, for certa, in other values Soc. 16, 193 (1964); E. W. Montroll and Q. W. Weiss,
of b, numberical results of Lorenz (reported in J. Math. Phys. 6, 167 (1965); E. W. Montroll, J. Math.
Ref. 1) "strongly suggest that the sequences are Phys. 10, 753 (1969).
truly chaotic." The purpose of this communica- 4K. Tomita, in Pattern Formation by Dynamic Sys-
tion was to use an independent and exact result tems and Pattern Recognition, edited by H. Haken
(Springer-Verlag, Heidelberg, 1979), pp. 90-97.
from the statistical-mechanical theory of d =1 5S. Thomae and S. Grossmann, J. Stat. Phys. 26,
random walks to test the randomness of the para- 485 (1981).
bolic map for parameter values where the exis- S. Qrossmann and S. Thomae, Z. Naturforsch. 32a,
tence of "true chaos" is still an open question. 1353 (1977).

E*perimen&al Tes& of Bell's Inequalities

Alain Aspect, Jean Dalibard,
Institut d'Optique
Using Time-Varying
and Gerard Roger
Theomque et APPliquee, F-9j406 Qxsay Cedex, France

(Received 27 September 1982)

Correlations of linear polarizations of pairs of photons have been measured with
time-varying analyzers. The analyzer in each leg of the apparatus is an acousto-opti-
cal switch followed by two linear polarizers. The switches operate at incommensurate
frequencies near 50 MHz. Each analyzer amounts to a polarizer which jumps between
two orientations in a time short compared with the photon transit time. The results
are in good agreement with quantum mechanical predictions but violate Bell's inequal-
ities by 5 standard deviations.
PACS numbers: 03.65.8z, 35.80.+s

Bell's inequalities apply to any correlated meas-

urement on two correlated systems. For in- PM1 ~ ~ PM2

stance, in the optical version of the Einstein- I(a) I I (b)

Podolsky-Rosen-Bohm Gedankenexperiment, ' a
source emits pairs of photons (Fig. 1). Measure- COINCIDENCE
ments of the correlations of linear polarizations MONITORING

are performed on two photons belonging to the FIG. l. Optical version of the Einstein-Podolsky-
same pair. For pairs emitted in suitable states, Bosen-Bohm GedankenexPeximent. The pair of photons
the correlations are strong. To account for these v, and v, is analyzed by linear polarizers I and II (in
correlations, Bell considered theories which in- orientations a and b) and photomultipliers. The coin-
voke common properties of both members of the cidence rate is monitored.

1804 1982 The American Physical Society


pair. Such properties are referred to as supple-

mentary parameters. This is very different from J~l-
I (~a)
Q~ -
"" I I (b)
the quantum mechanical formalism, which does L
not involve such properties. With the addition of I'(a')
a reasonable locality assumption, Bell showed
that such classical-looking theories are con-
strained by certain inequalities that are not al-
ways obeyed by quantum mechanical predictions.
Several experiments of increasing accuracy
have been performed and clearly favor quantum
mechanics. Experiments using pairs of visible FIG. 2. Timing experiment with optical switches.
Each switching device (C&, C&&) is followed by two po-
photons emitted in atomic radiative cascades
larizers in two different orientations. Each combina-
seem to achieve a good realization of the ideal tion is equivalent to a polarizer switched fast between
Gedankenexperiment. ' However, all these ex- two or ientations.
periments have been performed with static setups,
in which polarizers are held fixed for the whole
duration of a run. Then, one might question Shimony-Holt inequalities':
Bell's locality assumption, that states that the re- 1~S~O
sults of the measurement by polarizer II does not
depend on the orientation a of polarizer l (and with
vice versa), nor does the way in which pairs are
emitted depend on a or b. Although highly rea-
~(a, b) ~(a, 6 ) V(a, b)
Q(&x& ao) Q(oo m') Q(oo' ao)
sonable, such a locality condition is not prescribed
by any fundamental physical law. As pointed out
V(a', b ) N(a, , ) V(-, b)
by Bell, it is possible, in such experiments, to P7(~' ~~) ~(~~ ~) N(~ ~)'
reconcile supplementary-parameter theories and
the experimentally verified predictions of quan- The quantity & involves (i) the four coincidence
tum mechanics: The settings of the instruments counting rates [N(a, b), W(a', b), etc. ] measured
are made sufficiently in advance to allow them to in a single ~n; (ii) the four corresponding coin-
reach some mutual rapport by exchange of signals cidence rates [(~, ~), V(~', ~), etc. ] with all
with velocity less than or equal to that of light. polarizers removed; and (iii) two coincidence
If such interactions existed, Bell's locality condi- rates [N(a', ~), N(~, b)] with a polarizer removed
tion would no longer hold for static experiments, on each side. The measurements (ii) and (iii) are
nor would Bell's inequalities. performed in auxiliary runs.
Bell thus insisted upon the importance of ex- In this experiment, switching between the two
periments of the type proposed by Bohm and channels occurs about each 10 ns. Since this de-
Aharonov, ' in which the settings are changed dur- lay, as well as the lifetime of the intermediate
ing the flight of the particles. In such a timing level of the cascade (5 ns), is small compared
experiment, the locality condition would then be- to L/c (40 ns), a detection event on one side and
come a consequence of Einstein's causality, pre- the corresponding change of orientation on the
venting any f aster-than-light inf luenee. other side are separated by a spacelike interval.
In this Letter, we report the results of the first The switching of the light is effected by acousto-
experiment using variable polarizer s. F ollowing optical interaction with an ultrasonic standing
our proposal, ' we have used a modified scheme wave n water. ' As sketched in Fig. 3 the inci-
(Fig. 2). Each polarizer is replaced by a setup dence angle is equal to the Bragg angle, 6 g =5
involving a switching device followed by two po- &10 ' rad. It follows that light is either trans-
larizers in two different orientations: a and a' mitted straight ahead or deflected at an angle
on side I, and b and b' on side II. Such an optical 26 &. The light is completely transmitted when
switch is able to rapidly redirect the incident the amplitude of the standing wave is null, which
light from one polarizer to the other one. If the occurs twice during an acoustical period. A
two switches work at random and are uncorrelat- quarter of a period later, the amplitude of the
ed, it is possible to write generalized Bell's in- standing wave is maximum and, for a suitable
equalities in a form similar to Clauser-Horne- value of the acoustical power, light is then fully


nsducer 20ns
z li

e ~ ~


I IdeAt
fYl l
is ~ I+
~ A

~ i
~ ~
t ~
generator ~i ~ ~
~ ~ i ~

25 Mhz
~ ~

FIG. 3. Optical switch. The incident light is switched

at a frequency around 50 MHz by diffraction at the
Bragg angle on an ultrasonic standing wave. The inten- FIG. 4. Average normalized coincidence rate as a
sities of the transmitted and deflected beams as a func- function of the relative orientation of the polarizers.
tion of time have been measured with the actual source. Indicated errors are + 1 standard deviation. The dashed
The fraction of light directed towards other diffraction curve is not a fit to the data but the predictions by quan-
orders is negligible. tum mechanics for the actual experiment.

deflected. This optical switch thus works at rate (i.e., coincidences due to photons emitted
twice the acoustical frequency. by the same atom) are obtained by subtraction.
The ultrasonic standing-wave results from Simultaneously, a time-to-amplitude converter,
interf erenee between counterpropagating acoustic followed by a fourfold multichannel analyzer,
waves produced by two electroacoustical trans- yields four time-delay spectrums. Here, the
ducers driven in phase at about 25 MHz. In true coincidence rate is taken as the signal in
auxiliary tests with a laser beam, the switching the peak of the time-delay spectrum. '
has been found complete for an acoustical power
about 1 W. In the actual experiment, the light
beam has a finite divergence, and the switching
We have not been able to achieve collection ef-
ficiencies as large as in previous experiments,
since we had to reduce the divergence of the
is not complete (Fig. 3). beams in order to get good switching. Coinci-
The other parts of the experiment have already
been described in previous publications. The ' dence rates with the polarizers removed were
only a few per second, with accidental coincidence
high-efficiency well-stabilized source of pairs of rates about one per second.
correlated photons, at wavelengths &, =422. 7 nm A typical run lasts 12 000 s, involving totals of
and &, =551.3 nm, is obtained by two-photon exci- 4000 s with polarizers in place at a given set of
tation of a (& =0)- (&=1)-(&=0) cascade in calci- orientations, 4000 s with all polarizers removed,
ume and 4000 s with one polarizer removed on each
Since each switch is 6 m from the source, side. In order to compensate the effects of sys-
rather complicated optics are required to match tematic drifts, data accumulation was alternated
the beams with the switches and the polarizers. between these three configurations about every
We have carefully checked each channel for no 400 s. At the end of each 400-s period, the raw
depolarization, by looking for a cosine Malus law data were stored for subsequent processing with
when a supplementary polarizer is inserted in the help of a computer.
front of the source. These auxiliary tests are At the end of the run, we average the true coin-
particularly important for the channels which in- cidence rates corresponding to the same config-
volve two mirrors inclined at 11'. They also urations for the polarizers. We then compute the
yield the efficiencies of the polarizers, required relevant ratios for the quantity S. The statistical
for the quantum mechanical calculations. accuracy is evaluated according to standard sta-
The eoineidence counting electronics involve tistical methods f or photon counting. The proces-
f our double- coincidence- counting circuits with sing is perf ormed on both sets of data: that ob-
coincidence windows of 18 ns. I'or each relevant tained with coincidence circuits, and that obtained
pair of photomultipliers, we monitor nondelayed with the time-to-amplitude converter. The two
and delayed coincidences. The true coincidence methods have always been found to be consistent.


Two runs have been performed in order to test tronies. We are indebted to Jean-Pierre Pas-
Bell's inequalities. In each run, we have chosen serieux, from the Departement de Physique Nu-
a set of orientations leading to the greatest pre- cleaire et Basse Energie at Centre O'Etudes Na-
dicted conflict between quantum mechanics and tional de Saclay, for his assistance in fast-coin-
Bell's inequalities [(a, 5) = (6, a') = (a', b') =22. cidence techniques, and to Dr. Torguet and Dr.
(i, 6') =67.5']. The average of the two runs yields Gazalb, from Laboratoire d'Acoustooptique de
Valenciennes, for help with the optical switches.
Sgyp t 0 101 + 0 020 p
We also acknowledge the valuable help of Philippe
violating the inequality S ~ 0 by 5 standard devia- Grangier during the final runs.
tions. On the other hand, for our solid angles
and polarizer efficiencies, quantum mechanics
predicts SQM 0 112.
We have carried out another run with different
)Permanent address: Laboratoire de Physique de
orientations, for a direct comparison with quan- l'Ecole Normale Superieure, F-75231 Paris Cedex 05,
tum mechanics. Figure 4 shows that the agree- France.
ment is excellent. 'A. Einstein, B. Podolsky, and N. Bosen, Phys Hev.
The new feature of this experiment is that we 47, 777 (1935); D. Bohm, Quantum Theory (Prentice-
change the settings of the polarizers, at a rate Hall, Englewood Cliffs, N. J., 1951).
greater than c/&. The ideal scheme has not been 2J. S. Bell, Physics (N. Y.) 1, 195 (1965).
completed since the change is not truly random, 3J. F. Clauser and A. Shimony, Rep. Prog. Phys. 41,
1981 (1978). (This paper is an exhaustive review of
but rather quasiperiodic. Nevertheless, the two
the subject); F. M. Pipkin, inAdvances in Atomic and
switches on the two sides are driven by different MoLeculax Physics, edited by D. B. Bates and B. Beder-
generators at different frequencies. It is then son (Academic, New York, 1978) {a comprehensive
very natural to assume that they function in an un- review).
correlated way. A. Aspect, P. Grangier, and G. Roger, Phys. Bev.
A more ideal experiment with random and com- Lett. 47, 460 (1981).
5A. Aspect, P. Grangier, and G. Roger, Phys.
plete switching would be necessary for a fully Bev.
conclusive argument against the whole class of Lett. 49, 91 (1982).
'D. Bohm and Y. Aharonov, Phys. Bev. 108, 1070
supplementary-parameter theories obeying Ein- (1957). The suggestion of this thought-experiment was
stein's causality. However, our observed viola- already given by D. Bohm, Ref. l.
tion of Bell's inequalities indicates that the exper- 7A. Aspect, Phys. Lett. 54A, 117 (1975), and Phys.
imental accuracy was good enough for pointing Bev. D 14, 1944 (1976}.
out a hypothetical discrepancy with the predic- J. F. Clauser, M. A. Horne, A. Shimony, and B. A.
tions of quantum mechanics. No such effect was Holt, Phys. Rev. Lett. 23, 880 (1969).
observed. '0 A. Yariv, Quantum E/ect~onics (%'iley, New York,
This work has been carried out in the group of ' Let us emphasize that such
results cannot be taken
Professor Imbert, who we thank for his support. as providing the possiblity of faster-than-light com-
We thank all the technical staff of the Institut munication. See, for instance, A. Aspect, J. Phys.
d'Optique, especially Andre Villing for the elec- (Paris), Colloq. 42, C263 {1981).


Centres d'intérêt liés