References and Notes 1. A ‘‘fluffy-bun ny’’ is a cheap, manufactured toy given as a prize in British fairgrounds. 2. E. Schro ¨dinger, Naturwissenscha ften23, 807 (1935). 3. J. J. Bollinge r, W. M. Itano, D. J. Wineland, D. J. Heinzen, Phys. Rev. A54, R4649 (1996). 4. Z. Y. Ou, Phys. Rev. A55, 2598 (1997). 5. V. Giovann etti, S. Lloyd, L. Macco ne, Science306, 1330 (2004). 6. O. Carnal, J. Mlynek, Phys. Rev. Lett.66, 2689 (1991). 7. D. W. Keith, C. R. Ekstro m, Q. A. Turchette , D. E. Pritchard, Phys. Rev. Lett.66, 2693 (1991). 8. M. Kasevi ch, S. Chu, Phys. Rev. Lett.67, 181 (1991). 9. M. Arndt et al., Nature401, 680 (1999). 10. L. Hackermu ¨ller, K. Hornberger, B. Brezger, A. Zeilinger, M. Arndt, Nature427, 711 (2004). 11. S. M. Tan, D. F. Walls , Phys. Rev. A47, 4663 (1993). 12. R. Bac h, K. Rz a $z ˙ ewski, Phys. Rev. Lett.92, 200401 (2004). 13. M. Brun e et al., Phys. Rev. Lett.77, 4887 (1996). 14. W. Marshall , C. Simon, R. Penro se, D. Bouwme ester , Phys. Rev. Lett. 91, 130401 (2003). 15. Another way to find evidence for a cat, not discussed here, is to disen tangl e the particle s, but this also amounts to recombination and destroys the cat. 16. E. Joos, H. D. Ze h, Z. Phys. B59, 223 (1985). 17. W. H. Zurek , Phys. Today44, 36 (1991). 18. G. C. Ghirardi, A. Rimini , T. Weber , Phys. Rev. D34, 470 (1986). 19. W. H. Zurek,Rev. Mod. Phys.75, 715 (2003). 20. J. Javanainen, S. M. Yoo, Phys. Rev. Lett.76, 161 (1996). 21. J. A. Dunningha m, K. Burnett, Phys. Rev. Lett.82, 3729 (1999). 22. A. V. Rau, J. A. Dunningham, K. Burnet t, Science301, 1081 (2003). 23. J. A. Dunning ham, A. V. Rau, K. Burnett, J. Mod. Opt. 51, 2323 (2004). 24. This work was supported by the UK Engineering and Physical Sciences Research Council and the Royal Society and Wolfson Foundation. 10.1126/science.1109545 R E V I E W Time and the Quantum: Erasing the Past and Impacting the Future Yakir Aharonov 1,2 and M. Suhail Zubairy 3 * The quantum er as er ef fe ct of Sc ul ly an d Dru ¨ hl dramatica lly unde rsco res the difference between our classical conceptions of time and how quantum processes can unfold in time. Such eyebrow-raising features of time in quantum mechanics have been labeled ‘‘the fallacy of delayed choice and quantum eraser’’ on the one hand and described ‘‘as one of the most intriguing effects in quantum mechanics’’ on the othe r. In the prese nt pap er, we dis cuss how the ava ila bil ity or erasure ofinformation generated in the past can affect how we interpret data in the present. The quantum eraser concept has been studied and extended in many differ ent experiments and scenarios, for example, the entanglement quantum eraser, the kaon quantum eraser, and the use of quantum eraser entanglement to improve micro- scopic resolution. The Bclassical[ notion of time was summedup by Newton:BIabsolute and mathematical time, of itself, and from its own nature, flows equally wit hout rel atio n to anythin g ext er- nal.[In the present article, we go beyond ourclassica l experi ence by present ing counter - intuiti ve feat ur es of ti me as it evo lves in certain experimen ts in quantum mechanics. To illustrate this point, an excellent example is the dela yed-cho ice qua ntum era ser, pro- posed by Marlan O. Scully and Kai Dr[hl (1), which was described as an idea thatBshookthe physics communi ty[ whe n it was firs t published in 1982 ( 2). They analy zed a photon correlation experiment designed to probe the extent to which information ac- cessible to an observer and its erasure affects measured results. The Scully-Dr[hl quantum eraser idea as it was described inNewsweektel ls the st or y well ( 3), and Fig . 1 is an adaptati on of their account of this fascinating effect. In his bookThe Fabric of the Cosmos( 4), Brian Greene sums up beautifully the counter- intuitiv e outcome of the experiment al real- izations of the Scully-Dr[hl quantum eraser(p. 149): These experiments are a magnif icent af- front to our convention al notions ofspace and time. Something that takes place long after and far away from some- thin g els e nev ert hel ess is vit al to ourdes cri ptio n of that something els e. By any class ical-c ommon sense -reck oning, that_ s, well, crazy. Of course, that_ s the point: classical reckoning is the wrongkin d of reck oni ng to us e in a qua ntu m universe I. For a fe w da ys af t er I learned of these experiments, I remem- ber feeli ng elate d. I felt I_d been given a gl impse into a veiled side of real- ity. Common expe- rience—mundane, ordinary, day-to- day activities—sud den ly see medpart of a classica lcharade, hiding the tru e natu re of ourquantum world. Th e worl d of the everyday suddenly seemed nothing butan inverte d magi c act, lulling its audi- ence int o bel iev ingin the usual, famil- iar con ce pti ons ofspa ce and ti me, whi le the astonish- in g tr ut h of qu ant um reality lay carefully gua rde d by natur e _ s sle igh ts of hand . 1 School of Physics and Astronomy, Tel Aviv Univer- sity, Tel Aviv 69978, Israel. 2 Department of Physics, Unive rsity of South Carolin a, Columbia, SC 29208, USA. 3 Institute for Quantum Studies and Department of Physics, Texas A&M University, College Station, TX 77843, USA. *To who m corr espo ndence sho uld be addressed. E-mail: [email protected]As Thomas Young taught us two hundred years ago, photons interfere. But now we knowthat: Knowledge of path (1 or 2) is t he reason why interference is lost. It'sas if the photon knows it is being watched. But now we discover that: Erasing the knowledge of photon pathbringsinterference back. Erasing Knowledge! “No wonder Einstein was conf used.” Fig. 1. Schematics for the Young’s double-slit experiment. The which- path information wipes out the interference pattern. The interference pattern can be restored by erasing the which-path information. EI N S T E I N ’ S LE G A C Y EI N S T E I N ’ S LE G A C Y www.sc iencema g.o rg SCIENCE VOL 307 11 FEBRUARY 2005 875 SPECIALSECTIO N o n A u g u s t 8 , 2 0 1 0 w w w . s c i e n c e m a g . o r g D o w n l o a d e d f r o m
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
References and Notes1. A ‘‘fluffy-bunny’’ is a cheap, manufactured toy given
as a prize in British fairgrounds.2. E. Schrodinger, Naturwissenschaften 23, 807 (1935).3. J. J. Bollinger, W. M. Itano, D. J. Wineland, D. J. Heinzen,
Phys. Rev. A 54, R4649 (1996).4. Z. Y. Ou, Phys. Rev. A 55, 2598 (1997).5. V. Giovannetti, S. Lloyd, L. Maccone, Science 306, 1330
(2004).6. O. Carnal, J. Mlynek, Phys. Rev. Lett. 66, 2689 (1991).7. D. W. Keith, C. R. Ekstrom, Q. A. Turchette, D. E. Pritchard,
Phys. Rev. Lett. 66, 2693 (1991).8. M. Kasevich, S. Chu, Phys. Rev. Lett. 67, 181 (1991).9. M. Arndt et al., Nature 401, 680 (1999).
10. L. Hackermuller, K. Hornberger, B. Brezger, A. Zeilinger,M. Arndt, Nature 427, 711 (2004).
11. S. M. Tan, D. F. Walls, Phys. Rev. A 47, 4663 (1993).12. R. Bach, K. Rza$zewski, Phys. Rev. Lett. 92, 200401 (2004).13. M. Brune et al., Phys. Rev. Lett. 77, 4887 (1996).14. W. Marshall, C. Simon, R. Penrose, D. Bouwmeester,
Phys. Rev. Lett. 91, 130401 (2003).15. Another way to find evidence for a cat, not discussed
here, is to disentangle the particles, but this alsoamounts to recombination and destroys the cat.
16. E. Joos, H. D. Zeh, Z. Phys. B 59 , 223 (1985).17. W. H. Zurek, Phys. Today 44, 36 (1991).18. G. C. Ghirardi, A. Rimini, T. Weber, Phys. Rev. D 34 ,
470 (1986).
19. W. H. Zurek, Rev. Mod. Phys. 75 , 715 (2003).20. J. Javanainen, S. M. Yoo, Phys. Rev. Lett. 76, 16
(1996).21. J. A. Dunningham, K. Burnett, Phys. Rev. Lett. 82, 37
(1999).22. A. V. Rau, J. A. Dunningham, K. Burnett, Science 30
1081 (2003).23. J. A. Dunningham, A. V. Rau, K. Burnett, J. Mod. Op
51, 2323 (2004).24. This work was supported by the UK Engineering a
Physical Sciences Research Council and the RoySociety and Wolfson Foundation.
10.1126/science.1109545
R E V I E W
Time and the Quantum: Erasing the Past andImpacting the Future
Yakir Aharonov1,2 and M. Suhail Zubairy 3*
The quantum eraser effect of Scully and Druhl dramatically underscores thedifference between our classical conceptions of time and how quantum processes
can unfold in time. Such eyebrow-raising features of time in quantum mechanicshave been labeled ‘‘the fallacy of delayed choice and quantum eraser’’ on the onehand and described ‘‘as one of the most intriguing effects in quantum mechanics’’ onthe other. In the present paper, we discuss how the availability or erasure of information generated in the past can affect how we interpret data in the present.The quantum eraser concept has been studied and extended in many differentexperiments and scenarios, for example, the entanglement quantum eraser, the kaonquantum eraser, and the use of quantum eraser entanglement to improve micro-scopic resolution.
The Bclassical[ notion of time was summed
up by Newton: BIabsolute and mathematical
time, of itself, and from its own nature, flows
equally without relation to anything exter-
nal.[ In the present article, we go beyond our classical experience by presenting counter-
intuitive features of time as it evolves in
certain experiments in quantum mechanics.
To illustrate this point, an excellent example
is the delayed-choice quantum eraser, pro-
posed by Marlan O. Scully and Kai Dr [hl (1),
which was described as an idea that Bshook
the physics community[ when it was first
published in 1982 (2). They analyzed a
photon correlation experiment designed to
probe the extent to which information ac-
cessible to an observer and its erasure affects
measured results. The Scully-Dr [hl quantum
eraser idea as it was described in Newsweek tells the story well (3), and Fig. 1 is an
adaptation of their account of this fascinating
effect.
In his book The Fabric of the Cosmos (4),
Brian Greene sums up beautifully the counter-
intuitive outcome of the experimental real-
izations of the Scully-Dr [hl quantum eras
(p. 149):
These experiments are a magnificent af-
front to our conventional notions of
space and time. Something that takes
place long after and far away from some-
thing else nevertheless is vital to our
description of that something else. By
any classical-common sense-reckoning,
that _ s, well, crazy. Of course, that _ s the
point: classical reckoning is the wrong
kind of reckoning to use in a quantum
universe I. For a few days after I
learned of these experiments, I remem-
ber feeling elated.
I felt I _d been givena glimpse into a
veiled side of real-
ity. Common expe-
rience—mundane,
ordinary, day-to-
day ac tivit ie s—
sud den ly see med
part of a classical
charade, hiding the
true nature of our
quantum wor ld.
The world of the
everyday suddenly
seemed nothing but an inverted magic
act, lulling its audi-
ence into believing
in the usual, famil-
iar conceptions of
spa ce and time,
while the astonish-
ing truth of quantum
reality lay carefully
guarded by nature_ s
sleights of hand.
1School of Physics and Astronomy, Tel Aviv Univer-sity, Tel Aviv 69978, Israel. 2Department of Physics,University of South Carolina, Columbia, SC 29208,USA. 3 Institute for Quantum Studies and Departmentof Physics, Texas A&M University, College Station, TX77843, USA.
*To whom correspondence should be addressed.E-mail: [email protected]
As Thomas Young taught us two
hundred years ago, photons interfere.
But now we know that:
Knowledge of path (1 or 2) is the reason
why interference is lost. It's as if the photon
knows it is being watched.
But now we discover that:
E rasing the know ledge of pho ton path
brings interference back.
Erasing Knowledge!
“No wonder Einstein was confused.”
Fig. 1. Schematics for the Young’s double-slit experiment. The which-path information wipes out the interference pattern. The interferencepattern can be restored by erasing the which-path information.
E I N S T E I N ’ S L E G A C YE I N S T E I N ’ S L E G A C Y
www.sciencemag.org SCIENCE VOL 307 11 FEBRUARY 2005
a m is l ea di n g w o rd in g t heauthors even appear to endorse
this interpretation.
In a later paper, however, the author
retracts this statement (8).
In fact, many people had a similar
mind set, and it is only by carefully
considering and analyzing several
experiments (real and gedanken) that
the issue is made clear.
We now turn to the particularly
clear treatment of Shih and co-
workers as depicted in Fig. 3. We
again consider two atoms of the type
shown in Fig. 2C located at sites 1and 2. A pair of photons g and f are
emitted either by the atom located at
1 or by the atom located at 2. The g
photon, as before, proceeds to the
screen on the right and is detected
by a detector on screen D at a loca-
tion x0
. A repeat of this experiment
yields an essentially random distri-
bution of photons on the screen.
What about the appearance and
disappearance of interference fringes
discussed above? For this purpose,
we look at the f photon that pro-
ceeds to the left. We consider onlythose instances where the f photon
scattered from the atom located at 1
proceeds to the beam splitter B1
and
the f photon scattered from the atom
located at 2 proceeds to B2
. At either
of these 50/50 beam splitters, the f
photon has a 50% probability of
procee ding to det ect ors D3
(for
photon scattered from 1) and to D4
(for photon scattered from 2). On the
other hand, there is also a 50%
probability that the photon will be
reflected from the respective beam
splitter and proceed to another 50/50 beam splitter, B. For these photons,
there is an equal probability of being
detected at detectors D1
and D2
.
If the f photon is detected at the detector
D3, it has necessarily come from the atom
located at 1 and could not have come from the
atom located at 2. Similarly, detection at D4
means that the f photon came from the atom
located at 2. For such events, we can also
conclude that the corresponding g photon
was also scattered from the same atom. That
is, we have ‘‘which-way’’ information if
detectors D3
or D4
register a count.
Returning to the quantum erasure proto-
col, if the f photon is detected at D1, there is
an equal probability that it may have come
from the atom located at 1, following the path
1 B1 BD
1, or it may have come from the atom
located at 2, following the path 2 B2 BD
1.
Thus, we have erased the information about
which atom scattered the f photo
and there is no which-path informa
tion available for the correspondin
g photon. The same can be sa
about the f photon detected at D
The difference between counts in D
and D2
is a phase shift such that
click at D1
gives the fringes co
responding to g1 þ g
2, whereas
click at D2
correlates with g1
– g2
After this experiment is done
large number of times, we shall havroughly 25% of f photons detecte
each at D1, D
2, D
3, and D
4 becaus
of the 50/50 nature of our bea
splitters. The corresponding spati
distribution of g photons will be, a
mentioned above, completely ra
dom. Next we do a sorting proces
We separate out all the events wher
the f photons are detected at D1
, D
D3
, and D4
. For these four groups o
events, we locate the positions of th
detected g photons on the screen D.
The key result is that, for th
events corresponding to the detectioof f photons at detectors D
3 and D
the pattern obtained by the g photon
on the screen D is the same as w
would expect if these photons ha
scattered from atoms at sites 1 and
respectively. That is, there are n
interference fringes, as would b
expected when we have which-pa
information available. On the co
trary, we obtain conjugate (p pha
shifted) interference fringes for tho
events where the f photons are d
tected at D1
and D2
. For this set o
data, there is no which-path informtion available for the corresponding
photons.
Suppose we place the f photo
detectors far away. Then the futu
measurements on these photons influ
ence the way we think about the
photons measured today (or ye
terday!). For example, we can con
clude that g photons whose f partne
were successfully used to ascerta
which-path information can be d
scribed as having (in the past) orig
nated from site 1 orsite2. Wecan als
conclude that g photons whose partners had their which-path info
mation erased cannot be described a
having (in the past) originated from site 1 o
site 2 but must be described, in the same sens
as having come from both sites. The futu
helps shape the story we tell of the past.
Here again the eloquent and insightful Bria
Greene says it well (p. 197):
Notice, too, perhaps the most dazzling
result of all: the three additional beam
Fig. 4. (A) The four kaons K S
, K L, K 0, and K 0 have characteristic
signatures; (the short-lived) kaon K S
decays into two p particles,whereas (the long-lived) kaon K
L decays into three p particles; the
K 0 kaon (strangeness þ1) mostly passes through matter, but theK 0 (strangeness –1) interacts much more strongly with matter(nuclei) and is stopped. (B) The K 0 and K 0 states are super-posit ions of K
S an d K
L, i . e. , kK 0À 0 ( kK SÀ þ kK
LÀ)=
ffiffiffi
2p
andkK 0À 0 (kK SÀ j kK LÀ)=
ffiffiffi
2p
. Now K S
and K L have masses m
S and m
L
so that kK 0(t)À 0 (ejim S
t kK SÀ þ ejimLt kK
LÀ)=
ffiffiffi
2p
. Thus, if we pro-duce K 0 particles in plate I and they propagate for a time t to
plate II then the probability for passage through plate II iskbK 0kK 0(t)Àk2 which shows oscillations in time. (C) A kaon quantumeraser may be realized by noting that p p collisions generate theentangled states moving to the right (r ) and left (l) which can bewritten in terms of which-way (K
S, K
L) or which-wave (K 0, K 0).
Quantum erasing is achieved by the left-moving kaon as themeasured kaon (which will or will not show oscillations), and theright tag or ancilla kaon will serve to select the which-waveensemble (K 0, K 0) if we put in plate II and measure K
r
0. However, if we do not put in the second plate then we must describe thephysics by the which-way subensemble. Thus, the entangled kaonstate can be used to demonstrate quantum erasure bysubensemble selection just as in the original photon case.However, if K
S or K
L propagates from I to II, the state of the kaon
just before it enters II is kK S
(t)À 0 ejim
S
t kK S
À and kbK 0kK S
(t)Àk2 0 1/2with a similar result for K
L. In this sense, K
S and K
L are ‘‘which-
way’’ (short or long lived) states like photons going through slit 1
or 2, i.e., do not show oscillations. K 0
and K 0
, however, do showoscillation behavior and in this sense may be called ‘‘which-wave.’’
E I N S T E I N ’ S L E G A C YE I N S T E I N ’ S L E G A C Y
www.sciencemag.org SCIENCE VOL 307 11 FEBRUARY 2005
References and Notes1. M. O. Scully, K. Druhl, Phys. Rev. A. 25, 2208 (1982).2. S. P. Walborn, M. O. T. Cunha, S. Padua, C. H.
Monken, Am. Sci. 91, 336 (2003).3. S. Begley, Newsweek , 19 June 1995, p. 67.4. B. Greene, The Fabric of the Cosmos (Alfred A. Knopf,
New York, 2004).5. R. P. Feynman, R. Leighton, M. Sands, The Feynman
Lectures on Physics, Vol. III (Addison Wesley,Reading, MA, 1965).6. Y.-H. Kim, R. Yu, S. P. Kulik, Y. Shih, M. O. Scully,
Phys. Rev. Lett. 84, 1 (2000).7. U. Mohrhoff, Am. J. Phys. 64 , 1468 (1996).8. U. Mohrhoff, Am. J. Phys. 67 , 330 (1999).9. Mathematically we can understand the essential
results of the Scully-Druhl quantum eraser by firstrealizing that the photon state emitted by the atomslocated at sites 1 and 2 is given by
ky 0À 0 1
ffiffiffi
2p (kg1Àkf1 À þ kg2 Àkf2 À)
i.e., either the photon pair g1, f
1 is emitted by the
atom located at site 1 or the pair g2, f
2 is emitted by
the atom at site 2. Thus if the f photon is detectedby D
3, the quantum state reduces to kg
1À. A similar
result is obtained for the f photon detection by thedetector D
4. This is the situation when the which-
path information is available and the sorted data yields no interference fringes. The physics behind theretrieval of the fringes is made clear by rewriting thestate ky
0À as
ky 0À 0 1
ffiffiffi
2p (kgþÀkfþÀ þ kgjÀkfjÀ)
where gT
and fT
are the symmetric and antisymmetric combinations.
kgT À 0 1
ffiffiffi
2p (kg1À þ kg2À)
kfTÀ 0 1
ffiffiffi
2p ðkf1 À þ kf2À)
The state of the f photon after passage through tbeam splitter B is either kf
þÀ or kf
–À. Thus, a click
detectors D1
or D2, reduces the state of the g photo
to kgþ
À or kg–
À, respectively, leading to a retrieval the interference fringes.
10. M. O. Scully, B.-G. Englert, H. Walther, Nature 35
111 (1991).11. B.-G. Englert, J. Schwinger, M. O. Scully, Found. Phy
18, 1045 (1988).12. M. O. Scully, M. S. Zubairy, Quantum Optics (Cam
bridge, London, 1997).13. S. Durr, T. Nonn, G. Rempe, Nature 395, 33 (199814. P. Storey, S. Tan, M. Collett, D. Walls, Nature 36
626 (1994).15. R. Garisto, L. Hardy, Phys. Rev. A. 60, 827 (1999).16. G. Teklemariam, E. M. Fortunato, M. A. Pravia, T.
Havel, D. G. Cory, Phys. Rev. Lett. 86, 5845 (200117. M. S. Zubairy, G. S. Agarwal, M. O. Scully, Phys. Re
A. 70, 012316 (2004).18. A. Bramon, G. Garbarino, B. C. Hiesmayr, Phys. Re
Lett. 92, 020405 (2004).19. M. O. Scully, unpublished results.20. We thank E. Fry, A. Muthukrishnan, R. Ooi, and
Patnaik for their help in the preparation of th
manuscript. We also gratefully acknowledge suppofrom U.S. Air Force Office of Scientific ResearcDefense Advanced Research Projects Agency, aTexas A&M University’s Telecommunication aInformatics Task Force initiative.
10.1126/science.1107787
R E V I E W
Astrophysical Observations:Lensing and Eclipsing Einstein’s Theories
Charles L. Bennett
Albert Einstein postulated the equivalence of energy and mass, developed the theory of special relativity, explained the photoelectric effect, and described Brownian motion infive papers, all published in 1905, 100 years ago. With these papers, Einstein providedthe framework for understanding modern astrophysical phenomena. Conversely,astrophysical observations provide one of the most effective means for testingEinstein’s theories. Here, I review astrophysical advances precipitated by Einstein’sinsights, including gravitational redshifts, gravitational lensing, gravitational waves, theLense-Thirring effect, and modern cosmology. A complete understanding of cosmology,from the earliest moments to the ultimate fate of the universe, will requiredevelopments in physics beyond Einstein, to a unified theory of gravity and quantumphysics.
Einstein_s 1905 theories form the basis for
much of modern physics and astrophysics. In
1905, Einstein postulated the equivalence of
mass and energy (1), which led Sir Arthur
Eddington to propose (2) that stars shine by
converting their mass to energy via E 0 mc2,
and later led to a detailed understanding of
how stars convert mass to energy by nuclear
burning (3, 4). Einstein explained the photo-
electric effect by showing that light quanta
are packets of energy (5), and he received
the 1921 Nobel Prize in physics for this
work. With the photoelectric effect, astron-
omers determined that ultraviolet photons
emitted by stars impinge on interstellar dust
and overcome the work function of the grains
to cause electrons to be ejected. The photo-
electrons emitted by the dust grains excite the
interstellar gas, including molecules wi
molecular sizes of È1 nm, as estimated b
Einstein in 1905 (6 ). Atoms and molecul
emit spectral lines according to Einstein
quantum theory of radiation (7 ). The co
cepts of spontaneous and stimulated emissio
explain astrophysical masers and the 21-cm
hydrogen line, which is observed in emissio
and absorption. The interstellar gas, which
heated by starlight, undergoes Brownian mo
tion, as also derived by Einstein in 1905 ( 8
Two of Einstein_s five 1905 papers introduced relativity (1, 9). By 1916, Einstein ha
generalized relativity from systems movin
with a constant velocity (special relativity) t
accelerating systems (general relativity).
Space beyond Earth provides a uniqu
physics laboratory of extreme pressures an
temperatures, high and low energies, wea
and strong magnetic fields, and immen
dimensions that cannot be reproduced
laboratories or under terrestrial condition
The extreme astrophysical environmen
Department of Physics and Astronomy, The JohnsHopkins University, 3400 North Charles Street,Baltimore, MD 21218, USA. E-mail: [email protected]
E I N S T E I N ’ S L E G A C YE I N S T E I N ’ S L E G A C Y
www.sciencemag.org SCIENCE VOL 307 11 FEBRUARY 2005