et al.
et al.
et al.
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
REPORTS<br />
794<br />
matrices for this graph at momenta –p/4 and<br />
–p/2 are<br />
Sswitchð−p=4Þ ¼<br />
0 0 e−ip=4 0 −1 0<br />
e−ip=4 0<br />
@<br />
1<br />
A<br />
0 0<br />
0<br />
1<br />
Sswitchð−p=2Þ ¼@<br />
0<br />
0<br />
0<br />
1<br />
0<br />
−1 A ð5Þ<br />
0 −1 0<br />
The momentum switch has perfect transmission<br />
b<strong>et</strong>ween vertices 1 and 3 at momentum –p/4 and<br />
perfect transmission b<strong>et</strong>ween vertices 2 and 3 at<br />
momentum –p/2. Thus, the path a particle follows<br />
through the switch depends on its momentum:<br />
A particle with momentum –p/2 follows the<br />
double line in Fig. 3A, whereas a particle with<br />
momentum –p/4 follows the single line.<br />
The graph used to implement the Cq gate is<br />
shown in Fig. 3B [see section S4 of (32) forthe<br />
numbers of vertices on each of the paths]. To see<br />
why this graph implements a Cq gate, consider<br />
the movement of two particles as they pass<br />
through the graph. If either particle begins in the<br />
state j0in〉, it travels <strong>al</strong>ong a path to the output<br />
without interacting with the second particle. When<br />
either particle begins in the state j1in〉,itisrouted<br />
onto the vertic<strong>al</strong> path as it passes through the first<br />
momentum switch and is routed to the right as it<br />
passes through the second switch. If both particles<br />
begin in the state j1in〉, they interact on the<br />
vertic<strong>al</strong> path and the wave function acquires a<br />
phase e iq .<br />
To implement a circuit, the subgraphs representing<br />
circuit elements are connected by paths.<br />
Figure 4 depicts a graph corresponding to a simple<br />
two-qubit computation. Timing is important:<br />
Wave pack<strong>et</strong>s must me<strong>et</strong> on the vertic<strong>al</strong> paths for<br />
interactions to occur. We achieve this by choosing<br />
the numbers of vertices on each of the segments<br />
in the graph appropriately, taking into account<br />
the different propagation speeds of the two wave<br />
pack<strong>et</strong>s [see section S4 of (32)]. In section S3.1<br />
of (32), we present a refinement of our scheme<br />
using planar graphs with maximum degree four.<br />
By an<strong>al</strong>yzing the full (n +1)–particle interacting<br />
many-body system, we prove that our <strong>al</strong>gorithm<br />
performs the desired quantum computation<br />
up to an error term that can be made arbitrarily<br />
sm<strong>al</strong>l (32). Our an<strong>al</strong>ysis goes beyond the scattering<br />
theory discussion presented above; we take<br />
into account the fact that both the wave pack<strong>et</strong>s<br />
and the graphs are finite. Specific<strong>al</strong>ly, we prove<br />
that by choosing the size of the wave pack<strong>et</strong>s, the<br />
number of vertices in the graph, and the tot<strong>al</strong><br />
evolution time to be polynomi<strong>al</strong> functions of both<br />
n and g, the error in simulating an n-qubit, g-gate<br />
quantum circuit is bounded above by an arbitrarily<br />
sm<strong>al</strong>l constant [section S5 of (32)]. For<br />
example, for the Bose-Hubbard model and for<br />
the nearest-neighbor interaction model, we prove<br />
that the error can be made arbitrarily sm<strong>al</strong>l by<br />
choosing the size of the wave pack<strong>et</strong>s to be<br />
O(n 12 g 4 ), the tot<strong>al</strong> number of vertices in the<br />
graphtobeO(n 13 g 5 ), and the tot<strong>al</strong> evolution time<br />
to be O(n 12 g 5 ). The bounds we prove, <strong>al</strong>though<br />
<strong>al</strong>most certainly not optim<strong>al</strong>, are sufficient to establish<br />
univers<strong>al</strong>ity with only polynomi<strong>al</strong> overhead.<br />
Because it is <strong>al</strong>so possible to efficiently<br />
simulate a multiparticle quantum w<strong>al</strong>k of the type<br />
we consider using a univers<strong>al</strong> quantum computer,<br />
this model exactly captures the power of<br />
quantum computation.<br />
References and Notes<br />
1. E. Farhi, S. Gutmann, Phys. Rev. A 58, 915 (1998).<br />
2. D. Aharonov, A. Ambainis, J. Kempe, U. Vazirani,<br />
“Quantum w<strong>al</strong>ks on graphs,” Proceedings of the 33rd<br />
ACM Symposium on Theory of Computing (2001), pp. 50–59.<br />
3. A. Ambainis, E. Bach, A. Nayak, A. Vishwanath,<br />
J. Watrous, “One-dimension<strong>al</strong> quantum w<strong>al</strong>ks,”<br />
Proceedings of the 33rd ACM Symposium on Theory of<br />
Computing (2001), pp. 37–49.<br />
4. A. M. Childs <strong>et</strong> <strong>al</strong>., “Exponenti<strong>al</strong> <strong>al</strong>gorithmic speedup by<br />
quantum w<strong>al</strong>k,” Proceedings of the 35th ACM Symposium<br />
on Theory of Computing (2003), pp. 59–68.<br />
5. A. Ambainis, SIAM J. Comput. 37, 210 (2007).<br />
6. E. Farhi, J. Goldstone, S. Gutmann, Theory of Computing<br />
4, 169 (2008).<br />
7. A. M. Childs, Phys. Rev. L<strong>et</strong>t. 102, 180501 (2009).<br />
8. B. Do <strong>et</strong> <strong>al</strong>., J. Opt. Soc. Am. B 22, 499 (2005).<br />
9. M. Karski <strong>et</strong> <strong>al</strong>., Science 325, 174 (2009).<br />
10. H. B. Per<strong>et</strong>s <strong>et</strong> <strong>al</strong>., Phys. Rev. L<strong>et</strong>t. 100, 170506 (2008).<br />
11. Y. Bromberg, Y. Lahini, R. Morandotti, Y. Silberberg,<br />
Phys. Rev. L<strong>et</strong>t. 102, 253904 (2009).<br />
12. A. Peruzzo <strong>et</strong> <strong>al</strong>., Science 329, 1500 (2010).<br />
13. J. O. Owens <strong>et</strong> <strong>al</strong>., New J. Phys. 13, 075003 (2011).<br />
14. L. Sansoni <strong>et</strong> <strong>al</strong>., Phys. Rev. L<strong>et</strong>t. 108, 010502 (2012).<br />
15. I. L. Chuang, Y. Yamamoto, Phys. Rev. A 52, 3489 (1995).<br />
16. G. K. Brennen, C. M. Caves, P. S. Jessen, I. H. Deutsch,<br />
Phys. Rev. L<strong>et</strong>t. 82, 1060 (1999).<br />
17. W. S. Bakr, J. I. Gillen, A. Peng, S. Fölling, M. Greiner,<br />
Nature 462, 74 (2009).<br />
Photonic Boson Sampling<br />
in a Tunable Circuit<br />
18. A. J. Hoffman <strong>et</strong> <strong>al</strong>., Phys.Rev.L<strong>et</strong>t.107, 053602 (2011).<br />
19. J.K.Gamble,M.Friesen,D.Zhou,R.Joynt,S.N.Coppersmith,<br />
Phys.Rev.A81, 052313 (2010).<br />
20. J. Smith, Electron. Notes Discr<strong>et</strong>e Math. 38, 795 (2011).<br />
21. Y. Omar, N. Paunković, Phys. Rev. A 74, 042304 (2006).<br />
22. P. K. Pathak, G. S. Agarw<strong>al</strong>, Phys. Rev. A 75, 032351 (2007).<br />
23. Y. Lahini <strong>et</strong> <strong>al</strong>., Phys. Rev. A 86, 011603 (2012).<br />
24. A. Schreiber <strong>et</strong> <strong>al</strong>., Science 336, 55 (2012).<br />
25. A. Ahlbrecht <strong>et</strong> <strong>al</strong>., New J. Phys. 14, 073050 (2012).<br />
26. P. P. Rohde, A. Schreiber, M. Štefaňák,I.Jex,C.Silberhorn,<br />
New J. Phys. 13, 013001 (2011).<br />
27. B.M.Terh<strong>al</strong>,D.P.DiVincenzo,Phys.Rev.A65, 032325 (2002).<br />
28. S. Aaronson, A. Arkhipov, “The computation<strong>al</strong> complexity<br />
of linear optics,” Proceedings of the 43rd ACM<br />
Symposium on Theory of Computing (2011), pp. 333–342.<br />
29. R. Ionicioiu, P. Zanardi, Phys. Rev. A 66, 050301 (2002).<br />
30. A. Mizel, D. A. Lidar, M. Mitchell, Phys. Rev. L<strong>et</strong>t. 99,<br />
070502 (2007).<br />
31. A. M. Childs, D Goss<strong>et</strong>, J. Math. Phys. 53, 102207<br />
(2012).<br />
32. Materi<strong>al</strong>s and m<strong>et</strong>hods are available as supplementary<br />
materi<strong>al</strong>s on Science Online.<br />
33. M. V<strong>al</strong>iente, Phys. Rev. A 81, 042102 (2010).<br />
34. B. A. Blumer, M. S. Underwood, D. L. Feder, Phys. Rev. A<br />
84, 062302 (2011).<br />
Acknowledgments: This work was supported in part by<br />
MITACS; Natur<strong>al</strong> Sciences and Engineering Research Council of<br />
Canada; the Ontario Ministry of Research and Innovation; the<br />
Ontario Ministry of Training, Colleges, and Universities; and<br />
the U.S. Army Research Office.<br />
Supplementary Materi<strong>al</strong>s<br />
www.sciencemag.org/cgi/content/full/339/6121/791/DC1<br />
Materi<strong>al</strong>s and M<strong>et</strong>hods<br />
Supplementary Text<br />
Figs. S1 to S10<br />
References<br />
10 September 2012; accepted 17 December 2012<br />
10.1126/science.1229957<br />
Matthew A. Broome, 1,2 * Alessandro Fedrizzi, 1,2 S<strong>al</strong>eh Rahimi-Keshari, 2 Justin Dove, 3<br />
Scott Aaronson, 3 Timothy C. R<strong>al</strong>ph, 2 Andrew G. White 1,2<br />
Quantum computers are unnecessary for exponenti<strong>al</strong>ly efficient computation or simulation if the<br />
Extended Church-Turing thesis is correct. The thesis would be strongly contradicted by physic<strong>al</strong><br />
devices that efficiently perform tasks believed to be intractable for classic<strong>al</strong> computers. Such a task<br />
is boson sampling: sampling the output distributions of n bosons scattered by some passive, linear<br />
unitary process. We tested the centr<strong>al</strong> premise of boson sampling, experiment<strong>al</strong>ly verifying that<br />
three-photon scattering amplitudes are given by the permanents of submatrices generated from a<br />
unitary describing a six-mode integrated optic<strong>al</strong> circuit. We find the protocol to be robust, working<br />
even with the unavoidable effects of photon loss, non-ide<strong>al</strong> sources, and imperfect d<strong>et</strong>ection.<br />
Sc<strong>al</strong>ing this to large numbers of photons should be a much simpler task than building a univers<strong>al</strong><br />
quantum computer.<br />
Amajor motivation for sc<strong>al</strong>able quantum<br />
computing is Shor’s <strong>al</strong>gorithm (1),<br />
which enables the efficient factoring of<br />
large composite numbers into their constituent<br />
primes. The presumed difficulty of this task is the<br />
basis of the majority of today’s public-key encryption<br />
schemes. It may be that sc<strong>al</strong>able quantum<br />
computers are not re<strong>al</strong>istic if, for example,<br />
quantum mechanics breaks down for large numbers<br />
of qubits (2). If, however, quantum com-<br />
15 FEBRUARY 2013 VOL 339 SCIENCE www.sciencemag.org<br />
puters are re<strong>al</strong>istic physic<strong>al</strong> devices, then the<br />
Extended Church-Turing (ECT) thesis—that<br />
any function efficiently computed on a re<strong>al</strong>istic<br />
physic<strong>al</strong> device can be efficiently computed<br />
on a probabilistic Turing machine—means that<br />
a classic<strong>al</strong> efficient factoring <strong>al</strong>gorithm exists.<br />
Such an <strong>al</strong>gorithm, long sought after, would<br />
enable us to break public-key cryptosystems<br />
such as RSA. A third possibility is that the ECT<br />
thesis itself is wrong.<br />
Downloaded from<br />
www.sciencemag.org on February 14, 2013