et al.
et al.
et al.
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
PERSPECTIVES<br />
768<br />
A Start B<br />
Start<br />
Finish X Finish X<br />
C<strong>al</strong>culating with quantum random w<strong>al</strong>ks. (A) A<br />
classic<strong>al</strong> random w<strong>al</strong>k process in which M particles<br />
are initi<strong>al</strong>ly located in N paths or channels. There is<br />
some probability that a particle will hop to an adjacent<br />
channel at any moment in time. (B) A quantum<br />
random w<strong>al</strong>k process in which the wavelike nature<br />
of each particle spreads out among many channels.<br />
Non<strong>et</strong>heless, each particle will be found in a<br />
single channel when its position is measured at the<br />
end of the process, as indicated by the two Xs at the<br />
bottom in this example. The probability of d<strong>et</strong>ecting<br />
the photons is proportion<strong>al</strong> to the output of certain<br />
matrix c<strong>al</strong>culations (the permanent) and <strong>al</strong>lows<br />
these c<strong>al</strong>culations to be performed much faster than<br />
with classic<strong>al</strong> <strong>al</strong>gorithms for large N.<br />
The output probability distribution from<br />
a quantum random w<strong>al</strong>k process does not<br />
simply correspond to a sum of the independent<br />
probabilities for the individu<strong>al</strong> particles<br />
even if there is no physic<strong>al</strong> interaction<br />
b<strong>et</strong>ween the particles (panel B of the<br />
fi gure). The extra interactions arise because<br />
the state of the system must obey the rule<br />
that Ψ(x 1, x 2) = ±Ψ(x 2, x 1) when two indistinguishable<br />
particles are swapped or interchanged.<br />
The plus sign applies to particles<br />
that are known for historic<strong>al</strong> reasons as<br />
bosons, whereas the minus sign applies to<br />
particles known as fermions. Applying this<br />
rule to the wave function produces an effective<br />
attraction b<strong>et</strong>ween identic<strong>al</strong> bosons and<br />
an effective repulsion b<strong>et</strong>ween identic<strong>al</strong> fermions.<br />
For example, the probability of fi nding<br />
two identic<strong>al</strong> fermions at the same location<br />
x 1 = x 2 is zero because that corresponds<br />
to Ψ(x 1, x 1) = –Ψ(x 1, x 1). These effects are<br />
commonly referred to as exchange forces,<br />
even though there is no physic<strong>al</strong> interaction<br />
b<strong>et</strong>ween the particles. Roughly speaking,<br />
the exchange forces provide an effective<br />
interaction that can be used to perform<br />
certain c<strong>al</strong>culations.<br />
Spring <strong>et</strong> <strong>al</strong>., Broome <strong>et</strong> <strong>al</strong>., and Tillmann<br />
<strong>et</strong> <strong>al</strong>. used indistinguishable particles of light<br />
(photons) to implement a quantum random<br />
w<strong>al</strong>k of this kind c<strong>al</strong>led boson sampling.<br />
The photons propagated through a series of<br />
conducting channels known as waveguides<br />
that were fabricated on the surface of a chip.<br />
Neighboring channels were coupled to each<br />
other by bringing them sufficiently close<br />
tog<strong>et</strong>her that a photon had some probability<br />
of hopping to the adjacent waveguide. The<br />
probabilities of d<strong>et</strong>ecting the photons in the<br />
various output channels were then measured<br />
with single-photon d<strong>et</strong>ectors. It is possible<br />
to fabricate a much larger number of waveguides<br />
on the surface of a chip than were<br />
used in these examples, so full-sc<strong>al</strong>e implementations<br />
should be possible in the future.<br />
The probability of d<strong>et</strong>ecting a photon<br />
(a boson) in each of the output channels is<br />
proportion<strong>al</strong> to the so-c<strong>al</strong>led permanent of<br />
a matrix ( 5). The permanent of an N × N<br />
matrix is defi ned as the sum of <strong>al</strong>l products<br />
of N elements of the matrix chosen in such<br />
a way that each row and column appears<br />
only once. The permanent is similar to the<br />
more familiar d<strong>et</strong>erminant aside from the<br />
minus signs that appear in the d<strong>et</strong>erminant.<br />
There are effi cient m<strong>et</strong>hods for c<strong>al</strong>culating<br />
the d<strong>et</strong>erminant of a matrix that use classic<strong>al</strong><br />
computers, but the best-known classic<strong>al</strong><br />
<strong>al</strong>gorithm for c<strong>al</strong>culating the permanent<br />
requires an exponenti<strong>al</strong>ly large number of<br />
computation<strong>al</strong> steps and is not feasible for<br />
large N. The relevant matrices are related<br />
to the coupling coeffi cients b<strong>et</strong>ween the N<br />
input and output channels, which can be<br />
controlled experiment<strong>al</strong>ly.<br />
Experiments of this kind provide a simple<br />
demonstration of the ability of a quantum<br />
system to perform a potenti<strong>al</strong>ly useful<br />
MOLECULAR BIOLOGY<br />
New Tool for Genome Surgery<br />
John van der Oost<br />
A bacteri<strong>al</strong> system that uses RNA to edit DNA is harnessed for engineering mamm<strong>al</strong>ian genomes.<br />
Gene therapy is the holy grail of<br />
human medicine. Many diseases<br />
are caused by a defective gene,<br />
som<strong>et</strong>imes with a mutation as subtle as a<br />
single-nucleotide variation. Before restoration<br />
of such a mutation in a patient’s<br />
genome can take place, the targ<strong>et</strong> nucleotide<br />
sequence has to be cleaved at a single<br />
position, out of 3 billion possibilities.<br />
This degree of precise surgery requires an<br />
enzyme with highly selective targ<strong>et</strong> recognition.<br />
Successful editing of eukaryotic<br />
genomes has been accomplished with<br />
DNA nucleases designed to bear a unique<br />
site that binds to a specifi c DNA sequence.<br />
Laboratory of Microbiology, Wageningen University, 6703 HB<br />
Wageningen, N<strong>et</strong>herlands. E-mail: john.vanderoost@wur.nl<br />
15 FEBRUARY 2013 VOL 339 SCIENCE www.sciencemag.org<br />
Published by AAAS<br />
computation without the need for the quantum<br />
logic operations required for a gener<strong>al</strong>purpose<br />
quantum computer. For larger v<strong>al</strong>ues<br />
of N, this approach may eventu<strong>al</strong>ly provide<br />
the fi rst demonstration of an actu<strong>al</strong> c<strong>al</strong>culation<br />
that can be done faster using quantum techniques<br />
than could be achieved with a classic<strong>al</strong><br />
computer. In addition, Childs <strong>et</strong> <strong>al</strong>. ( 6) have<br />
shown that any c<strong>al</strong>culation can be performed<br />
using quantum random w<strong>al</strong>ks if quantum logic<br />
operations ( 7, 8) b<strong>et</strong>ween the photons are <strong>al</strong>so<br />
included. The combination of these two techniques<br />
may eventu<strong>al</strong>ly lead to the building of a<br />
full-sc<strong>al</strong>e quantum computer.<br />
References<br />
1. J. B. Spring <strong>et</strong> <strong>al</strong>., Science 339, 798 (2013);<br />
10.1126/science.1231692.<br />
2. M. A. Broome <strong>et</strong> <strong>al</strong>, Science 339, 794 (2013);<br />
10.1126/science.1231440.<br />
3. M. Tillmann <strong>et</strong> <strong>al</strong>., http://arxiv.org/abs/1212.2240<br />
(2012).<br />
4. Y. Aharonov, L. Davidovich, N. Zagury, Phys. Rev. A 48,<br />
1687 (1993).<br />
5. S. Aaronson, A. Arkhipov, in Proceedings of ACM Symposium<br />
on the Theory of Computing, STOC (Association for<br />
Computing Machinery, New York, 2011), pp. 333–342,<br />
http://dl.acm.org/citation.cfm?id=1993682.<br />
6. A. M. Childs, D. Goss<strong>et</strong>, Z. Webb, Science 339, 791<br />
(2013).<br />
7. E. Knill, R. Lafl amme, G. J. Milburn, Nature 409, 46<br />
(2001).<br />
8. T. B. Pittman, M. J. Fitch, B. C. Jacobs, J. D. Franson,<br />
Phys. Rev. A 68, 032316 (2003).<br />
10.1126/science.1234061<br />
A major drawback of these protein-guided<br />
systems to “engineer” genomes, however,<br />
is that each new targ<strong>et</strong> sequence requires<br />
laboriously adjusting the specifi city of the<br />
nuclease’s DNA binding site. On pages 819<br />
and 823 of this issue, Cong <strong>et</strong> <strong>al</strong>. ( 1) and<br />
M<strong>al</strong>i <strong>et</strong> <strong>al</strong>. ( 2) describe effi cient genome<br />
editing in human cells based on an RNAguided<br />
system.<br />
Upon identifying the exact genomic<br />
targ<strong>et</strong> site, an endonuclease will cleave<br />
the DNA. Depending on the nature of the<br />
consequenti<strong>al</strong> DNA damage (single-strand<br />
nicks or double-strand breaks) and on the<br />
type of DNA repair system that is activated<br />
in response to the damage (homologous<br />
or nonhomologous recombination), strand<br />
religation may either seamlessly revert to<br />
on February 14, 2013<br />
www.sciencemag.org<br />
Downloaded from