Massachusetts Institute of Technology

Department of Electrical Engineering and Computer Science

6.453 Quantum Optical Communication

Lecture Number 6
Fall 2016

Jeffrey H. Shapiro
c 2006, 2008, 2010, 2012, 2014

Date: Tuesday, September 27, 2016

Reading: For squeezed states:

• C.C. Gerry and P.L. Knight, Introductory Quantum Optics (Cambridge Uni-
versity Press, Cambridge, 2005) Sect. 7.1.

• H.P. Yuen, “Two-photon coherent states of the radiation field,” Phys. Rev. A
13, 2226-2243 (1976).

• L. Mandel and E. Wolf, Optical Coherence and Quantum Optics (Cambridge

University Press, Cambridge, 1995) Sects. 21.1–21.6.

For continuous-spectrum eigenkets:

• W.H. Louisell, Quantum Statistical Properties of Radiation (McGraw-Hill, New

York, 1973) Sect. 1.10.

Introduction
Last time we showed that the annihilation operator, despite its not being Hermitian,
has an overcomplete set of eigenkets—the coherent states {|αi : α ∈ C}—and that
in the limit |α| → ∞ these states give the classical limit for the quantum harmonic
oscillator. In particular, for the coherent state |αi we have the following quantum
measurement statistics

• Measurement of the number operator, N ˆ , yields a classical random variable

2 2
with mean value |α| and variance |α| , so that the signal-to-noise ratio of the
photon number measurement is

hN̂ i2
SNRnumber ≡ = |α|2 → ∞, as |α| → ∞. (1)
ˆ
h∆N i 2

ˆ -measurement
The Chebyschev inequality then implies that the probability that the N
2 2
outcome deviates from |α| by |α| will go to zero, as |α| → ∞, for all  > 0.

1
 • Measurement of the quadrature operator, â1 (t), yields a classical random vari-
able with mean value Re(αe−jωt ), and variance 1/4, so that the signal-to-noise
ratio of the quadrature operator measurement is
hâ1 (t)i2
SNRquad ≡ 2
= 4[Re(αe−jωt )]2 → ∞, for |Re(αe−jωt )| → ∞. (2)
hƉ1 (t)i
Here, the Chebyschev inequality leads to the conclusion that quadrature mea-
surement behavior is essentially noiseless sinusoidal oscillation in the limit of
|α| → ∞.1
The variance-1/4 quadrature fluctuations,
hƉ21 (t)i = hƉ22 (t)i = 1/4, (3)
satisfy the Heisenberg uncertainty principle with equality, and hence the coherent
states are minimum uncertainty-product (MUP) states for the inequality
h∆â21 (t)ih∆â22 (t)i ≥ 1/16. (4)
An especially interesting special case is then the coherent state with zero eigenvalue,
|0i, because this state is also a photon number eigenket, and hence an energy eigenket,
viz.,
ˆ |0i = 0, H|0i
ˆ ~ω
â|0i = 0, N = |0i. (5)
2
The last of these three eigenket-eigenvalue relations shows that the energy in the
state |0i is ~ω/2, the zero-point energy. The second of these three eigenket-eigenvalue
relations shows that there are zero energy quanta (zero photons) in the state |0i, so
it is appropriate and conventional to call this the vacuum state, because it represents
an unexcited state of the oscillator. However, because |0i is a coherent state, its
quadrature variances satisfy (3). The noise that is quantified by these variances is
called zero-point fluctuations, because it is associated with the zero-point energy in
the vacuum state. That zero-point energy leads to a deterministic non-zero result
when the Hamiltonian H ˆ measurement is performed on the vacuum state, whereas a
zero-mean variance-1/4 random variable results when a quadrature operator—â1 (t) or
â2 (t)—measurement is performed on this state is a clear manifestation of something
that we noted earlier: the state of a quantum system and the measurement that is
made determine the statistics of the resulting outcome.
Today’s lecture will be devoted to learning more about MUP states for the Heisen-
berg inequality (4). We begin by developing the minimum uncertainty-product prop-
erty from the perspective of wave functions. We’ll then connect this formulation back
to what we have already seen, from our Dirac-notation treatment of the coherent
states, and then move on to introduce the squeezed states, which are MUP states
with unequal fluctuations in the two quadratures, i.e., these states satisfy
hƉ21 (t)ihƉ22 (t)i = 1/16 with hƉ21 (t)i =
6 hƉ22 (t)i. (6)
1
The same behavior also applies to the other quadrature, â2 (t).

2
Quadrature-Operator Measurements and Wave Functions
On Problem Set 4 you will determine key properties of the quadrature-operator eigen-
kets. Specifically, for â1 = Re(â) and â2 = Im(â), with â = â(0), these eigenkets are
as follows:
âk |αk ik = αk |αk ik , for k = 1, 2, (7)
where the eigenvalues span the continuum −∞ < αk < ∞, and the k subscript on
the right angle-bracket in the ket is to indicate for which quadrature this state is an
eigenket. From Axiom 3a, we know that these eigenkets are infinite length, satisfying
the delta-function orthonormality relation,

k hαk |βk ik = δ(αk − βk ), for k = 1, 2. (8)

We also know that they resolve the identity according to

Z ∞
Iˆ = dαk |αk ikk hαk |, for k = 1, 2, (9)
−∞

and diagonalize their associated quadrature operators,

Z ∞
âk = dαk αk |αk ikk hαk |, for k = 1, 2. (10)
−∞

Furthermore, if the oscillator is in the (finite-energy ↔ unit-length) state |ψi, then

measurement of the quadrature operator âk yields a continuous, classical random
variable whose probability density function is2

p(αk ) = |k hαk |ψi|2 , for k = 1, 2. (11)

The final result that we need from Problem Set 4 is the inner product relation that
links |α1 i1 to |α2 i2 :
e−2jα2 α1
2 hα |α i
2 1 1 = √ . (12)
π
First treatments of quantum mechanics often represent states as wave functions.
We can now connect our Dirac-notation work to that approach. Let |ψi be some
arbitrary (unit-length) state. Then |ψi has two equivalent representation—which
follow from resolving the identity using the |α1 i1 and |α2 i2 kets, respectively—namely,
Z ∞
ˆ
|ψi = I|ψi = dαk (k hαk |ψi)|αk ik , for k = 1, 2. (13)
−∞

2
Because |αk ik has infinite length, the projection postulate clearly does not apply to quadrature
measurements. Later this term, when we see how this measurement is realized via optical homodyne
detection, you will become more comfortable with abandoning the projection postulate for this case.

3
So, with
ψ(α1 ) ≡ 1 hα1 |ψi and Ψ(α2 ) ≡ 2 hα2 |ψi, (14)
we have that
Z ∞ Z ∞
|ψi = dα1 ψ(α1 )|α1 i1 and |ψi = dα2 Ψ(α2 )|α2 i2 , (15)
−∞ −∞

where the wave functions, {ψ(α1 ), Ψ(α2 )}, have squared magnitudes that integrate
to unity over the infinite interval −∞ < αk < ∞, for k = 1, 2, respectively.3 In terms
of these wave functions, the quadrature measurement statistics for â1 and â2 become
the following. If the state of the system is |ψi, we get a classical random variable
with probability density
p(α1 ) = |ψ(α1 )|2 , (16)
when we measure â1 , and we get a classical random variable with probability density
p(α2 ) = |Ψ(α2 )|2 , (17)
when we measure â2 . Now, the inner product relation between the â1 and â2 eigenkets
implies the following relations between the wave functions, {ψ(α1 ), Ψ(α2 )}, for the
state |ψi:
Z ∞
Ψ(α2 ) = 2 hα2 |ψi = dα1 2 hα2 |α1 i1 ψ(α1 ) (18)
−∞

∞
e−2jα2 α1
Z
1
= dα1 ψ(α1 ) √ = √ F [ψ(t)]|f =α2 /π , (19)
−∞ π π
and
Z ∞
ψ(α1 ) = 1 hα1 |ψi = dα2 1 hα1 |α2 i2 Ψ(α2 ) (20)
−∞

∞
e2jα2 α1
Z
1
= dα2 Ψ(α2 ) √ = √ F −1 [Ψ(f )] t=α1 /π
. (21)
−∞ π π
Here,
Z ∞ Z ∞
−j2πf t −1
F[x(t)] ≡ dt x(t)e and F [X(f )] ≡ df X(f )ej2πf t , (22)
−∞ −∞

denote the Fourier transform and its inverse. Simply stated, the â1 and â2 wave
functions for the state |ψi comprise a Fourier transform pair. Applying the Fourier
transform uncertainty principle that you proved on Problem Set 2 then tells us that
Z ∞ Z ∞
2 2 2 2
hâ1 ihâ2 i = dα1 |α1 | |ψ(α1 )| dα2 |α2 |2 |Ψ(α2 )|2 ≥ 1/16. (23)
−∞ −∞
3
Thus, these wave functions live in the Hilbert space L2 [−∞, ∞].

4
If hâi = 0, i.e., if the quadrature measurements are both zero-mean random variables,
then the mean-squares that appear in this inequality become variances, and we regain
the Heisenberg uncertainty principle for the quadratures that we originally derived
in Dirac notation.4
Suppose, for now, that |ψi does have zero-mean quadrature measurements. Then,
from our work on Problem Set 2, we know that equality will occur in the wave-
function form of the Heisenberg uncertainty principle for the quadratures when the
wave functions associated with |ψi are the Gaussian-function Fourier pair
2 2 2 2
e−α1 /4h∆â1 i e−α2 /4h∆â2 i
ψ(α1 ) = and Ψ(α2 ) = , (24)
(2π h∆â21 i)1/4 (2πh∆â22 i)1/4

with hƉ22 i = 1/16hƉ21 i.

Minimum Uncertainty-Product States

The first time you were taught about the Heisenberg uncertainty principle—whether
in qualitative terms in high school, or more quantitatively in undergraduate physics—
it was probably in the context of position and momentum, not the quadrature com-
ponents of the photon annihilation operator. Nevertheless, you were surely told that
you could get a state with very low uncertainty in position, but this came at the
expense of a very large uncertainty in momentum. Conversely, so the story goes, you
could get a state with a very low uncertainty in momentum, but it carried a very
large uncertainty in position. The same is evidently true for the zero-mean MUP
state that we found at the end of the last section: we can have a very low hƉ21 i
but it comes with a very large hƉ22 i, and, conversely, a very low hƉ22 i carries with
it a very large hƉ21 i. So, the coherent states are rather special, because they are
MUP states with equal uncertainties in the two quadratures. Before we can find their
wave function representations, however, we must generalize the work of the preceding
section to allow for non-zero quadrature-measurement mean values. Here, the basic
idea is obvious from Fourier theory, and the form taken by the probability densities
for the â1 and â2 quadrature measurements.
Suppose we have a state whose â1 measurement statistics—governed by its ψ(α1 )
wave function—are zero mean and variance σ12 :

hâ1 i = 0 and hâ21 i = σ12 . (25)

Another state, whose wave function is ψ(α1 − m1 ) will have then have

hâ1 i = m1 and h∆â21 i = h(â1 − m1 )2 i = σ12 , (26)

4
If we had wanted to complicate the notation on Problem Set 2, we could have derived the
Fourier transform uncertainty principle in a form that would have led our wave function result to
immediately yield (4). Instead, we have chosen to take that step in this lecture.

5
as you should verify by checking that ψ(α1 − m1 ) is a valid wave function for a
unit-length ket, and using the probability density function for the â1 measurement to
relate the mean and variance for this shifted wave-function state to the corresponding
quantities for the original state. Because a shift in the time domain corresponds to
a phase shift in the frequency domain, the Fourier transform relation between ψ(α1 )
and Ψ(α2 ) tells us how the â2 wave function of a state is affected when that state’s â1
wave function has been shifted. Obviously, we can do the same thing if we start from
a state whose â2 measurement statistics—governed by its Ψ(α2 ) wave function—are
zero mean and variance σ22 : the state Ψ(α2 − m2 ) will have

hâ2 i = m2 and h∆â22 i = h(â2 − m2 )2 i = σ22 , (27)

and its ψ(α1 ) wave function will undergo a phase shift. With these concepts in
hand—and after some tedious algebra—we have that
exp[2j hâ2 iα1 − jhâ1 ihâ2 i − (α1 − hâ1 i)2 /4h∆â21 i]
ψ(α1 ) = (28)
(2πh∆â21 i)1/4

exp[−2jhâ1 iα2 + j hâ1 ihâ2 i − (α2 − hâ2 i)2 /4h∆â22 i]

Ψ(α2 ) = , (29)
(2πh∆â22 i)1/4
are the wave functions, in the â1 and â2 eigenket representations, of a minimum
uncertainty-product state whose quadrature-measurement means are {hâ1 i, hâ2 i}, whose
quadrature-measurement variances are {hƉ21 i, hƉ22 i}, with the former being arbi-
trary real numbers and the latter being positive numbers obeying hƉ21 i = 1/16hƉ22 i.
We can now obtain the wave function representations of the coherent states. Let
|βi be a coherent state, so that â|βi = β|βi. We know that â1 and â2 measurements
made on this state lead to the following first and second moments:

hâ1 i = β1 , hâ2 i = β2 , h∆â21 i = h∆â22 i = 1/4. (30)

It immediately follows, from our general wave function results for an MUP state, that
exp[2jβ2 α1 − jβ1 β2 − (α1 − β1 )2 ]
ψ(α1 ) = (31)
(π/2)1/4

exp[−2jβ1 α2 + jβ1 β2 − (α2 − β2 )2 ]

Ψ(α2 ) = , (32)
(π/2)1/4
are its wave functions. Note that deriving these wave functions has been more than
a sterile reprise of the minimum uncertainty-product property of the coherent states.
Because |ψ(α1 )|2 and |Ψ(α2 )|2 are the probability densities for the â1 and â2 quadra-
ture measurements, we learn that a coherent state has Gaussian probability densities
for its quadrature-measurement statistics. Indeed, for our general MUP state—which
can have unequal variances in its two quadratures—the quadrature-measurement

6
probability densities are Gaussian. Moreover, from our basic probability theory we
know that these densities are completely characterized by their means and variances.
Let’s take another look at minimum uncertainty-product states for the quadrature
measurements â1 and â2 , this time connecting back to the equality condition that we
established when we derived the Heisenberg uncertainty principle. There we showed
that
hƉ21 ihƉ22 i = 1/16 (33)
will prevail if and only if the state |ψi satisfies

∆â1 |ψi = −jλ∆â2 |ψi, where λ is a real-valued constant. (34)

Rearranging this condition, as shown below, we can reduce it to an eigenket-eigenvalue

relation:
(â1 + jλâ2 )|ψi = (hâ1 i + jλhâ2 i)|ψi. (35)
We are not assured that there are solutions to this eigenket-eigenvalue relation for
all values of λ, but some must exist because we have already identified MUP states
for the quadrature operators’ Heisenberg uncertainty principle via our wave function
analysis.
Suppose that λ is positive. Writing the quadrature
√ operators in terms of the
annihilation and creation operators, and dividing by λ then leads to
1+λ 1−λ
(µâ + ν↠)|ψi = (µhâi + νh↠i)|ψi, where µ ≡ √ and ν ≡ √ . (36)
2 λ 2 λ
If we define a new operator by
b̂ ≡ µâ + ν↠, (37)
then, because µ and ν are real-valued and satisfy µ2 − ν 2 = 1, we have that
ˆ ˆb† ] = (µâ + ν↠)(µâ† + νâ) − (µâ† + νâ)(µâ + ν↠) = µ2 [a,
[b, ˆ ↠] + ν 2 [↠, â] = 1, (38)

i.e., ˆb and ˆb† have the same commutator as the annihilation operator and its adjoint,
the creation operator. The transformation from {a, ˆ ˆb† } is known as a Bo-
ˆ ↠} to {b,
goliubov transformation, and what we have just seen is that it preserves commutator
brackets. Commutator preservation is an essential issue for us, in that it will dictate
the need for quantum noise injection whenever there is loss or amplification of an
electromagnetic field. For now, however, it will provide us with another window into
MUPs with unequal quadrature fluctuations, one which we will ultimately see is how
such states have been generated for the electromagnetic field.
We know—from our wave function results—that there are kets that satisfy (36),
and we will denote them |β; µ, νi, whence

b̂|β; µ, νi = β|β; µ, νi, for β ∈ C. (39)

7
Now, by using the inverse of the Bogoliubov transformation that defines ˆb, i.e.,
â = µbˆ − νbˆ† , (40)
we find that
hβ; µ, ν|â|β; µ, νi = µβ − νβ ∗ , (41)
so that putting the oscillator into this state produces simple harmonic motion in the
mean, viz.,
hβ; µ, ν|â(t)|β; µνi = (µβ − νβ ∗ )e−jωt , (42)
but the eigenket-eigenvalue relation (39) tells us much more. Because ˆb and ˆb† have
the same commutator as â and ↠, it follows that
ˆb ≡ ˆb† b̂
N (43)
behaves like a number operator, i.e., it has non-negative integer eigenvalues with
associated eigenkets {|n; µ, νi} that obey
∞
X ∞
X
hm; µ, ν |n; µ, ν i = δmn , ˆb =
N n|n; µ, νihn; µ, ν|, Iˆ = |n; µ, νihn; µ, ν|.
n=0 n=0
(44)
Note that we have carried along µ and ν as parameters of these number-like states.
This is to emphasize that unless µ = 1 and ν = 0, these states are not the photon-
number eigenkets. However, using the preceding properties of the {|n; µ, νi}, we can
show that
∞ ∞
X √ X √
b̂ = n |n − 1; µ, νihn; µ, ν| and ˆb† = n + 1 |n + 1; µ, νihn; µ, ν|, (45)
n=1 n=0

and hence the {|β; µ, νi} can be regarded as coherent states of the ˆb operator. Later
in the term we will show that second-order nonlinear optics can be used to perform
Bogoliubov transformations. These systems will allow us to start with a coherent
state—such as is emitted by an ideal laser—and transform it into an MUP state with
unequal variances in its two quadratures.
Our next task is to study the quadrature-measurement statistics for an MUP state
with unequal quadrature-measurement variances. We will do so by building on the
Bogoliubov transformation work, which was begun above, generalized to allow for
complex-valued {µ, ν}. We define ˆb by
b̂ = µâ + ν↠, where µ, ν ∈ C and |µ|2 − |ν|2 = 1. (46)
You should verify the [b,ˆ ˆb† ] = 1 still holds. The result of this commutator-bracket
preservation is that Nˆb ≡ ˆb†ˆb still has a complete orthonormal set of eigenkets, {|n :
µ, νi}, associated with its non-negative integer eigenvalues, and that
b̂|β; µ, νi = β|β; µ, νi, for β, µ, ν ∈ C and |µ|2 − |ν|2 = 1, (47)

8
still defines the coherent states of ˆb. To get at the quadrature-measurement statistics—
in this case just the mean and variance behavior—we start with the inverse of the
Bogoliubov transformation that defines ˆb, i.e.,

â = µ∗ b̂ − νbˆ† . (48)

From this we find that

hβ; µ, ν|â(t)|β; µ, νi = (µ∗ β − νβ ∗ )e−jωt , (49)

which is a slight generalization of what we had already shown for the case of µ and ν
real valued. The mean values of â1 (t) and â2 (t) measurements, when the oscillator is
in the state |β; µ, νi, are then just the real and imaginary parts, respectively, of this
result. To find the variances of the quadrature measurements, we must work harder.
The details for hƉ21 (t)i are given below, along with the answer for hƉ22 (t)i. The
derivation of the latter formula is left for you to work out.
We know that h∆â21 (t)i = hâ21 (t)i − hâ1 (t)i2 , with hâ1 (t)i = Re[(µ∗ β − νβ ∗ )e−jωt ].
For the mean-square term, we have that
a2 ie−2jωt + hâ†2 ie2jωt + 2h↠âi + 1
hˆ
hâ21 (t)i = , (50)
4
ˆ ↠] = 1. The average photon number in the state |β; µ, νi,
where we have used [a,
which appears in the preceding expression, is interesting in its own right:

hN̂ i = h↠âi = h(µ∗ b̂ − νbˆ† )† (µ∗ b̂ − νbˆ† )i = h(µbˆ† − ν ∗ˆb)(µ∗ˆb − νbˆ† )i (51)

= |µ|2 hˆb†ˆbi + |ν |2 (hˆb†ˆbi + 1) − µν hˆb†2 i − µ∗ ν ∗ hbˆ2 i (52)

= (|µ|2 + |ν|2 )|β|2 − 2Re(µ∗ ν ∗ β 2 ) + |ν|2 (53)

= |µ∗ β − νβ ∗ |2 + |ν|2 = |hâi|2 + |ν|2 . (54)

The last equality is especially interesting. It shows that the state |0; µ, νi with ν =
6 0,
2 5
which has hâi = 0, has a non-zero average photon number equal to |ν | . To complete
our derivation of hƉ21 (t)i, we note that

hâ2 i = hâ†2 i∗ = h(µ∗ b̂ − νbˆ† )2 i = µ∗2 β 2 + ν 2 β ∗2 − 2µ∗ ν|β|2 − µ∗ ν. (55)

Putting everything together, and doing yet more algebra (which will be omitted),
yields the final result
|µ − νe−2jωt |2
hƉ21 (t)i = . (56)
4
5
When µ = 1 and ν = 0 we have ˆb = â, so that |β; 1, 0i is the coherent state |βi of the annihilation
operator â. The zero-mean state |0; 1, 0i is thus the vacuum state |0i, which has no photons in it.
The zero-mean state |0; µ, νi with ν =
6 0, however, has a nonzero average photon number.

9
A similar derivation for the other quadrature results in
|µ + νe−2jωt |2
hƉ22 (t)i
= . (57)
4
A natural question to ask, at this point, is when (for what values of t) does |β; µ, νi
with |ν| > 0 yield a minimum uncertainty product for the quadrature measurements
â1 (t) and â2 (t). This happens when µ∗ νe−2jωt is positive, in which case
(|µ| − |ν|)2 (|µ| + |ν|)2
hƉ21 (t)i = < 1/4 and hƉ22 (t)i = > 1/4. (58)
4 4
It also happens when µ∗ νe−2jωt is negative, in which case6
(|µ| + |ν|)2 (|µ| − |ν|)2
hƉ21 (t)i = > 1/4 and hƉ22 (t)i = < 1/4. (59)
4 4
At other times, however, |β; µ, νi is not an MUP state for the quadrature operators.
We can get an intuitive feeling for why the Bogoliubov transformation from â to
b̂ creates unequal quadrature variances by looking at the simple case of µ and ν real
valued. We then have that
∆â1 = (µ − ν)∆bˆ1 and ∆â2 = (µ + ν)∆bˆ2 . (60)
We know that |β; µ, νi is a coherent state of ˆb, and so it has
h∆bˆ2 i = h∆bˆ2 i = 1/4.
1 2 (61)
It thus follows that
h∆â21 i = (µ − ν)2 /4 and h∆â22 i = (µ + ν)2 /4, (62)
which shows that when µν > 0 holds we are attenuating the noise in the first quadra-
ture and amplifying the noise in the second quadrature, and vice versa when µν < 0.
We say that this noise transformation is phase sensitive, because a π/2-rad phase
shift is what distinguishes the two quadratures. We say that this noise transforma-
tion is a squeezing transformation, because it is reducing the noise (squeezing the
noise) in one quadrature and, in order to preserve the minimum uncertainty product,
appropriately increasing the noise in the other quadrature. Thus the state |β; µ, νi
is referred to as a squeezed state. Also, because the zero-mean coherent state |0i is
the vacuum state, we call the zero-mean squeezed state, |0; µ, νi, a squeezed vacuum
state. The squeezed vacuum state plays an essential role in the quantum waveguide
tap that was described in Lecture 1.
We’ll conclude today’s lecture with a qualitative look at two special cases of
squeezed states, shown on Slide 12. In one case, we have reduced the noise that
is in phase with the mean, obtaining an amplitude-squeezed state. In the other case,
we have reduced the noise that is in quadrature with (π/2 rad out of phase from) the
mean, obtaining a phase-squeezed state.
6
Note that h∆â21 (t)ih∆â22 (t)i = (|µ| − |ν|)2 (|µ| + |ν|)2 /16 = (|µ|2 − |ν|2 )2 /16 = 1/16, for the two
mininmum uncertainty product cases we have identified, as advertised.

10
The Road Ahead
The squeezed states are an important class of non-classical states, i.e., putting a single-
mode electromagnetic field into one of these states results in quantum photodetection
statistics that cannot be described by classical electromagnetism and shot noise. We
are almost, but not quite, ready to begin a treatment of quantum photodetection for
single-mode fields. The last things to learn before doing so will be covered in the next
lecture, where we will introduce quantum characteristic functions, and describe what
it means to “measure” the annihilation operator.7

7
Because â is a non-Hermitian operator whose real and imaginary parts are non-commuting
observables, nothing we have done so far implies what it means to say that â can be “measured.”

11
MIT OpenCourseWare
https://ocw.mit.edu

6.453 Quantum Optical Communication

Fall 2016

For information about citing these materials or our Terms of Use, visit: https://ocw.mit.edu/terms.