Quantum Input-Output Theory For Optical Cavities W
Quantum Input-Output Theory For Optical Cavities W
Quantum Input-Output Theory For Optical Cavities W
net/publication/256926513
Article in Physical review A, Atomic, molecular, and optical physics · September 2013
DOI: 10.1103/PhysRevA.88.043819 · Source: arXiv
CITATIONS READS
24 1,248
2 authors:
All content following this page was uploaded by Michael Raymer on 17 September 2014.
Quantum Input-Output Theory for Optical Cavities with Arbitrary Coupling Strength: Application to
Two-Photon Wave-Packet Shaping
M. G. Raymer1 and C. J. McKinstrie2
1
Department of Physics, University of Oregon, Eugene, Oregon 97403, USA
2
Bell Laboratories, Alcatel-Lucent, Holmdel, New Jersey 07733, USA
We develop quantum-optical input-output theory for resonators with arbitrary coupling strength, and for input fields whose
spectrum can be wider than the cavity free-spectral range, while ensuring that the field-operator commutator relations in
space-time variables are correct. The cavity-field commutator exhibits a series of space-time ‘echoes,’ representing causal
connections of certain space-time points by light propagation. We apply the theory to two-photon wave-packet shaping by
cavity reflection, which displays a remarkable illustration of dispersion cancellation. We also show that the theory is
amenable to inclusion of intracavity absorbing and emitting atoms, allowing, for example, dissipative losses within the cavity
to be incorporated in a quantum mechanically correct way.
1
arXiv:1309.5915 [quant-ph] /Oct 5 2013
and in the absence of backscattering and nonlinear normalized electric field amplitude (or photon annihilation
coupling, the two counter-propagating fields evolve operator) C(z,t) obeys the partial differential traveling-
independently of each other and so we can ignore one of wave Maxwell equation:
them. (However, one should be aware that if there is a
structure in the cavity medium that is capable of acting as a
nonlinear coupling or as a diffraction grating, such as a
( ∂ + v ∂ ) C(z,t) = α P(z,t)
,
(1)
t z
As mentioned above, it is simplest to consider only a These relations enter the theory here as cavity-field
single input beam and a single output beam, in which case boundary conditions rather than as the (sometimes) ad hoc
the field in the cavity is a traveling-wave one with no relations that are postulated in order to maintain
standing-wave effects. It is straightforward to generalize to commutation relations in free space. [14] As in those
the case of two counter-propagating inputs and therefore treatments, the minus sign on − ρ makes the matrix in Eq.
standing waves in the cavity. In addition, only a single (4) unitary. In fact, we will show that, rather than maintain
transverse mode is considered, although it is free-space commutation relations, Eqs. (1) – (3) lead to
straightforward to include more such modes. significant alterations to the commutation relations in a
In the presence of a local electric-dipole polarization way that maintains causality and unitarity in the theory.
P(z,t) (which equals zero for an empty cavity), the
2
arXiv:1309.5915 [quant-ph] /Oct 5 2013
∫ dte
iω t
f (z, ω ) = F(z,t)
.
(7)
−∞
(Throughout the paper we denote frequency-domain
functions by lower-case letters.) Then Eqs. (1) – (5) imply
3
arXiv:1309.5915 [quant-ph] /Oct 5 2013
output field differs from the input field only by a unit- C(0 + ,t) = ∫ 2π e
−iω t
τ a(ω )∑ ρ n einω T
magnitude factor with a frequency-dependent phase θ (ω ) . −∞ n=0
∞ ∞
This fact is consistent with energy conservation.
=τ ∫ dt 'A(t ')∑ ρ δ (t − nT )
(22)
n
= exp(−i θ (ω ))b(ω )
.
(18)
The first term in the sum is the beam-splitter transfer
= Gba (ω ) b(ω ) * function, while the remaining terms are delayed and
attenuated replicas (‘echoes’) of the input. This, too, can be
expressed in linear-response form, by introducing the time-
The inverse Green function is simply the complex
domain Green function G ca (t) , which is the Fourier-
conjugate of the forward one: Gab (ω ) = Gba (ω )* . As can
transform of Gca (ω ) . Then
be seen from (17), the complex conjugate corresponds to
∞
simply replacing T by –T in Gba (ω ) .
The above relations are common lore in cavity theory,
C(0 + ,t) = ∫ dt 'G
−∞
ca (t − t ')A(t ')
,
(23)
but to the best of our knowledge they have not previously where
been exploited to derive simple input-output relations for ∞
quantum fields. To further this goal, we derive the
G ∑
ca (t) = τ ρ nδ ( t − nT )
.
(24)
quantum-mechanical ramifications of the above relations. n=0
τ
2
= ∑ρ |k|
δ (t − t '− kT ) .
k=−∞
= 2πδ (ω − ω ') .
1− ρ exp(iω T )
(20)
This shows that the cavity field at position z = 0 +
commutes with itself at all times except those separated by
This commutator is different from the free-space one, and integer multiples of the cavity round-trip time; that is, at
reflects the increase of the density of states near the cavity those times that are causally connected by the speed of
resonances. light in the cavity. In the limit that there is no cavity, i.e.
The cavity field can be expressed in a different way ρ → 0 , this recovers the free-space relation Eq. (5), as
using a Taylor-series expansion:
expected. For nonzero ρ the commutator decays as ρ |k| ,
indicating loss of memory or correlation between widely
τ a(ω ) ∞
c(0 + , ω ) = = τ a(ω )∑ ρ n einω T
.
(21)
separated times.
1− ρ exp(iω T ) n=0 The commutator in Eq. (25) can easily be generalized
to account for different positions in the cavity in similar
Transforming back to the time domain yields manner to Eq. (6) for the input field. Note that Eq. (1) with
P = 0 implies C(z,t) = C(0 + ,t − z / v) ; then Eq. (25)
implies
∞
⎡⎣C(z,t),C † (z ',t ') ⎤⎦ = ∑ρ |k|
δ (t − z / v − (t '− z '/ v) + kT ) .
k=−∞
(26)
4
arXiv:1309.5915 [quant-ph] /Oct 5 2013
For equal times this becomes Causality is satisfied. This can also be generalized to
account for different positions in the cavity, as in Eq. (26).
∞ The commutator between the cavity field and the output
⎡⎣C(z,t),C † (z ',t) ⎤⎦ = ∑ρ |k|
δ ((z '− z + k L) / v)
,
field can be found by similar means.
k=−∞
The output field B(t) , being a freely propagating field
= δ ((z '− z) / v)
in free space, must obey the same commutation relations as
(27)
does the input field, i.e., Eqs. (5) and (19). This is easy to
where the sum does not contribute because z is contained see in the frequency domain, where the commutator, from
within the interval z ∈[0, L] . This agrees with the Eqs. (16)–(19), is
fundamental field commutator result verified in [7], where
it was noted that Eq. (27) is the same inside and outside the 2
⎡⎣b(ω ),b † (ω ') ⎤⎦ = Gba (ω ) 2πδ (ω − ω ')
cavity, as it must be. What that report left unsaid is that
(29)
when different times are considered, as in Eq. (26), the = 2πδ (ω − ω ') .
commutator inside the cavity is not the same as that of the
input field. The temporal evolution – here manifested as
On the other hand, the mapping between the input and
echoes – affects the commutator, consistent with causality.
output fields is nontrivial when expressed in the time
Figure 3 illustrates the space-time structure of the
domain. To derive this, express the output-field Green
commutator, from Eq. (26). Slanted white lines indicate
function in the frequency domain, Eq. (17), as
values of t and z where the commutator is non-zero. For
t=0 (horizontal white lines), it can be seen that as z’
increases from 0 to L, the spatial position where the ⎛ τ 2 exp(iω T ) ⎞
Gba (ω ) = − ρ + ⎜
commutator is nonzero (indicated by the line crossings) ⎝ 1− ρ exp(iω T ) ⎟⎠
moves with z’. There is a single crossing for any fixed
(30)
∞
value of z’, in agreement with Eq. (27). In contrast, for = −ρ + τ 2
∑ρ n−1 i nω T
e .
fixed z, z’, and t’ values, there are an infinite number of n=1
∞
G ∑
ba (t) = − ρδ (t) + τ 2 ρ n−1δ (t − nT )
,
(31)
n=1
Therefore, the output field in the time domain is:
∞
Fig. 3 (Color online) Modulus of the cavity-field
commutator, versus z and t, for t ' = 0 , and (a) z ' = 0 ,
B(t) = ∫ dt 'G
−∞
ba (t − t ')A(t ')
,
(32)
(b) z ' = 0.333 , and (c) z ' = 0.666 . The cavity length
or
L=1; the speed of light v = 1; time t is measured in ∞
units L/v. Horizontal white lines indicate t=0 regions.
B(t) = − ρ A(t) + τ 2 ∑ ρ n−1 A(t − nT )
.
(33)
Mirror reflectivity ρ 2 = 0.998 . n=1
We can also calculate the commutator between the The first term here is the beam-splitter transfer function,
cavity field at position z = 0 + and the input field: while the terms in the sum are ‘echoes.’ It can be shown
that Eq. (33) is consistent with Eqs. (3), (12), and (22).
∞
From this result, the commutation relation for the output
⎡⎣C(0 + ,t), A† (t ') ⎤⎦ = τ ∑ ρ n ⎡⎣ A ( t − nT ) , A† (t ') ⎤⎦ field in the time domain can be derived as a consistency
n=0 check, and indeed is found to be
∞
= τ ∑ ρ δ (t − t '− nT ) .
n
⎡⎣ B(t), B† (t ') ⎤⎦ = δ (t − t ')
,
(34)
n=0
(28)
The cavity field C(0 + ,t) commutes with the input field that is, the same free-space commutator that is obeyed by
A (t ') for all times t < t ' because later values of the input
† the input field. The derivation is given in Appendix 2. The
generalized form in Eq. (6) also holds for the output field.
field cannot affect the cavity field at earlier times.
This reflects the fact that the output field, because it is
5
arXiv:1309.5915 [quant-ph] /Oct 5 2013
traveling in free space, can in principle be measured with If the conditions i - iii are met, meaning that κ T << 1 , then
arbitrarily high precision simultaneously at distinct space- the sum in Eq. (39) can be well approximated by an
time points not connected by causal propagation. integral:
The inverse relation can be written in the time domain t
τ
∫ dt ' T e
−κ (t−t ')
as well; from Eq. (18), we saw that it corresponds to
C(0 + ,t) ≈ A(t ')
,
(40)
simply replacing T by –T. Thus −∞
∞ where we used t ' = t − nT and noted that n ≥ 0 implies
A(t) = ∫ dt 'G ab (t − t ')B(t ')
,
(35)
t ' ≤ t . Comparing this to Eq. (23) shows that in this limit
−∞ the Green function can be effectively replaced by
where
∞
G ∑
ab (t) = − ρδ (t) + τ 2 ρ n−1δ (t + nT )
,
(36)
ca (t − t ') = τ e−κ (t−t ')Θ(t − t ')
,
(41)
G eff
n=1 T
so,
∞
where Θ(x) is the Heaviside step (theta) function.
A(t) = − ρ B(t) + τ 2
∑ρ n−1
B(t + nT )
.
(37)
Transforming this to the frequency domain gives
n=1
∞
τ −κ t τ /T
∫ dte
iω t
This result can be verified explicitly by substituting Eq.
Gca (ω )eff = e Θ(t) =
,
(42)
(33) into Eq. (37) and performing the sums. −∞
T κ − iω
3. Reduction to Standard High-Q Input-Output Theory a complex Lorentzian, as expected for a single, narrow
resonance of a cavity. The same result is obtained directly
To verify that the theory above includes the standard I- by considering Eq. (14) in the limit ω T → 0 , and using
O theory [1] as a limiting case, we consider the limit ρ = e−κ T ≈ 1− κ T ≈ 1− τ 2 / 2 , which implies κ ≈ (1− ρ ) / T
regime satisfying three conditions:
i. The junction transmission coefficient τ is very and κ ≈ τ 2 / 2T , which is standard in high-Q cavity theory.
small, so the cavity storage time is long. Near the resonance, the shape of the spectral response
ii. The cavity round-trip time T is very small, in Eq. (42) is similar to the exact result given by Eq. (14),
compared to the duration of the input field pulse. so it might seem tempting to apply the approximate form
iii. The input field is narrow band, so it is well even in the intermediate-loss regime, where ρ is
contained within a single FSR of the cavity. significantly different than 1. The problem with this idea is
For a transform-limited input field, conditions ii and iii are that the approximate form Eq. (42) has a maximum value
equivalent. τ / ln(1 / ρ ) , whereas the exact result has maximum value
τ / (1− ρ ) . These agree only in the limit ρ → 1 , which is
Define a cavity damping rate κ by κ = (1 / T )ln(1 / ρ ) ,
the high-Q regime. Therefore, we restrict application of
so that ρ = e−κ T , which gives two ways to write the Eq. (42) to the high-Q regime.
attenuation factor suffered by the field on each trip around Standard I-O theory in the high-Q regime can be
the cavity. Then we can write, without approximation, recovered easily by noting that the solution in Eq. (40)
(valid in the limit ω T → 0 ) satisfies the following
∞ differential equation:
G ∑
ca (t) = τ e−κ nT δ ( t − nT )
.
(38)
n=0
τ
∂t C(0 + ,t) = −κ C(0 + ,t) + A(t)
.
(43)
T
In order to consider how G ca (t) behaves in the high-Q
limit, note that it is a distribution (not a function), so it has This fundamental equation of motion for the cavity field is
meaning only as a factor inside an integral. The relevant supplemented with the output-field equation, Eq. (3):
integral is Eq. (23), which gives
B(t) = τ C(L− ,t) − ρ A(t)
.
(44)
∞
C(0 + ,t) = ∑ τ e −κ nT
A(t − nT )
.
(39)
n=0 The goal of standard I-O theory is to be able to treat
the cavity field as an effective single mode, called a
‘quasimode,’ with annihilation operator C(t) that obeys
6
arXiv:1309.5915 [quant-ph] /Oct 5 2013
the commutator [C(t),C † (t)] = 1 . To this end, we note that Eq. (47), using κ ≈ (1− ρ ) / T . (a) ρ = 0.97 , (b)
in the limit ω T → 0 , the effect of one round trip is ρ = 0.70 . (Delta functions are represented by narrow
negligible, so Eq. (11) and (12) imply that Gaussians for visualization.)
c(L− , ω ) ≈ c(0 + , ω ) and C(L− ,t) ≈ C(0 + ,t) , so we define
4. Cavity Shaping of Time-Frequency Entangled Two-
C(t) = T C(0 + ,t) , where we also introduced a scaling
Photon Wave-Packets
factor T . This makes the cavity field dimensionless.
Then, also using κ ≈ τ 2 / 2T , we find A standard example of non-classical light is the time-
frequency-entangled photon pair. [16, 17, 18, 19] It can
exhibit violations of Bell inequalities [16], violation of
∂t C(t) = −κ C(t) + 2κ A(t)
.
(45)
classical Maxwell electromagnetic theory [20, 21], and is
useful in quantum cryptographic key distribution [22],
The output-field equation becomes, in the limit ρ → 1 among other applications. It is easily created using
spontaneous parametric down-conversion in crystals [23,
24] or spontaneous four-wave mixing in fibers [25, 26, 27,
B(t) = 2κ C(t) − A(t)
.
(46)
28], and since it contains only two photons is fully
characterized by its fourth-order electric-field correlation
The commutator of the (rescaled) quasimode operator is function,
easily found from Eq. (40) to be:
f (t1 ,t 2 ) = Ψ A† (t1 )A† (t 2 )A(t 2 )A(t1 ) Ψ
,
(48)
[C(t),C (t ')] = exp(−κ | t − t ' |)
.
(47)
†
where the field operator is (if the light is not too broad
This reduces to [C(t),C (t)] = 1 for equal times, justifying
† band)
the scaling factor that we used. Equations (45) - (47) are
dω
the standard I-O theory for high-Q cavities, originally
A(t) = ∫ a(ω )e−iω t
.
(49)
derived using a master-equation method. [1] Note that the 2π
commutator Eq. (34) is exactly upheld even with the
approximations made in arriving at Eqs. (45) and (47). The state can be expressed equivalently in the frequency or
It is helpful to compare graphically the forms of the time domains as
commutator in the exact and approximate theories, as in
Fig. 4. For high junction reflectivity, ρ = 0.97 , the Ψ = (2π )−2 ∫ dω ∫ dω 'ψ (ω , ω ')a † (ω )a † (ω ') vac
approximate commutator (using κ ≈ (1− ρ ) / T ) acts like
= ∫ dt ∫ dt 'ψ (t,t ') A† (t)A† (t ') vac ,
an accurate envelope for the exact commutator, which is a
sum of delta functions. However, for ρ = 0.70 , the
(50)
approximate result deviates significantly from the true
where ψ (t,t ') is the double Fourier transform of ψ (ω , ω ')
envelope of the delta functions. By using the exact and the field operators obey the commutators Eq. (19) or
expression for the damping rate, κ = (1 / T )ln(1 / ρ ) , the Eq. (5). (Note that we are working in the Heisenberg
approximate commutator can be made to decay at the exact picture, where the state Ψ is time independent.) The
same rate as the exact commutator. But then, as stated modulus-squared of the two-photon probability amplitude
above, the magnitudes of the Green functions in the 2
frequency domain do not agree quantitatively unless ψ (t1 ,t 2 ) gives the joint probability to detect photons at
(1− ρ ) << 1 , which is the high-Q limit. both times t1 and t 2 , and is determined by the properties
of the down-conversion crystal and the laser field used to
pump it. [29, 30, 31] Likewise, ψ (ω , ω ') 2 gives the joint
spectral density – the probability to detect photons at both
frequencies ω and ω ' .
The correlation function in Eq. (48) is the inner
product of A(t 2 )A(t1 ) Ψ with its hermitian conjugate. So
Fig. 4 (Color online) Cavity-field commutator we evaluate:
†
[C(t),C (t ')] versus time difference; solid - exact
from Eq. (28); dashed - standard approximation from
7
arXiv:1309.5915 [quant-ph] /Oct 5 2013
(59)
unimodular, this confirms that there is no change of the
joint spectral density as a consequence of passing through
the cavity.
8
arXiv:1309.5915 [quant-ph] /Oct 5 2013
This clearly holds the possibility for photon-counting remarkable example of the cancellation of dispersion that
coincidences to occur at any combinations of delays is known for two-photon light with perfect frequency
suffered separately by the two photons. But two-photon anticorrelation. [37]
quantum interference can eliminate some of these The case of a non-stationary two-photon source is also
possibilities under certain conditions. of interest. This occurs if the pump field is pulsed. In a
As an example, first consider the common case of a special case we can model the wave function as a two-
stationary down-conversion source, pumped by a constant dimensional Gaussian, with parameter σ giving the
(cw) laser field. In this case the two-time wave function correlation time, and a second parameter β giving the
depends only on the time difference, Φ(t1 ,t 2 ) = D(t1 − t 2 ) , pulse duration,
with the width of the function D being the coherence time.
This implies a two-time wave function in the frequency Φ(t1 ,t 2 ) = exp[−(t1 + t 2 )2 / 2 β 2 ]exp[−(t1 − t 2 )2 / 2σ 2 ]
.
domain proportional to
(63)
Then, from Eq. (59) and Eq. (63),
ϕ (ω , ω ') ∝ δ (ω + ω ') ∫ dτ ei(ω −ω ')τ D(τ )
,
(60)
(
Φb (t1 ,t 2 ) = τ 2 F0 + ρ 2 exp[−(t1 + t 2 )2 / 2 β 2 ] )
× exp[−(t1 − t 2 ) / 2σ ]
2 2
showing the perfect frequency anticorrelation that is
∞
characteristic of this form of time-frequency entanglement.
To evaluate the two-time wave function of the output field (
+τ 2 ∑ Fm − ρ m exp[−(t1 + t 2 − mT )2 / 2 β 2 ] )
in this case, we use the math relation
m=1
(64)
×exp[−(t1 − t 2 + mT )2 / 2σ 2 ]
∞
( )
∞ ∞
ρ 2 ∞ where
= ∑ ρ |k| D(t1 − t2 + kT ) . ∞
1− ρ 2 k=−∞
Fm = τ 2 ∑ ρ 2 s−2 exp[−(t1 + t 2 − 2sT )2 / 2 β 2 ]
.(65)
s=|m|/2+1
Then we find The function Fk goes to ρ |k| in the limit β → ∞ , thus
∞
recovering the result Eq.(62).
Φ out (t1 ,t 2 ) = ρ 2 D(t1 − t 2 ) − τ 2 ∑ ρ m D(t1 − t 2 + mT ) Consider the case of equal σ and β , so the wave
m=1 function of the input field is separable (expressible as a
∞
product of two function, one in t1 and one in t 2 ). The
−τ 2 ∑ ρ n D(t1 − t 2 − nT )
n=1
magnitude of the two-photon wave function, Eq. (64) is
τ 4 ∞ ∞ plotted in Fig. 5.
+ ∑ ∑ ρ n+m D(t1 − t2 − (n − m)T )
ρ 2 n=1 m=1
= D(t1 − t 2 ) .
(62)
9
arXiv:1309.5915 [quant-ph] /Oct 5 2013
Fig. 6 (Color online) Magnitude of two-photon wave
Fig. 5 (Color online) Magnitude of two-photon wave function Φb (t1 ,t 2 ) vs. t1 and t 2 , from Eq.(64), for
function Φb (t1 ,t 2 ) vs. t1 and t 2 , from Eq.(64), for
parameter values σ = 0.2, β = 0.7 , T = 1 , and the
parameter values σ = 0.3, β = 0.3 , T = 1 , and the junction transmission varied as: (a) τ = 0.999 , (b)
junction transmission varied as: (a) τ = 0.999 , (b) τ = 0.95 , (c) τ = 0.85 , (d) τ = 0.60 .
τ = 0.95 , (c) τ = 0.85 , (d) τ = 0.60 .
5. Inclusion of Dissipative Loss
For τ = 0.999 the dominant correlation peak occurs at To include dissipative loss in the arbitrary-coupling-
(t1 ,t 2 ) = (1,1) because the light takes one full round trip in strength model, we introduce an absorbing molecular
the nearly nonreflecting cavity before emerging. For medium throughout the cavity. Then Eq. (1) reads:
τ = 0.60 the dominant correlation peak occurs at
(t1 ,t 2 ) = (0,0) because the light reflects from the junction,
( ∂ + v ∂ ) C(z,t) = α P(z,t)
t z
(66)
without delay, into the output beam. The output wave ∂t P(z,t) = −γ P(z,t) − β C(z,t) + F(z,t) ,
function is separable, as it was at the input. See Appendix
4. (This can be understood by noting that the output wave
where α , β are coupling parameters, γ is the damping
function in the frequency domain, in Eq. (53) retains its
separability if the input state is separable. And by noting rate for the molecular electric dipole polarization P(z,t) ,
that separability in frequency implies separability in time.) and F(z,t) is a quantum Langevin fluctuation operator
Figure 6 shows a case in which the input field’s wave obeying the commutator [38]
function is non-separable. The wave function retains this
non-separability at the output, as it develops ‘echoes.’
⎡⎣ F(z,t), F † (z ',t ') ⎤⎦ = 2γ δ (t − t ')δ (z − z ')
.
(67)
Integrate the equation for the dipole polarization, assuming
the molecular damping is fast, making the absorber broad
band:
t
P(z,t) = ∫ dt 'e
− γ (t−t ')
( − β C(z,t ') + F(z,t '))
−∞
(68)
≈ −(β / γ )C(z,t) + FP (z,t) ,
where the effective Langevin fluctuation operator for the
dipole polarization is
10
arXiv:1309.5915 [quant-ph] /Oct 5 2013
Calculate the commutator for the dipole fluctuation We thank Ian Walmsley, Josh Nunn, Steven van Enk,
operator: Kartik Srinivasan, and an anonymous reviewer for helpful
comments. MGR was supported by the National Science
⎡ FP (z,t), FP † (z ',t ') ⎤⎦ = exp(−γ | t − t ' |)δ (z − z ') Foundation, EPMD (ENG) and AMOP (Physics).
⎣
(70)
→ 2γ δ (t − t ')δ (z − z ') ,
Appendix 1: Commutator for cavity field
where the final step results from assuming the molecular
damping is fast. That is, in the limit we consider, one can To verify Eq. (25), write
idealize the P-fluctuations as delta-correlated. Absorber
models were introduced previously in I-O theory. [4, 5] ⎡⎣C(0 + ,t),C † (0 + ,t ') ⎤⎦ =
Here it leads to the modified cavity-field propagation ∞ ∞
equation, = τ 2 ∑ ∑ ρ m ρ n ⎡⎣ A ( t − nT ) , A† ( t '− mT ) ⎤⎦ (A1)
n=0 m=0
k=0 s=0
6. Discussion ∞
= ∑ ρ kδ (t − t '− kT ) ,
The main results of this study are: (1) Showing that k=0
the standard classical field propagation equations, Eqs. (1)-
(3), provide a proper quantum mechanical description of where we used τ 2 + ρ 2 = 1 and the general relation
input, cavity and output fields when the input coupling
strength takes on arbitrary values; (2) Deriving explicit ∞ n ∞ ∞
formulas for Green functions and commutators for the ∑ ∑ f (n + m)g(n − m) =∑ ∑ f (k + 2s)g(k) . (A3)
three fields in both space-frequency and space-time n=0 m=0 k=0 s=0
packet state.
⎡⎣C(0 + ,t),C † (0 + ,t ') ⎤⎦t≤t ' = ∑ρ −k
δ (t − t '+ kT ) . (A4)
k=−∞
The I-O theory formulated here is amenable to
inclusion of intracavity absorbing and emitting atoms. Combining the two cases gives:
Given that the equal-time field commutator agrees with the
fundamentally required one Eq. (27) [7], the effects of
atoms in the cavity may be accounted for by using the ⎡⎣C(0 + ,t),C † (0 + ,t ') ⎤⎦ =
standard minimal-coupling atom-field interaction ∞ ∞ . (A5)
Hamiltonian. The theory can also account for dynamical = ∑ρ δ (t − t '+ kT ) =
|k|
∑ρ δ (t − t '− kT )
|k|
11
arXiv:1309.5915 [quant-ph] /Oct 5 2013
B(t) = −
1 τ
A(t) + C(0 + ,t) . (A6)
= ∫ dt ∫ dt ' G(
ab (t − t )G
1
ab (t '− t ) + G
2
ab (t − t )G
2
ab (t '− t ) ψ (t,t ')
1 )
ρ ρ
(A12)
Use this and Eq. (28) to write This can be written as
1 τ
(A13)
= 2 ⎡⎣ A(t), A† (t ') ⎤⎦ − 2 ⎡⎣C(0 + ,t), A† (t ') ⎤⎦
ρ ρ
Appendix 4: Separability of two-photon state
τ τ2
− 2 ⎡⎣ A(t),C † (0 + ,t ') ⎤⎦ + 2 ⎡⎣C(0 + ,t),C † (0 + ,t ') ⎤⎦
ρ ρ Proof that if the state is separable at the input, then it is
1 τ 2 ∞ separable at the output: If Φ(t,t ') = φ1 (t)φ2 (t ') , then from
= δ (t − t ') − 2 ∑ ρ nδ (t − t '− nT ) Eq. (59),
ρ2 ρ n=0
τ2 ∞ n τ2 ⎛ ∞ ⎞
2 ∑
− ρ δ (t − t '+ nT ) + 2 ⎜ ∑ ρ |k|δ (t − t '− kT ) ⎟
Φ out (t1 , t 2 ) = ψ 1out (t1 )ψ 2out (t 2 )
(A14)
ρ n=0 ρ ⎝ k=−∞ ⎠
1 τ2 ⎛ ∞
⎞ where
= 2 δ (t − t ') − 2 ⎜ δ (t − t ') + ∑ ρ nδ (t − t '− nT )⎟
ρ ρ ⎝ n≠0 ⎠ ab (t '− t)φ (t)
ψ 1out (t) = ∫ dt 'G 1
τ ⎛
2 ∞
⎞ τ2 ∞ n
− δ (t − t ') + ∑ ρ nδ (t − t '+ nT )⎟
2 ⎜
ρ ⎝ n≠0 ⎠ = − ρφ1 (t) + ∑ ρ φ1 (t − nT ) ,
ρ n=1
(A15)
τ2 ⎛ ∞ ∞
⎞ ab (t '− t)φ (t ')
ψ 2out (t) = ∫ dt 'G
+ 2 ⎜
ρ ⎝
δ (t − t ') + ∑ ρ n
δ (t − t '− nT ) + ∑ ρ |k|δ (t − t '+ kT ) ⎟
⎠
2
n≠0 k≠0
τ2 ∞ n
⎛ 1 τ2 ⎞
= ⎜ 2 − 2 ⎟ δ (t − t ') = δ (t − t ')
= − ρφ2 (t) + ∑ ρ φ2 (t − nT )
ρ n=1
⎝ρ ρ ⎠
Plotting this form Eq. (A14) for the example shown in Fig.
(A7) 5 gives results identical to those shown there.
B(t 2 )B(t1 ) Ψ = [1] M. J. Collett and C. Gardiner, Phys. Rev. A 30, 1386
∞ ∞
,
(A8)
(1984).
= ∫ dτ ∫ dτ ' F(τ ,τ ')B(t
−∞ −∞
2 )B(t1 )B† (τ )B† (τ ') vac [2] C. W. Gardiner and C. M. Savage, A multimode
quantum theory of a degenerate parametric amplifier in a
cavity, Opt. Commun., 50, 173 (1984)
where
ab (t − τ )G
F(τ , τ ') = ∫ dt ∫ dt 'G ab (t '− τ ')ψ (t,t ') .
(A9)
[3] B. Yurke, Phys. Rev. A 29, 408 (1984).
[4] Carlos Viviescas and Gregor Hackenbroich, Field
Repeated use of commutators gives quantization for open optical cavities, Phys. Rev. A 67,
013805 (2003)
B(t 2 )B(t1 ) Ψ = ( F(t1 ,t 2 ) + F(t 2 ,t1 )) vac (A10)
[5] M. Khanbekyan, L. Knöll, D.-G. Welsch, A. A.
Semenov, and W. Vogel, QED of lossy cavities: Operator
The input two-photon function is: and quantum-state input-output relations, Phys. Rev. A.,
72, 053813 (2005).
Φ(t1 ,t 2 ) = vac A(t 2 )A(t1 ) Ψ = ψ (t1 ,t 2 ) + ψ (t 2 ,t1 )
(A11)
[6] D. F. Walls and G. J. Milburn, Quantum Optics
12
arXiv:1309.5915 [quant-ph] /Oct 5 2013
(Springer, Berlin, 2008). driven by a narrow pump pulse, Phys. Rev. A 56, 1534
(1997)
[7] S. M. Barnett, C. R. Gilson, B. Huttner and N. Imoto,
Field Commutation Relations in Optical Cavities, Phys. [24] W. Grice and I. Walmsley, Spectral information and
Rev. Lett. 77, 1739 (1996) distinguishability in type-II down-conversion with a
broadband pump, Phys. Rev. A 56, 1627 (1997).
[8] H. Risken and K. Nummedal, J. Appl. Phys. 39, 4662
(1968) [25] J. E. Sharping, J. Chen, X. Li, P. Kumar and R. S.
Windeler, Quantum-correlated twin photons from
[9] R. Graham and H. Haken, Z. Phys. 213, 420 (1968)
microstructure fiber, Opt. Express 12, 3086 (2004)
[10] E. M. Pessina, G. Bonfrate, F. Fontana, and L. A.
[26] J. G. Rarity, J. Fulconis, J. Duligall, W. J. Wadsworth
Lugiato, Experimental observation of the Risken-
and P. S. J. Russell, Photonic crystal fiber source of
Nummedal-Graham-Haken multimode laser
correlated photon pairs, Opt. Express 13, 534 (2005)
instability, Phys. Rev. A 56, 40868 (1997)
[27] J. Fan and A. Migdall, A broadband high spectral
[11] T. Voigt, M.O. Lenz, F. Mitschke, E. Roldán, G.J.
brightness fiber-based two-photon source, Opt. Express 15,
deValcárcel, Experimental investigation of Risken–
2915 (2007)
Nummedal–Graham–Haken laser instability in fiber
ring lasers, Appl. Phys. B 79, 175–183 (2004) [28] K. Garay-Palmett, H. J. McGuinness, Offir Cohen, J.
S. Lundeen, R. Rangel-Rojo, M. G. Raymer, C. J.
[12] A. E. Siegman, Lasers (University
Science
Books,
McKinstrie, S. Radic, A. B. U’Ren and I. A.Walmsley,
Sausalito,
1986))
Photon pair-state preparation with tailored spectral
[13] Optical Coherence and Quantum Optics, by Leonard properties by spontaneous four-wave mixing in photonic-
Mandel and Emil Wolf (Cambridge University Press, crystal fiber, Opt. Express, 15, 14870-14886 (2007)
(Cambridge, 1995)
[29] M. H. Rubin, D. N. Klyshko, Y. H. Shih, A. V.
[14] Rodney Loudon, Fermion and boson beam- Sergienko, Theory of two-photon entanglement in type-II
splitter statistics, Phys. Rev. A. 58, 4904 (1998). optical parametric down-conversion, Phys. Rev. A 50,
5122 (1994)
[15] R. Lang, M. O. Scully, and W. E. Lamb, Jr., Why is
the Laser Line So Narrow? A Theory of Single-Quasimode [30] W. P. Grice, A. B. U’Ren and I. A. Walmsley,
Laser Operation, Phys. Rev. A 7, 1788 (1973); M. Sargent, Eliminating frequency and space-time correlations in
M. O. Scully and W. E. Lamb, Laser Physics (Addison- multiphoton states, Phys. Rev. A 64, 063815 (2001)
Wesley, Reading, 1974)
[31] A. B. U’Ren, C. Silberhorn, K. Banaszek, I. A.
[16] J. D. Franson, Bell Inequality for Position and Time, Walmsley, R. Erdmann, W. P. Grice, and M.G. Raymer,
Phys. Rev. Lett. 62, 2205 (1989) “Generation of pure-state single-photon wave packets by
conditional preparation based on spontaneous parametric
[17] P. G. Kwiat, W. A. Vareka, C. K. Hong, H. Nathel, down conversion,” Laser Physics, 15, 146 (2005).
and R. Y. Chiao, Correlated two-photon interference in a
dual-beam Michelson interferometer, Phys. Rev. A 41, [32] T. B. Pittman, D. V. Strekalov, D. N. Klyshko, M. H.
2910 (1990). Rubin, A.V. Sergienko, and Y. H. Shih, Phys. Rev. A 53,
2804 )1996)
[18] J. D. Franson, Two-photon interferometry over large
distances, Phys. Rev. A 44, 4552–4555 (1991) [33] S. P. Walborn, A. N. de Oliveira, S. Pádua, and C. H.
Monken, Phys. Rev. Lett. 90, 143601 (2003)
[19] Z. Y. Ou, X. Y. Zou, L. J. Wang, and L. Mandel,
Observation of Nonlocal Interference in Separated Photon [34] Scully, M. O., M. S. Zubairy, Qautum Optics.
Channels, Rev. Lett. 65, 321 (1990). Cambridge University Press: Cambridge, 1997.
[20] J. D. Franson, Violations of a Simple Inequality for [35] Brian J. Smith and M. G. Raymer, Photon wave
Classical Fields, Phys. Rev. Lett. 67, 290 (1991) functions, wave-packet quantization of light, and
coherence theory, New J. Phys. 9, 414 (2007)
[21] Z. Y. Ou and L. Mandel, "Classical treatment of the
Franson two-photon correlation experiment," J. Opt. Soc. [36] B. Dayan, A. Pe'er, A. A. Friesem, Y. Silberberg, Two
Am. B 7, 2127 (1990). photon absorption and coherent control with broadband
down-converted light, Phys. Rev. Lett. 93, 023005-1-4.
[22] Nicolas Gisin and Rob Thew, Quantum (2004)
communication, Nature Photonics 1, 165 - 171 (2007)
[37] J. D. Franson, Nonclassical nature of dispersion
[23] T. E. Keller and M. H. Rubin, Theory of two-photon cancellation and nonlocal interferometry, Phys. Rev. A 80,
entanglement for spontaneous parametric down-conversion 032119 (2009)
13
arXiv:1309.5915 [quant-ph] /Oct 5 2013
14