Manipulation of optical-pulse-imprinted memory in a $\Lambda$ system
Extracted source text ยท arXiv 1605.03923
\documentclass[reprint,aps,pra,superscriptaddress,showpacs,
amsmath,amssymb]{revtex4-1}
\usepackage{mathtools}
\usepackage{graphicx}
\usepackage{booktabs,multirow}
\begin{document}
\newcommand{\pd}[2]{\frac{\partial #1}{\partial #2}}
% for partial derivatives
\newcommand{\ket}[1]{\left| #1 \right>} % for Dirac bras
\newcommand{\bra}[1]{\left< #1 \right|} % for Dirac kets
\renewcommand{\b}{\mathbf}
\title{Manipulation of optical-pulse-imprinted memory in a $\Lambda$ system}
\author{Rodrigo Guti\'{e}rrez-Cuevas}
\email{rgutier2@ur.rochester.edu}
\affiliation{Center for Coherence and Quantum Optics, University of
Rochester, Rochester, New York 14627, USA}
\affiliation{Institute of Optics, University of Rochester, Rochester, New
York 14627, USA}
\author{Joseph H. Eberly}
\affiliation{Center for Coherence and Quantum Optics, University of
Rochester, Rochester, New York 14627, USA}
\affiliation{Department of Physics and Astronomy, University of Rochester,
Rochester, New York 14627, USA}
\date{\today}
\begin{abstract}
We examine coherent memory manipulation in a $\Lambda$-type medium, using the
second order solution presented by Groves, Clader and Eberly [J. Phys.
B: At. Mol. Opt. Phys. \textbf{46}, 224005 (2013)] as a guide. The analytical
solution obtained using the Darboux transformation and a nonlinear
superposition principle describes complicated soliton-pulse dynamics which,
by an appropriate choice of parameters, can be simplified to a well-defined
sequence of pulses interacting with the medium. In this report, this solution
is reviewed and put to test by means of a series of numerical simulations,
encompassing all the parameter space and adding the effects of homogeneous
broadening due to spontaneous emission. We find that even though the
decohered results deviate from the analytical prediction they do follow a
similar trend that could be used as a guide for future experiments.
\end{abstract}
\pacs{42.50.Gy,42.50.Md,42.65.Tg,42.65.Sf}
\maketitle
\section{Introduction}
The seminal work by McCall and Hahn \cite{mccall1967self,mccall1969self}
showed the relevance of a semiclassical treatment of light-matter
interactions for strong fields with intensities far above the one-photon
limit. In this regime, disagreements with quantum electrodynamics are not
noticeable. Their discovery of self-induced transparency (SIT) showed that
new kinds of interactions beyond the well-known Beer's law were possible.
This paved the way for a number of interesting phenomena such as coherent
population trapping \cite{gray1978coherent}, electromagnetically induced
transparency (EIT)
\cite{kasapi1995electromagnetically,harris2008electromagnetically}, and slow
and fast light \cite{garrett1970propagation,boyd2002slow,milonni2004fast}, to
mention a few. There have been a number of discussions dealing with the
validity of the semiclassical theory against a full QED treatment (see for
example \cite{milonni1976semiclassical}), but no one argues about its
utility. Even today we continue to reap the benefits from this ``incomplete''
theory.
Some of these phenomena have been used to achieve light storage and
manipulation \cite{grobe1994formation}. Light can be slowed up to the point
where it stops and is stored in the medium \cite{fleischhauer2000dark}.
Then it can be regenerated as was observed in \cite{liu2001observation}.
Some other schemes have been employed such as a
combination of EIT and four-wave mixing in hot atomic vapor
\cite{camacho2009four}. The main potential application of the storage and
retrieval of light is towards quantum memories. Quantum optical systems are
desirable for this purpose as they have small decoherence and short
interaction times \cite{milonni2004fast}. Here we test the fidelity of the
complicated atom-pulse dynamics given by the second order solution
derived in \cite{groves2013jaynes} to one of the major sources of
decoherence: spontaneous emission.
The interaction of strong electromagnetic fields with atomic systems leads to
nonlinear dynamics, which makes it difficult to solve analytically, but it is
worth the effort. The SIT solution to the Maxwell-Bloch equation for a
two-level atom made clear the importance of the pulse area for the
interaction. This is defined as
\begin{equation}
\theta(x,t)=\int^t_{-\infty}\Omega(x,t)dt.
\label{area}
\end{equation}
When one takes the limit of infinite time we get the entire area of the given
pulse which follows the predictions of the area theorem, namely, the pulse
area tends to the closest even multiple of $\pi$. This results from the
smoothing properties of Doppler broadening \cite{allen2012optical} by taking
the average over the corresponding inhomogeneous distribution function of the
atomic part in the evolution equation for the field. Here we further explore
the usage of nonlinear optical interaction for light storage and memory
manipulation in a $\Lambda$-type system (see Fig.~\ref{lambda}) of ultra cold
atoms, where it is appropriate to neglect the effects of collisional and
Doppler broadening.
\begin{figure}
\includegraphics[scale=1]{fig1.eps}
\caption{\label{lambda} (Color online) Three-level
atom in a $\Lambda$ configuration
interacting with two fields in two-photon resonance via the common detuning
$\Delta$, with spontaneous emission $\Gamma_3$ from the excited state.}
\end{figure}
\section{Mathematical Model}
We consider the interaction of two fields with a $\Lambda$-system in
two-photon resonance with each field addressing a different atomic transition
as shown in Fig.~\ref{lambda}.
Each field interacts with the atomic system via the dipole moment operator
which only links levels 1 to 3 and 2 to 3;
$\hat{\vec{d}}=\vec{d}_{13}\ket{1}\bra{3}+\vec{d}_{23}\ket{2}\bra{3}
+\vec{d}_{31}\ket{3}\bra{1}+\vec{d}_{32}\ket{3}\bra{2}$.
We write the fields in carrier-envelope form:
\begin{align}
\vec{E}(x,t)=&\vec{\mathcal E}_{13}(x,t)e^{i(k_{13}x
-\omega_{13}t)}\nonumber\\
&+\vec{\mathcal E}_{23}(x,t)e^{i(k_{23}x-\omega_{23}t)}+c.c
\label{carrier}
\end{align}
where $\omega_{13}$ and $\omega_{23}$ are the field frequencies, $k_{13}$ and
$k_{23}$ the vacuum wave numbers and $\vec{\mathcal E}_{13}(x, t )$ and
$\vec{\mathcal E}_{23}(x,t)$ the slowly-varying field envelopes. We
assume that the envelopes change slowly over many cycles of the optical
frequency, thus justifying the slow-varying envelope approximation (SVEA).
Following \cite{groves2013jaynes} we refer to the 1-3 field as the signal
pulse and the 2-3 field as the control pulse. In the rotating-wave
approximation (RWA) the bare frequencies $\omega_{13}$ and $\omega_{23}$ are
eliminated in favor of $\Delta$, their common detuning, and the total
Hamiltonian takes the form:
\begin{equation}
\hat H=-\frac{\hbar}{2} \left(
\begin{array}{ccc}
0&0&\Omega_{13}^*\\
0&0&\Omega_{23}^*\\
\Omega_{13}&\Omega_{23}&-2\Delta
\end{array}
\right),
\label{hrwa}
\end{equation}
where we defined the Rabi frequencies, $\Omega_{13}(x,t)=2\vec{d}_{31}\cdot
\vec{\mathcal E}_{13}(x, t )/\hbar$ and $\Omega_{23}(x,t)=2\vec{d}_{32}\cdot
\vec{\mathcal E}_{23}(x, t )/\hbar$, and the detuning $\Delta=(E_3-
E_1)/\hbar-\omega_{13}=(E_3-E_2)/\hbar-\omega_{23}$ ($E_i$ corresponds to the
energy of level $\ket i$). The dynamics of the system are dictated by the von
Neumann equation for the density matrix of the atomic sample:
\begin{equation}
i\hbar \pd{\hat \rho}{t}=[\hat H,\hat \rho],
\label{neumann}
\end{equation}
and by Maxwell's wave equation in the SVEA for the field evolution:
\begin{subequations}
\label{meqs}
\begin{align}
\left(\pd{ }{x}+\frac{1}{c}\pd{}{t}\right)\Omega_{13}&=i\mu_{13} \rho_{31}\\
\shortintertext{and}
\left(\pd{ }{x}+\frac{1}{c}\pd{}{t}\right)\Omega_{23}&=i\mu_{23} \rho_{32}.
\end{align}
\end{subequations}
Here, we defined the atom-field coupling parameters $\mu_{j3}=N\omega_{j3}|
d_{j3}|^2/\hbar \epsilon_0c$ with $j=1,\,2$.
We consider the case of coherent short pulses for which it is justified to
neglect homogeneous relaxation processes due to the fast interaction with the
medium. This gives us a set of eight nonlinear partial differential equations
that need to be solved simultaneously. As has been shown by Park and Shin
\cite{park1998matched} and Clader and Eberly \cite{clader2007two}, for the
case of two-photon resonance and equal atom-field coupling parameters,
$\mu_{13}=\mu_{23}=\mu$, the system of equations given by Eqs.~\eqref{neumann}
and
\eqref{meqs} become integrable and thus can be solved by methods such as
inverse scattering \cite{lamb1980elements}, the B\"acklund transformation
\cite{miura1976backlund}, and the Darboux transformation
\cite{gu2006darboux,cieslinski2009algebraic}. This can be easily shown by the
introduction of the constant matrix
\begin{equation}
\hat W=i\ket{3}\bra 3 = \left(
\begin{array}{ccc}
0&0&0\\
0&0&0\\
0&0&i
\end{array}
\right),
\end{equation}
so that Eqs.~\eqref{neumann} and \eqref{meqs} in the traveling-wave
coordinates $T=t-x/c$ and $Z=x$ can be expressed as:
\begin{subequations}
\label{mb}
\begin{equation}
\label{neumann2}
i\hbar \pd{\hat \rho}{T}=[\hat H,\hat \rho]
\end{equation}
and
\begin{equation}
\pd{\hat H}{Z}=-\frac{\hbar \mu}{2}[\hat W,\hat \rho].
\end{equation}
\end{subequations}
By combining these two equations it is clear that the Lax equation,
\begin{equation}
\partial_Z \hat U-\partial_T \hat V +[\hat U,\hat V]=0,
\end{equation}
is satisfied where the Lax operators are defined as $\hat U=-(i/\hbar)\hat H
- \lambda \hat W$ and $\hat V=(i\mu/2\lambda)\hat \rho$, and $\lambda$ is a
constant known as the spectral parameter. This effectively shows that the
Maxwell-Bloch equations \eqref{mb} are integrable.
The solution obtained in \cite{groves2013jaynes} is a second-order solution
obtained from the nonlinear superposition of two first-order solutions, which
in turn were obtained by means of the Darboux transformation from the trivial
solution of a quiescent medium, $\hat \rho=\ket 1 \bra 1 $, and no fields,
$\Omega_{13}=\Omega_{23}=0$. With an appropriate choice of parameters, this
complicated solution can be reduced to a well-defined sequence of pulses
interacting with the medium (see Fig.~\ref{pulses}), transferring information
back and forth.
\begin{figure}
\includegraphics[scale=0.9]{fig2}
\caption{\label{pulses} (Color online)Pulse evolution dictated by the
second order analytical solution obtained by Groves \emph{et al.}
\cite{groves2013jaynes}. Appropriate parameters were chosen so that the
intricate analytical solution could be simplified into a well-defined
sequence of pulses, composed of two steps (separated by the ellipsis). In the
first step, at $t/\tau_a=-10$, a $2\pi$-signal pulse comes in; and as it
interacts with the medium, it gives way to a control pulse while storing its
information at $\kappa_a x_1=0$. During the storage process, which takes
place between $t/\tau_a=-7.5$ and $2.5$, the areas of the individual pulses
are no longer equal to $2\pi$ but the total pulse area [see Eq.~\eqref{atot}]
remains constant and equal to $2\pi$. At $t/\tau_a=2.5$ we can see that the
initial $2\pi$ signal pulse is gone and has been taken over by a $2\pi$
control pulse propagating away at the speed of light. The second step starts
at $t/\tau_a=12.5$ as a $2\pi$ control pulse comes in and collides with the
information imprint left by the signal pulse. During the collision, which
takes place between times $t/\tau_a=15$ and $22.5$, the initial signal pulse
is retrieved and redeposited at $\kappa_a x_2=3$, effectively pushing the
imprint farther into the medium. Here again, during the interaction the total
pulse area is conserved and equal to $2\pi$. When the re-encoding has taken
place, the control pulse recovers its original pulse area of $2\pi$ and
propagates away at the speed of light. The corresponding imprint is depicted
in Fig.~\ref{density}.}
\end{figure}
As was done in \cite{clader2007two} we will define the total pulse area as
\begin{equation}
\theta_\text{tot}=\sqrt{|\theta_{13}|^2+|\theta_{23}|^2},
\label{atot}
\end{equation}
and we have that $\theta_\text{tot}=2\pi$ for the two first-order solutions.
This concept can be extended to the second-order solution by applying it to
the pulses in each step of the sequence, as long as they are sufficiently
separated, as in Fig.~\ref{pulses} where the two steps are separated by an
ellipsis.
We will label the parameters pertaining to different steps of the sequence of
pulses by the letters $a$ and $b$.
For the first step, in the limit $t/\tau_a\ll-1$ we have a SIT-like signal
pulse propagating, driving population from the ground state $\ket 1$ into the
excited state $\ket 3$ and coherently driving it back, thus obtaining the
characteristic $2\pi$ pulse shaped as an hyperbolic secant. As the control
pulse is only zero in the limit of infinite negative time, some of the
excited population is coherently driven into the ground state $\ket 2$, thus
amplifying the seed of the control pulse and depleting the signal pulse.
During this transfer the signal pulse encodes its information into the
ground-state elements $\rho_{11}$, $\rho_{22}$, and $\rho_{12}$ of the density
matrix. This encoding we refer to as an imprint. After the storage process is
over we get a $2\pi$ control pulse propagating away at the speed of light as
it is decoupled from the medium. Both signal and control pulses have a
duration of $\tau_a$, and they are time-matched. Therefore the ratio between
their Rabi frequencies is independent of time and given by:
\begin{equation}
\frac{\Omega_{13}^a(x,t)}{\Omega_{23}^a(x,t)}=e^{-\kappa_a( x- x_1)},
\end{equation}
where the absorption coefficient is given by $\kappa_a=\mu\tau_a/2$, since we
are neglecting the effects of Doppler broadening, and $x_1$ is the location
of the imprint. This relation shows us how we should map the analytical
solution to appropriate initial conditions for the numerical computation. It
is easy to see that, by integrating the previous equation with respect to $T$
and by considering that our medium starts at $x=0$, we can get $x_1$ in terms
of the pulse areas:
\begin{equation}
\kappa_a x_1=\ln \left(\frac{\theta_{13}^a(x=0)}{\theta_{23}^a(x=0)}\right).
\label{ximp}
\end{equation}
For the second step we start with a $2\pi$-control pulse of duration $\tau_b$
decoupled from the medium. When this control pulse collides with the imprint,
the signal pulse is retrieved which, upon interaction with the medium, stores
its information in a displaced location. When the re-encoding has taken
place, the control pulse recovers its original pulse area of $2\pi$ and
propagates away at the speed of light. The end result is the displacement of
the imprint farther into the medium with a $\pi$-phase shift
for $\rho_{12}$ if $\tau_b<\tau_a$. The displacement is controlled by the
phase-lag parameter defined as:
\begin{equation}
\delta^{ab}=\kappa_a x_2 -\kappa_a x_1=\ln \left|\frac{\tau_a+\tau_b}
{\tau_a-\tau_b}\right|,
\label{dis}
\end{equation}
where $x_2$ is the new location of the imprint. Note that the addition of
Doppler broadening would affect the definition of the absorption coefficient
and thus change the group velocity of the pulses in the medium as was shown
in \cite{clader2007two}, but the storage procedure would carry through. The
results of creation and displacement of the imprint are shown in
Fig.~\ref{density}. Continuous lines show the first imprinted density matrix
elements, and the dashed lines show their displacements.
\begin{figure}[b]
\includegraphics{fig3}
\caption{\label{density} (Color online) Information
encoding into a $\Lambda$-system:
Imprint as it has been encoded in the ground state density matrix elements
before (continuous lines) and after (dashed lines) the displacement. The
imprint was generated by the pulse sequence depicted in Fig.~\ref{pulses},
the snapshots were taken at times $t/\tau_a=5$ for the initial imprint and
$t/\tau_a=25$ for the displaced one.}
\end{figure}
Having reviewed the main results of the analytical solution we now address
the question of how these ideas can be used for storage and retrieval of
optical pulses. The first step is to consider media of finite length and use
Eq.~\eqref{ximp} to determine the pulses' input areas in order to store the
signal pulse at the desired location. This location must be chosen so that
most of the imprint fits inside the medium for optimal information storage.
Signal pulse storage takes place as described by the dynamics of the first
step in the sequence of pulses, where the control pulse overtakes the signal
pulse. Now that we have the information stored inside the medium we want to
be able to retrieve it. To do this we inject a second control pulse of area
$2\pi$ and duration such that according to Eq.~\eqref{dis} the imprint is
pushed outside the medium. When the control pulse collides with the imprint
it retrieves the signal pulse that was stored in the medium, as described by
the second step in the sequence. According to the previous results, the
signal pulse travels until it reaches the location where the imprint is
supposed to be re-encoded. However, this never takes place because the
location lies outside the medium. Thus by frustrating the signal pulse
re-storage by means of the end face of the medium we are able to retrieve it.
Of course this retrieval can be done in any number of steps, displacing the
imprint closer to the end face before retrieving it.
\section{Numerical Results}
\subsection{Initial considerations}
The analytical solution is fairly restrictive because it assumes an infinite
medium and $2\pi$ pulses with asymptotic hyperbolic secant shape with
infinitely long tails. Additionally, we have neglected the effects of
homogeneous relaxation phenomena and the atom-field coupling parameters were
kept equal. Since we are considering the propagation of pulses in an ultra-cold
atomic system, we can safely omit the effects of collisional and Doppler
broadening, but spontaneous emission $\Gamma_3$ is still present and could
have a noticeable effect. We modify the von Neumann equation \eqref{neumann2}
to account for this:
\begin{subequations}
\begin{align}
\pd{\rho_{11}}{T}=&\frac{i}{2}\Omega_{13}^*\rho_{31}-\frac{i}
{2}\Omega_{13}\rho_{13}+\frac{\Gamma_3}{2}\rho_{33}\\
\pd{\rho_{22}}{T}=&\frac{i}{2}\Omega_{23}^*\rho_{32}-\frac{i}
{2}\Omega_{23}\rho_{23}+\frac{\Gamma_3}{2}\rho_{33}\\
\pd{\rho_{33}}{T}=&-\frac{i}{2}\Omega_{13}^*\rho_{31}+\frac{i}
{2}\Omega_{13}\rho_{13}-\frac{i}{2}\Omega_{23}^*\rho_{32}+\frac{i}
{2}\Omega_{23}\rho_{23} \nonumber\\
&-\Gamma_3\rho_{33}\\
\pd{\rho_{12}}{T}=&\frac{i}{2}\Omega_{13}^*\rho_{32}-\frac{i}
{2}\Omega_{23}\rho_{13}\\
\pd{\rho_{13}}{T}=&i\Delta\rho_{13}-\frac{i}
{2}\Omega_{23}^*\rho_{12}+\frac{i}
{2}\Omega_{13}^*(\rho_{33}-\rho_{11})-\frac{\Gamma_3}{2}\rho_{13}\\
\pd{\rho_{23}}{T}=&i\Delta\rho_{23}-\frac{i}
{2}\Omega_{13}^*\rho_{21}+\frac{i}
{2}\Omega_{23}^*(\rho_{33}-\rho_{22})-\frac{\Gamma_3}{2}\rho_{23}
\end{align}
\end{subequations}
and for the field:
\begin{subequations}
\begin{align}
\pd{\Omega_{13}}{Z}&=i\mu_{13} \rho_{31}\\
\pd{\Omega_{23}}{Z}&=i\mu_{23} \rho_{32}.
\end{align}
\label{meqsn}
\end{subequations}
The time and length scales are respectively defined in terms of the duration
and absorption coefficient associated with the first signal pulse. We abandon
the idealized conditions of infinitely long media and pulses by considering a
medium ten absorption lengths long (unless otherwise noted) and setting the
Rabi frequencies to zero when $|\tau_a\Omega|<10^{-5}$. For each simulation
we will consider three cases: hyperbolic-secant-shaped pulses with no decay
channels, Gaussian-shaped pulses with no decay channels, and hyperbolic-secant
shaped pulses with decay channels. This allows the effects of shape and
homogeneous broadening to be studied separately. The shape of the pulses we
will use throughout are:
\begin{subequations}
\begin{align}
\Omega=\frac{\theta}{\pi\tau}\text{sech}\left(\frac{T}{\tau}\right)\\
\shortintertext{and}
\Omega= \frac{\theta}{\sqrt{2\pi}\tau}e^{-\frac{T^2}{2\tau^2}}.
\end{align}
\end{subequations}
For the simulations, we consider a sample of $^{87}$Rb using the $D_2$ line
and consider a pulse duration such that $\tau_a \Gamma_3=0.01$, so
$\tau_a\approx 0.26\,\text{ns}$. For the atom-field coupling parameters we
have that $\mu_{23}/\mu_{13}=0.99998$ \cite{steck2001rubidium}. From these
estimates we clearly see that the approximations made in order to get the
analytical solution were justified. Nonetheless we need to study their
effects to determine whether these pulse dynamics are an experimentally
realizable scenario. For simplicity the detuning is taken to be zero.
Another thing worth noting is that we cannot choose an
arbitrary pulse duration because we need to be able to resolve the hyperfine
splitting of the ground state but not of the excited state in order to have a
$\Lambda$-system. For the case of Rb we have the condition that $0.15\,
\text{ns}<\tau_a<2\,\text{ns}$. We could have considered Cs atoms to attain
smaller values for $\tau_a \Gamma_3$ by using shorter pulses, down to $0.1\,
\text{ns}$ \cite{steck2003cesium}. Or if we want to use longer pulses, we
could use K but with the compromise of a larger value for $\tau_a
\Gamma_3$($>0.03$) \cite{tiecke2010properties}. Note that the important
quantity is $\tau_a \Gamma_3$ and not just $\Gamma_3$, because
what matters here is
how long the spontaneous decay time is with respect to the interaction time
between the pulses and the atomic system.
\subsection{Location of initial imprint}
\begin{figure*}
\includegraphics[width=.8\linewidth]{fig4}
\caption{\label{figx1} (Color online) Location of the initial imprint:
(a) as a function of
the control pulse area with a $2\pi$ signal pulse, (b) as a function of the
total pulse area keeping the ratio of the pulses' areas constant so that the
predicted location is $\kappa_a x_1=5$, and (c) as a function of the duration
of the control pulse with a $2\pi$ signal pulse and a predicted location of
$\kappa_a x_1=5$. These plots compare the location of the initial imprint for
four different cases. The black solid line is the formula given by the
analytical solution, Eq.~\eqref{ximp}. The three plots with the markers
represent the results of the numerical simulations for which we considered a
medium ten absorption lengths long, finite pulses taking $\Omega=0$ when
$|\tau_a\Omega|<10^{-5}$, and an initial $2\pi$ signal pulse. The plot with
blue circle markers represents the case of a hyperbolic-secant pulse shape
with $\tau_a \Gamma_3=0$, the one with green upside down
triangle markers adds the effects
of spontaneous emission to the previous one with $\tau_a \Gamma_3=0.01$, and
finally the plot with red square markers considers a Gaussian-shaped pulse
with $\tau_a \Gamma_3=0$.}
\end{figure*}
First, we consider how the shape and finiteness of the pulses, as well as the
effects of homogeneous relaxation, affect the location of the first imprint.
The results of the numerical simulations are summarized in Fig.~\ref{figx1}.
For the dependence on the control pulse area, Fig.~\ref{figx1}(a), we notice
that the hyperbolic secant-shaped pulse completely overlaps the curve given
by the theory. We see that the effect of spontaneous emission is to lower the
curve very slightly while maintaining the same shape. The Gaussian-shaped
pulse clearly deviates from the expected behavior, but keeps a predictable
trend and in the grand scheme has the same dependence, i.e., the location of
the imprint increases when the control pulse area decreases.
The effects of spontaneous emission on the Gaussian-shaped pulse curve are
analogous to those of the hyperbolic secant-shaped pulse, namely, to lower
the curve while keeping the same type of dependence (the data for this case
are not presented for the sake of clarity in the figures).
The analytical solution sets the total pulse area of each step of the process
equal to $2\pi$, but we are free to change that in the simulations. From an
experimental point of view, it is important to know this dependence, since it
might be difficult to control the area of the pulses with much precision.
Setting the ratio of the signal to control pulse area such that
Eq.~\eqref{ximp} predicts an imprint location of $\kappa_a x_1=5$, we vary
the total pulse area [see Fig.~\ref{figx1}(b)]. We find that for all three
cases there appears to be a linear dependence with similar positive slope,
$\Delta (\kappa_a x_1)/\Delta \theta_\text{tot}^a\approx 5/\pi$ for the
hyperbolic secant shaped- and $\Delta (\kappa_a
x_1)/\Delta \theta_\text{tot}^a \approx 6.5/\pi$ for the Gaussian-shaped
pulses. Here we only consider small variations inherent to any realistic
experimental scenario and so this behavior cannot be extrapolated to
arbitrarily large and small pulse areas. In particular, if we consider areas
smaller than $\pi$ then the pulses are not strong enough to promote the
necessary population transfer for the initial SIT propagation and then the
transfer from signal to control pulses. For pulse areas larger than $3\pi$ we
would get pulse breakup and thus a different kind of interaction.
Another restriction
from the solution is that signal and control pulses are time matched (this
might not be ideal for an experiment because one would have to change the duration
of the control pulses between consecutive steps). Relaxing this condition we
find a behavior similar to the previous case [see Fig.~\ref{figx1}(c)]: the
imprint location increases with the duration of the control pulse. Once again
all three cases seem to have similar positive slopes, $\Delta
(\kappa_a x_1)/\Delta (\tau_{c}/\tau_s)\approx 0.5$ for the hyperbolic
secants and $\Delta (\kappa_a x_1)/\Delta
(\tau_{c}/\tau_s) \approx 0.73$ for
the Gaussian. It is also worth mentioning that the shape of the imprint is
the same as the one predicted theoretically (Fig.~\ref{density}) for the two
cases where the pulses are hyperbolic secants, but for the Gaussian-shaped
pulse it is approximately $1.4$ times wider. This is a consequence of the
reshaping of the Gaussian pulse to a corresponding hyperbolic-secant-shaped
pulse with slightly different time duration.
\begin{figure*}
\includegraphics[width=.8\linewidth]{fig5}
\caption{\label{figdel} (Color online) Displacement of the
imprint as a function of: (a) the
duration of the second control pulse, (b) the control pulse area, and (c) the
location of the initial imprint. We consider the same three cases as those
described in Fig.~\ref{figx1}, represented by the same markers. For (a) and
(b) we considered an initial $2\pi$ signal pulse with the necessary pulse
area of the corresponding control pulse so that the location of the first
imprint is at $\kappa_a x_1=3$. Then for (a) we kept the area of the second
control pulse equal to $2\pi$. The dashed light gray line
marks the end of the medium
and the black solid line is the result predicted by Eq.~\eqref{dis}. For (c)
we considered a longer medium, 15 absorption lengths long, and we chose the
control pulse duration so that we had a displacement of $\delta^{ab}=3$ when
the initial imprint was at $\kappa_a x_1=3$. }
\end{figure*}
\subsection{\label{secdis} Displacement of the imprint}
Now that we know we can imprint the information of the signal pulse into the
atomic system given non-idealized conditions, we have to study the next step
of the pulse sequence: the displacement of the imprint. To do this,
we select the initial pulse area of the control pulse for each case using the
results from the previous section, so that the initial imprint is made at
$\kappa_a x_1=3$ combined with a $2\pi$ signal pulse. Then we vary the
control pulse duration and find the new position of the imprint. The results
are summarized in Fig.~\ref{figdel}(a).
As predicted by the theory, the closer $\tau_b$ is to $\tau_a$, the more the
imprint is displaced. Additionally, the new imprint is identical to the
initial one except for a $\pi$ phase shift when the control pulse duration is
smaller than the signal pulse. We also notice that the different parameters
affect the displacement in a similar way regarding the location of the
initial imprint.
The hyperbolic-secant-shaped pulse closely follows the behavior dictated by
Eq.~\eqref{dis}, while the addition of decay causes a decrease in the
displacement. However, the Gaussian-shaped pulse causes a displacement that
is typically larger than that given by the analytical solution (here again
the effects of spontaneous emission for the Gaussian-shaped pulse are
analogous to those of the hyperbolic-secant-shaped pulse). The biggest
difference from the theoretical prediction is that there is an upper limit in
how much the imprint can be moved (this is true for all three cases). This
will define how close the imprint must be to the end face so that the signal
pulse can be retrieved. Another
feature is that if we consider the case $\tau_b>\tau_a$ we get similar
results but the imprint does not present the $\pi$ phase shift. This might be
desirable if we want the retrieved signal pulse to have the same phase as the
original (as we will see in the next section), but one must be careful
because homogeneous relaxation will affect the displacement even more if the
pulses interact with the medium for a longer time.
We also explore the dependence of the displacement with respect to the
control pulse area $\theta_{23}^b$ as noted before, a parameter not
accessible from the analytical solution. As shown in Fig.~\ref{figdel}(b), the
results are quite surprising. We find that the displacement increases as we
decrease the pulse area, until it reaches a maximum (these are hidden by the
plateaus which represent that the imprint has been pushed outside the medium)
and then starts decreasing again. For all three cases, the ``optimal'' pulse
area is actually less than the $2\pi$ predicted by the theory, and not only
do we have maxima for these smaller areas but also the
displacement can be much
greater than the one predicted in Eq.~\eqref{dis}.
There is also some dependence on the first imprint location, and there are two
possible tendencies as shown in Fig.~\ref{figdel}(c). The first is an increase
in the displacement as we increase $x_1$ until it reaches a steady value.
This behavior is obtained when there are no decay channels and is due to the
finiteness of the medium which cuts off part of the information deposited by
the signal pulse, so the closer we get to the center of the medium the more
room there is for the imprint to be made. As for the Gaussian-shaped pulse,
it is affected the most because the imprint is wider. When we consider
spontaneous emission, another process takes over: The further into the medium
the first imprint is made, the less it will be displaced by the second
control pulse. This can be understood by the fact that the longer it takes to
deposit the information of the signal pulse, the longer the decay is
effective, provoking some information loss and this ultimately leads to less
displacement.
\subsection{Retrieval of the signal pulse}
One of the most important effects of the boundaries in the medium is to cut
off the pulse transfer process of the analytical solution, thus providing a
way to retrieve the initial signal pulse. To quantify the accuracy
of the retrieval process we define the retrieval efficiency as
\begin{equation}
\eta=\frac{I_{13}^{\text{out}}}{I_{13}^{\text{in}}}=\frac{\int|
\Omega_{13}^{\text{out}}|^2 dt}{\int|\Omega_{13}^{\text{in}}|^2dt}.
\label{fidel}
\end{equation}
Clearly this quantity only gives information about the output signal pulse
intensity but does not take into account any possible reshaping of the pulse.
To account for this we calculate the correlation coefficient $r$ between the
input and output signal pulses.
This is particularly important for the Gaussian-shaped pulses which are
reshaped into hyperbolic secants as they propagate through the medium. This
has already been noted in the case of atomic vapors at room temperature
\cite{clader2007two,clader2008two}. From the previous sections it should be
clear that the retrieval process can take place in any number of steps, but
for the sake of clarity we will treat only two cases: two- and three-step
retrieval. Additionally, we will only consider control pulses of areas equal
to $2\pi$ and of duration times equal to or smaller than that of the initial
signal pulse for the displacement and retrieval steps.
{
\renewcommand{\arraystretch}{1.3}
\setlength{\tabcolsep}{5pt}
\begin{table}
\caption{\label{tab} Retrieval efficiency and correlation coefficient for one-
and two-step processes.}
\begin{tabular}{lcccc}
\hline
\hline
Case & \multicolumn{2}{c}{1 step} & \multicolumn{2}{c}{2 steps}\\
& $\eta$ & $r$ & $\eta$ & $r$\\
\hline
Sech& 98\%& 1.0000& 98\%& 1.0000\\
Decay& 65\%& 0.9998& 66\%& 0.9998\\
Gaussian& 94\%&0.9940 & 94\%& 0.9941 \\
\hline
\hline
\end{tabular}
\end{table}
}
Let us first consider the one-step process. Here we want to make the imprint
somewhere in the medium such that it can be retrieved without having to move
it closer to the boundary (this limit is set by the case where we consider
homogeneous relaxation). For this we consider the necessary pulse areas for
each case so that the initial imprint is made at $\kappa_a x_1=8$, and for
the second step we consider a control pulse with the same duration as the
signal pulse, giving us the maximum displacement. For the two-step process we
will make the initial imprint at $\kappa_a x_1=5$, then tailor the time
duration of the following control pulse so that we displace the imprint by
$\delta^{ab}=3$ and finally push the imprint outside the medium by means of
another control pulse of duration equal to the original signal pulse. The
results are summarized in Table \ref{tab}. We notice that, when no decay is
present, the efficiency is high (larger than 90\%) but as soon as homogeneous
relaxation is added it drops to 65\% for one-step and 66\%
for two-step processes. As
for the correlation coefficient, it is fairly close to unity in all cases,
indicating that the shape of the pulse is mostly preserved throughout the
storage and retrieval procedure. The Gaussian-shaped pulses are the ones that
have the lowest $r$ values as could have been expected because of the
reshaping during propagation. In
any case, we are able to retrieve a good portion of the initial signal pulse
and see that this does not depend on the number of steps involved. In the
one-step case, the retrieved signal pulse has a $\pi$ phase shift with
respect to the original. The correct phase can be obtained by inverting the
initial storing control pulse or by increasing the duration of the second
control pulse as discussed in Sec.~\ref{secdis}. In the two-step process, the
signal pulse comes out with the same phase due to the $\pi$-phase shift in
the displacement step.
\section{Conclusions}
In this report we have shown that previously predicted memory manipulation by
means of idealized atomic pulse dynamics is plausible even in non-idealized
conditions. The shape of the pulse and the effects of spontaneous emission
have an impact on the quantitative results but the storage and retrieval of
the signal pulse are achieved in all cases. We have quantified the deviations
in each case and even shown some features that could add more control to
the process and lift some restrictions. The length of the medium has no
effect on the memory manipulation, but should be chosen so that most of the
information can be deposited into the medium. This may no longer be true if
we consider the reflection of the pulses at the end face.
Another aspect that we noticed throughout this work, and that has been noted
elsewhere, is the stability against the total pulse area. When considering
pulses of different area than $2\pi$, be it bigger or smaller, the pulses are
reshaped as they propagate in an effort to obtain a $2\pi$ total area. This
is completely analogous to the predictions of the area theorem
\cite{mccall1967self,mccall1969self} for a homogeneously broadened two-level
atom. Additionally, we can ask which is the appropriate extension of the
total pulse area for higher order solutions, particularly when the two first-
order solutions overlap.
We can also think of extending this analysis to more complicated pulse
sequences. This would be the same as simulating higher-order solutions of the
Maxwell-Bloch equations obtained by further applications of the nonlinear
superposition rule. This could possibly pave the way for making multiple
imprints belonging to different signal pulses and then manipulating them by
means of control pulses.
\begin{acknowledgments}
This work was supported by NSF grant PHY-1203931, and and R.G.-C. acknowledges support as a
CONACYT Fellow.
\end{acknowledgments}
%merlin.mbs apsrev4-1.bst 2010-07-25 4.21a (PWD, AO, DPC) hacked
%Control: key (0)
%Control: author (8) initials jnrlst
%Control: editor formatted (1) identically to author
%Control: production of article title (-1) disabled
%Control: page (0) single
%Control: year (1) truncated
%Control: production of eprint (0) enabled
\begin{thebibliography}{25}%
\makeatletter
\providecommand \@ifxundefined [1]{%
\@ifx{#1\undefined}
}%
\providecommand \@ifnum [1]{%
\ifnum #1\expandafter \@firstoftwo
\else \expandafter \@secondoftwo
\fi
}%
\providecommand \@ifx [1]{%
\ifx #1\expandafter \@firstoftwo
\else \expandafter \@secondoftwo
\fi
}%
\providecommand \natexlab [1]{#1}%
\providecommand \enquote [1]{``#1''}%
\providecommand \bibnamefont [1]{#1}%
\providecommand \bibfnamefont [1]{#1}%
\providecommand \citenamefont [1]{#1}%
\providecommand \href@noop [0]{\@secondoftwo}%
\providecommand \href [0]{\begingroup \@sanitize@url \@href}%
\providecommand \@href[1]{\@@startlink{#1}\@@href}%
\providecommand \@@href[1]{\endgroup#1\@@endlink}%
\providecommand \@sanitize@url [0]{\catcode `\\12\catcode `\$12\catcode
`\&12\catcode `\#12\catcode `\^12\catcode `\_12\catcode `\%12\relax}%
\providecommand \@@startlink[1]{}%
\providecommand \@@endlink[0]{}%
\providecommand \url [0]{\begingroup\@sanitize@url \@url }%
\providecommand \@url [1]{\endgroup\@href {#1}{\urlprefix }}%
\providecommand \urlprefix [0]{URL }%
\providecommand \Eprint [0]{\href }%
\providecommand \doibase [0]{http://dx.doi.org/}%
\providecommand \selectlanguage [0]{\@gobble}%
\providecommand \bibinfo [0]{\@secondoftwo}%
\providecommand \bibfield [0]{\@secondoftwo}%
\providecommand \translation [1]{[#1]}%
\providecommand \BibitemOpen [0]{}%
\providecommand \bibitemStop [0]{}%
\providecommand \bibitemNoStop [0]{.\EOS\space}%
\providecommand \EOS [0]{\spacefactor3000\relax}%
\providecommand \BibitemShut [1]{\csname bibitem#1\endcsname}%
\let\auto@bib@innerbib\@empty
%</preamble>
\bibitem [{\citenamefont {McCall}\ and\ \citenamefont
{Hahn}(1967)}]{mccall1967self}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {S.~L.}\ \bibnamefont
{McCall}}\ and\ \bibinfo {author} {\bibfnamefont {E.~L.}\ \bibnamefont
{Hahn}},\ }\href {\doibase 10.1103/PhysRevLett.18.908} {\bibfield {journal}
{\bibinfo {journal} {Phys. Rev. Lett.}\ }\textbf {\bibinfo {volume} {18}},\
\bibinfo {pages} {908} (\bibinfo {year} {1967})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {McCall}\ and\ \citenamefont
{Hahn}(1969)}]{mccall1969self}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {S.~L.}\ \bibnamefont
{McCall}}\ and\ \bibinfo {author} {\bibfnamefont {E.~L.}\ \bibnamefont
{Hahn}},\ }\href {\doibase 10.1103/PhysRev.183.457} {\bibfield {journal}
{\bibinfo {journal} {Phys. Rev.}\ }\textbf {\bibinfo {volume} {183}},\
\bibinfo {pages} {457} (\bibinfo {year} {1969})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Gray}\ \emph {et~al.}(1978)\citenamefont {Gray},
\citenamefont {Whitley},\ and\ \citenamefont {Stroud}}]{gray1978coherent}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {H.~R.}\ \bibnamefont
{Gray}}, \bibinfo {author} {\bibfnamefont {R.~M.}\ \bibnamefont {Whitley}}, \
and\ \bibinfo {author} {\bibfnamefont {C.~R.}\ \bibnamefont {Stroud}},\
}\href {\doibase 10.1364/OL.3.000218} {\bibfield {journal} {\bibinfo
{journal} {Opt. Lett.}\ }\textbf {\bibinfo {volume} {3}},\ \bibinfo {pages}
{218} (\bibinfo {year} {1978})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Kasapi}\ \emph {et~al.}(1995)\citenamefont {Kasapi},
\citenamefont {Jain}, \citenamefont {Yin},\ and\ \citenamefont
{Harris}}]{kasapi1995electromagnetically}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {A.}~\bibnamefont
{Kasapi}}, \bibinfo {author} {\bibfnamefont {M.}~\bibnamefont {Jain}},
\bibinfo {author} {\bibfnamefont {G.~Y.}\ \bibnamefont {Yin}}, \ and\
\bibinfo {author} {\bibfnamefont {S.~E.}\ \bibnamefont {Harris}},\ }\href
{\doibase 10.1103/PhysRevLett.74.2447} {\bibfield {journal} {\bibinfo
{journal} {Phys. Rev. Lett.}\ }\textbf {\bibinfo {volume} {74}},\ \bibinfo
{pages} {2447} (\bibinfo {year} {1995})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Harris}(1997)}]{harris2008electromagnetically}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {S.~E.}\ \bibnamefont
{Harris}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {Phys.
Today}\ }\textbf {\bibinfo {volume} {50}},\ \bibinfo {pages} {36} (\bibinfo
{year} {1997})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Garrett}\ and\ \citenamefont
{McCumber}(1970)}]{garrett1970propagation}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {C.~G.~B.}\
\bibnamefont {Garrett}}\ and\ \bibinfo {author} {\bibfnamefont {D.~E.}\
\bibnamefont {McCumber}},\ }\href {\doibase 10.1103/PhysRevA.1.305}
{\bibfield {journal} {\bibinfo {journal} {Phys. Rev. A}\ }\textbf {\bibinfo
{volume} {1}},\ \bibinfo {pages} {305} (\bibinfo {year} {1970})}\BibitemShut
{NoStop}%
\bibitem [{\citenamefont {Boyd}\ and\ \citenamefont
{Gauthier}(2002)}]{boyd2002slow}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {R.~W.}\ \bibnamefont
{Boyd}}\ and\ \bibinfo {author} {\bibfnamefont {D.~J.}\ \bibnamefont
{Gauthier}},\ }\href {\doibase
http://dx.doi.org/10.1016/S0079-6638(02)80030-0} {\bibfield {journal}
{\bibinfo {journal} {Prog. Opt.}\ }\textbf {\bibinfo {volume} {43}},\
\bibinfo {pages} {497} (\bibinfo {year} {2002})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Milonni}(2005)}]{milonni2004fast}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {P.~W.}\ \bibnamefont
{Milonni}},\ }\href@noop {} {\emph {\bibinfo {title} {Fast Light, Slow Light
and Left-Handed Light}}}\ (\bibinfo {publisher} {Institute of Physics},\
\bibinfo {address} {Bristol},\ \bibinfo {year} {2005})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Milonni}(1976)}]{milonni1976semiclassical}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {P.~W.}\ \bibnamefont
{Milonni}},\ }\href {\doibase http://dx.doi.org/10.1016/0370-1573(76)90037-5}
{\bibfield {journal} {\bibinfo {journal} {Phys. Rep.}\ }\textbf {\bibinfo
{volume} {25}},\ \bibinfo {pages} {1} (\bibinfo {year} {1976})}\BibitemShut
{NoStop}%
\bibitem [{\citenamefont {Grobe}\ \emph {et~al.}(1994)\citenamefont {Grobe},
\citenamefont {Hioe},\ and\ \citenamefont {Eberly}}]{grobe1994formation}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {R.}~\bibnamefont
{Grobe}}, \bibinfo {author} {\bibfnamefont {F.~T.}\ \bibnamefont {Hioe}}, \
and\ \bibinfo {author} {\bibfnamefont {J.~H.}\ \bibnamefont {Eberly}},\
}\href {\doibase 10.1103/PhysRevLett.73.3183} {\bibfield {journal} {\bibinfo
{journal} {Phys. Rev. Lett.}\ }\textbf {\bibinfo {volume} {73}},\ \bibinfo
{pages} {3183} (\bibinfo {year} {1994})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Fleischhauer}\ and\ \citenamefont
{Lukin}(2000)}]{fleischhauer2000dark}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {M.}~\bibnamefont
{Fleischhauer}}\ and\ \bibinfo {author} {\bibfnamefont {M.~D.}\ \bibnamefont
{Lukin}},\ }\href {\doibase 10.1103/PhysRevLett.84.5094} {\bibfield
{journal} {\bibinfo {journal} {Phys. Rev. Lett.}\ }\textbf {\bibinfo
{volume} {84}},\ \bibinfo {pages} {5094} (\bibinfo {year}
{2000})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Liu}\ \emph {et~al.}(2001)\citenamefont {Liu},
\citenamefont {Dutton}, \citenamefont {Behroozi},\ and\ \citenamefont
{Hau}}]{liu2001observation}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {C.}~\bibnamefont
{Liu}}, \bibinfo {author} {\bibfnamefont {Z.}~\bibnamefont {Dutton}},
\bibinfo {author} {\bibfnamefont {C.~H.}\ \bibnamefont {Behroozi}}, \ and\
\bibinfo {author} {\bibfnamefont {L.~V.}\ \bibnamefont {Hau}},\ }\href@noop
{} {\bibfield {journal} {\bibinfo {journal} {Nature (London)}\ }\textbf
{\bibinfo {volume} {409}},\ \bibinfo {pages} {490} (\bibinfo {year}
{2001})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Camacho}\ \emph {et~al.}(2009)\citenamefont
{Camacho}, \citenamefont {Vudyasetu},\ and\ \citenamefont
{Howell}}]{camacho2009four}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {R.~M.}\ \bibnamefont
{Camacho}}, \bibinfo {author} {\bibfnamefont {P.~K.}\ \bibnamefont
{Vudyasetu}}, \ and\ \bibinfo {author} {\bibfnamefont {J.~C.}\ \bibnamefont
{Howell}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {Nature
Photonics}\ }\textbf {\bibinfo {volume} {3}},\ \bibinfo {pages} {103}
(\bibinfo {year} {2009})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Groves}\ \emph {et~al.}(2013)\citenamefont {Groves},
\citenamefont {Clader},\ and\ \citenamefont {Eberly}}]{groves2013jaynes}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {E.}~\bibnamefont
{Groves}}, \bibinfo {author} {\bibfnamefont {B.~D.}\ \bibnamefont {Clader}},
\ and\ \bibinfo {author} {\bibfnamefont {J.~H.}\ \bibnamefont {Eberly}},\
}\href@noop {} {\bibfield {journal} {\bibinfo {journal} {J. Phys. B: At.
Mol. Opt. Phys.}\ }\textbf {\bibinfo {volume} {46}},\ \bibinfo {pages}
{224005} (\bibinfo {year} {2013})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Allen}\ and\ \citenamefont
{Eberly}(1987)}]{allen2012optical}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {L.}~\bibnamefont
{Allen}}\ and\ \bibinfo {author} {\bibfnamefont {J.~H.}\ \bibnamefont
{Eberly}},\ }\href@noop {} {\emph {\bibinfo {title} {Optical Resonance and
Two-Level Atoms}}}\ (\bibinfo {publisher} {Dover},\ \bibinfo {address} {New
York},\ \bibinfo {year} {1987})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Park}\ and\ \citenamefont
{Shin}(1998)}]{park1998matched}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {{Q.~Han}}\ \bibnamefont
{Park}}\ and\ \bibinfo {author} {\bibfnamefont {H.~J.}\ \bibnamefont
{Shin}},\ }\href {\doibase 10.1103/PhysRevA.57.4643} {\bibfield {journal}
{\bibinfo {journal} {Phys. Rev. A}\ }\textbf {\bibinfo {volume} {57}},\
\bibinfo {pages} {4643} (\bibinfo {year} {1998})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Clader}\ and\ \citenamefont
{Eberly}(2007)}]{clader2007two}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {B.~D.}\ \bibnamefont
{Clader}}\ and\ \bibinfo {author} {\bibfnamefont {J.~H.}\ \bibnamefont
{Eberly}},\ }\href {\doibase 10.1103/PhysRevA.76.053812} {\bibfield
{journal} {\bibinfo {journal} {Phys. Rev. A}\ }\textbf {\bibinfo {volume}
{76}},\ \bibinfo {pages} {053812} (\bibinfo {year} {2007})}\BibitemShut
{NoStop}%
\bibitem [{\citenamefont {Lamb}(1980)}]{lamb1980elements}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {G.~L.}\ \bibnamefont
{Lamb}},\ }\href@noop {} {\emph {\bibinfo {title} {Elements of Soliton
Theory}}}\ (\bibinfo {publisher} {Wiley},\ \bibinfo {address} {New York},\
\bibinfo {year} {1980})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Miura}(1976)}]{miura1976backlund}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {R.~M.}\ \bibnamefont
{Miura}},\ }\href@noop {} {\emph {\bibinfo {title} {Backlund
Transformations}}}\ (\bibinfo {publisher} {Springer-Verlag},\ \bibinfo
{address} {Berlin},\ \bibinfo {year} {1976})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Gu}\ \emph {et~al.}(2005)\citenamefont {Gu},
\citenamefont {Hu},\ and\ \citenamefont {Zhou}}]{gu2006darboux}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {C.}~\bibnamefont
{Gu}}, \bibinfo {author} {\bibfnamefont {H.}~\bibnamefont {Hu}}, \ and\
\bibinfo {author} {\bibfnamefont {Z.}~\bibnamefont {Zhou}},\ }\href@noop {}
{\emph {\bibinfo {title} {Darboux Transformations in Integrable Systems}}}\
(\bibinfo {publisher} {Springer},\ \bibinfo {address} {Dordrecht},\ \bibinfo
{year} {2005})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Cie{\'s}li{\'n}ski}(2009)}]{cieslinski2009algebraic}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {J.~L.}\ \bibnamefont
{Cie{\'s}li{\'n}ski}},\ }\href@noop {} {\bibfield {journal} {\bibinfo
{journal} {J. Phys. A: Math. Theor.}\ }\textbf {\bibinfo {volume} {42}},\
\bibinfo {pages} {404003} (\bibinfo {year} {2009})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Steck}(2010{\natexlab{a}})}]{steck2001rubidium}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {D.~A.}\ \bibnamefont
{Steck}},\ }\href@noop {} {\enquote {\bibinfo {title} {Rubidium 87 {D} {L}ine
{D}ata},}\ }\bibinfo {howpublished} {available online at
\url{http://steck.us/alkalidata}} (\bibinfo {year} {revision 2.1.4, 23
December 2010}{\natexlab{a}})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Steck}(2010{\natexlab{b}})}]{steck2003cesium}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {D.~A.}\ \bibnamefont
{Steck}},\ }\href@noop {} {\enquote {\bibinfo {title} {Cesium {D} {L}ine
{D}ata},}\ }\bibinfo {howpublished} {available online at
\url{http://steck.us/alkalidata}} (\bibinfo {year} {revision 2.1.4, 23
December 2010}{\natexlab{b}})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Tiecke}(2011)}]{tiecke2010properties}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {T.~G.}\ \bibnamefont
{Tiecke}},\ }\href
{http://www.tobiastiecke.nl/archive/PotassiumProperties.pdf} {\enquote
{\bibinfo {title} {Properties of {P}otassium},}\ }\bibinfo {howpublished}
{available online at
\url{http://www.tobiastiecke.nl/archive/PotassiumProperties.pdf}} (\bibinfo
{year} {v1.02, May 2011})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Clader}\ and\ \citenamefont
{Eberly}(2008)}]{clader2008two}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {B.~D.}\ \bibnamefont
{Clader}}\ and\ \bibinfo {author} {\bibfnamefont {J.~H.}\ \bibnamefont
{Eberly}},\ }\href {\doibase 10.1103/PhysRevA.78.033803} {\bibfield
{journal} {\bibinfo {journal} {Phys. Rev. A}\ }\textbf {\bibinfo {volume}
{78}},\ \bibinfo {pages} {033803} (\bibinfo {year} {2008})}\BibitemShut
{NoStop}%
\end{thebibliography}%
\end{document}