1: \documentstyle[twocolumn,aps,epsf]{revtex}
2:
3: \input macros.tex
4:
5: \begin{document}
6:
7: \draft
8:
9: \title{Formation of Pairing Fields in Resonantly Coupled Atomic and
10: Molecular Bose-Einstein Condensates}
11:
12: \author{M. Holland, J. Park, and R. Walser}
13:
14: \address{JILA, University of Colorado and National Institute of
15: Standards and Technology, Boulder, Colorado 80309-0440}
16:
17: \date{\today}
18:
19: \wideabs{
20:
21: \maketitle
22:
23: \begin{abstract}
24: In this paper, we show that pair-correlations may play an important
25: role in the quantum statistical properties of a Bose-Einstein
26: condensed gas composed of an atomic field resonantly coupled with a
27: corresponding field of molecular dimers. Specifically,
28: pair-correlations in this system can dramatically modify the
29: coherent and incoherent transfer between the atomic and molecular
30: fields.
31: \end{abstract}
32:
33: \pacs{PACS number(s): 32.80.Pj, 03.75.Fi}
34:
35: }
36:
37: In quantum field theory, superfluidity and long-range order are
38: usually identified with a complex-valued order parameter defined as
39: the expectation value of the field operator. In the case of a dilute
40: Bose gas, the evolution of the order parameter is given in the
41: mean-field approximation by the well-known Gross-Pitaevskii equation.
42: Recently, this equation has been extensively applied to studies of the
43: zero-temperature collective behavior of metal-alkali Bose-Einstein
44: condensates confined in magnetic traps~\cite{stringari}. Quantitative
45: agreement has been found between theory and experiment on a wide range
46: of distinct phenomena (e.g.\ Refs.~\cite{collective,vortex}).
47:
48: Although identical in form to a non-linear Schr\"odinger equation, it
49: is notable that the Gross-Pitaevskii equation does not by itself
50: describe the full evolution of the quantum state. In general, a
51: complete description requires knowledge of the evolution of a
52: hierarchy of correlation functions, and not just the evolution of the
53: expectation value of the field. In particular, in low temperature
54: systems, it is well known that pairing fields associated with
55: two-particle correlations can play an important role. In a quantum
56: degenerate Fermi gas, pairing may radically alter the equilibrium
57: properties, giving rise to superfluidity in $^3$He and
58: superconductivity in electron systems. Correlations can also be
59: important in Bose systems. For example, squeezed states of
60: light---formed when photons are generated in pairs in nonlinear
61: media---have been studied extensively in the subject of quantum
62: optics~\cite{walls}.
63:
64: In this paper, we investigate the role of pair-correlations on the
65: macroscopic dynamics of a dilute Bose-Einstein condensate composed of
66: atoms and molecules which are resonantly
67: coupled~\cite{tim,verhaar,nist,drummond}. Such a coupling may be
68: generated experimentally~\cite{mit,rb85} by tuning the strength of an
69: external magnetic field in the proximity of a Feshbach resonance (see
70: Fig.~\ref{potens}). Two parameters characterize the Feshbach
71: resonance: (i) the energy mismatch~$\epsilon$ between the bound state
72: and the zero energy edge of the continuum states for the colliding
73: atom pair, and (ii) the inverse lifetime~$\kappa$ of the bound state.
74: As an alternative, photoassociation may be used to directly generate
75: the resonant coupling~\cite{photo,heinzen}. In photoassociation, a
76: two-photon Raman transition is used to couple the atomic continuum
77: states and a specific bound molecular level. In that case $\epsilon$
78: represents the detuning energy of the Raman lasers from the
79: atom-molecule transition, and $\kappa$ denotes the two-photon Rabi
80: frequency proportional to the laser intensities and to the usual
81: overlap integrals.
82:
83: \begin{figure}[b]
84: \begin{center}\
85: \epsfysize=45mm \epsfbox{potential.eps}
86: \end{center}
87: \caption{
88: Feshbach resonance. Atoms collide with relative kinetic energy near
89: zero as indicated by the dotted line. This energy is
90: quasi-degenerate with a bound state in a closed channel potential.
91: Typically, $\epsilon$ can be tuned experimentally by changing the
92: magnitude of an external magnetic field.}
93: \label{potens}
94: \end{figure}
95:
96: An effective Hamiltonian for this coupled atom-molecule system may be
97: written as
98: \begin{eqnarray}
99: \operator{H} &=& \int d^3\boldvec{x} \,
100: \fieldopdag{a}{x}\freeH{a}{x}\fieldop{a}{x} +
101: \fieldopdag{m}{x}\freeH{m}{x}\fieldop{m}{x}\nonumber\\
102: &&{}+ {\UAA\over2} \int d^3\boldvec{x} \, \fieldopdag{a}{x}
103: \fieldopdag{a}{x} \fieldop{a}{x}
104: \fieldop{a}{x}\nonumber\\
105: &&{}+ \UAM \int d^3\boldvec{x} \, \fieldopdag{a}{x}
106: \fieldopdag{m}{x} \fieldop{m}{x}
107: \fieldop{a}{x}\nonumber\\
108: &&{}+ {\UMM\over2} \int d^3\boldvec{x} \, \fieldopdag{m}{x}
109: \fieldopdag{m}{x} \fieldop{m}{x}
110: \fieldop{m}{x}\nonumber\\
111: &&{}+ {g\over2} \int d^3\boldvec{x} \, \fieldopdag{m}{x}
112: \fieldop{a}{x} \fieldop{a}{x}+\mbox{h.c.}\,,
113: \label{effective}
114: \end{eqnarray}
115: where $\fieldop{a}{x}$ and $\fieldop{m}{x}$ are bosonic field
116: operators which annihilate an atom or molecule respectively at
117: coordinate~$\boldvec{x}$. Atomic binary collisions are characterized
118: by the mean-field energy per unit density, $\UAA=4\pi\hbar^2a/m$,
119: where $m$ is the atomic mass, and $a$ is the scattering length for
120: atom-atom collisions. Equivalent definitions apply for $\UAM$ and
121: $\UMM$ corresponding to atom-molecule collisions and molecule-molecule
122: collisions. The process of inter-conversion of atom pairs into
123: molecules is characterized by $g=\sqrt{\kappa\,\UAA}$~\cite{tim}. The
124: free Hamiltonians are
125: \begin{eqnarray}
126: \freeH{a}{x}&=&\kinetic{2m}{x} + \poten{a}{x} - \mu_a\nonumber\\
127: \freeH{m}{x}&=&\kinetic{4m}{x} + \poten{m}{x} - \mu_m\,,
128: \end{eqnarray}
129: where we include $\poten{a}{x}$ and $\poten{m}{x}$ for generality to
130: allow for the possibility of external potentials. The chemical
131: potentials of the atomic and molecular fields are denoted by $\mu_a$
132: and $\mu_m$ respectively, such that $\epsilon=\mu_m-2\mu_a$.
133:
134: We solve for the evolution of the fields, assuming the principle of
135: attenuation of many-particle correlations in the dilute gas. That is,
136: we find the expectation value of the Heisenberg equations for
137: $\fieldop{a}{x}$ and $\fieldop{m}{x}$ and separate explicitly the mean
138: values of the field operators
139: \begin{eqnarray}
140: \meanop{a}{x} &=& \expect{\fieldop{a}{x}} \nonumber\\
141: \meanop{m}{x} &=& \expect{\fieldop{m}{x}}
142: \end{eqnarray}
143: from their zero-mean fluctuating components
144: \begin{eqnarray}
145: \fluctop{a}{x} &=& \fieldop{a}{x}-\meanop{a}{x} \nonumber\\
146: \fluctop{m}{x} &=& \fieldop{m}{x}-\meanop{m}{x}.
147: \end{eqnarray}
148: In order to close the dynamic equations, we expand products of three
149: of more fluctuating operators using Wick's theorem (i.e.\ we assume a
150: Gaussian reference distribution for the many-body quantum state).
151: Since we intend to determine the pairing statistics of atoms and do
152: not anticipate intrinsic pairing of molecules to play a significant
153: role, we assume a classical field for the molecular quantum state
154: (i.e.\ we specify the molecular field as an exact eigenstate of
155: $\fieldop{m}{x}$). Following this procedure, we derive a description
156: of the coupled atom-molecule system, which involves the following
157: dynamical quantities: the condensates $\meanop{a}{x}$ and
158: $\meanop{m}{x}$, and the atomic fluctuations as described by normal
159: densities $\GN{x}{y}$, and anomalous densities $\GA{x}{y}$, defined by
160: \begin{eqnarray}
161: \GN{x}{y} &=& \expect{\fluctopdag{a}{y}\fluctop{a}{x}}
162: \nonumber\\
163: \GA{x}{y} &=& \expect{\fluctop{a}{y}\fluctop{a}{x}}.
164: \end{eqnarray}
165: Following this prescription, we arrive at the following dynamical
166: equations for the condensates:
167: \begin{eqnarray}
168: i\hbar{d\meanop{a}{x}\over dt} &=&
169: \bigl(\freeH{a}{x}+\meanH{a}{x}\bigr)\meanop{a}{x} +
170: \meanO{x}\meanstar{a}{x} \nonumber\\
171: i\hbar{d\meanop{m}{x}\over dt} &=&
172: \bigl(\freeH{m}{x}+\meanH{m}{x}\bigr)\meanop{m}{x} + \meanS{x}\, ,
173: \label{meanfldeq}
174: \end{eqnarray}
175: with corresponding mean-field potentials
176: \begin{eqnarray}
177: && \meanH{a}{x} = \UAA\bigl(|\meanop{a}{x}|^2 + 2\GN{x}{x}\bigr) +
178: \UAM|\meanop{m}{x}|^2 \nonumber\\
179: && \meanH{m}{x} = \UAM\bigl(|\meanop{a}{x}|^2 + \GN{x}{x}\bigr) +
180: \UMM|\meanop{m}{x}|^2\, ,
181: \end{eqnarray}
182: and coupling elements
183: \begin{eqnarray}
184: \meanO{x} &=& \UAA\GA{x}{x} + g\meanop{m}{x} \nonumber\\
185: \meanS{x} &=& {g\over2}\bigl(\meanop{a}{x}^2+\GA{x}{x}\bigr)\,.
186: \end{eqnarray}
187: A concise representation for the evolution of the atomic fluctuations
188: is given by
189: \begin{equation}
190: i\hbar{d{\cal G}\over dt} = \Sigma{\cal G} - {\cal
191: G}\Sigma^{\dag}\,,
192: \label{densitymatrixeq}
193: \end{equation}
194: which is written in terms of the matrix
195: \begin{equation}
196: \bigG{x}{y}=\left(\begin{array}{cc}
197: \expect{\fluctopdag{a}{y}\fluctop{a}{x}} &
198: \expect{\fluctop{a}{y}\fluctop{a}{x}} \\
199: \expect{\fluctopdag{a}{y}\fluctopdag{a}{x}} &
200: \expect{\fluctop{a}{y}\fluctopdag{a}{x}}
201: \end{array}\right)\,,
202: \end{equation}
203: and Bogoliubov self-energy of general structure
204: \begin{eqnarray}
205: \bigSigma{x}{y}&=&\left(
206: \begin{array}{cc}
207: \fluctHTot{x}{y} & \fluctO{x}{y} \\
208: -\fluctO{x}{y}^* & -\fluctHTot{x}{y}^*
209: \end{array}
210: \right)\,.
211: \label{fluct}
212: \end{eqnarray}
213: The $2\times2$ elements of $\bigSigma{x}{y}$ are each diagonal in
214: $\boldvec{x}$ and $\boldvec{y}$ due to the assumption of a contact
215: potential for all scattering diagrams in Eq.~(\ref{effective}).
216: Accordingly, by defining
217: $\fluctHTot{x}{x}=\freeH{a}{x}+\fluctH{x}{x}$, the full solution is
218: \begin{eqnarray}
219: \fluctH{x}{x}&=& 2\UAA\bigl(|\meanop{a}{x}|^2 + \GN{x}{x}\bigr) +
220: \UAM|\meanop{m}{x}|^2\nonumber\\
221: \fluctO{x}{x}&=&\UAA\bigl(\meanop{a}{x}^2+\GA{x}{x}\bigr)
222: +g\meanop{m}{x}.
223: \end{eqnarray}
224:
225: This forms a universal description of the Hartree-Fock-Bogoliubov
226: theory of the coupled atom-molecule condensate system. As an example
227: of the implementation of these equations, we now apply the theory to
228: the case of a uniform gas with the assumption that we may ignore the
229: mean-field potentials (i.e.\ we set to zero $\UAA$, $\UAM$, and
230: $\UMM$), along with the external potentials $\poten{a}{x}$ and
231: $\poten{m}{x}$. Due to translational symmetry, the resulting atomic
232: and molecular condensates are spatially homogeneous, and the normal
233: and anomalous densities, $\GN{x}{y}$ and $\GA{x}{y}$, depend only on
234: the magnitude of the relative coordinate, i.e.\
235: $|\boldvec{x}-\boldvec{y}|$.
236:
237: We convert the equations to dimensionless form in the following
238: manner. We define the number density $n$ as the number of atoms plus
239: twice the number of molecules per unit volume. Then $g\sqrt{n}$
240: denotes a characteristic coupling energy of the atomic and molecular
241: mean-fields. A dimensionless time $\tau$ can therefore be defined by
242: $\tau=g\sqrt{n}\,t/\hbar$. A dimensionless coordinate
243: $r=|\boldvec{x}-\boldvec{y}|/2\zeta$ is associated with the length
244: scale $\zeta$ corresponding to the formation of molecules from atom
245: pairs. This is effectively a ``healing length'' for molecule formation
246: found by balancing the relative kinetic energy with the resonance
247: coupling energy, i.e.\ $\hbar^2/2\mu\zeta^2=g\sqrt{n}$ where $\mu=m/2$
248: is the reduced mass. Making systematic substitutions
249: $\phi_a=\meanop{a}{x}/\sqrt{n}$, $\phi_m=\meanop{m}{x}/\sqrt{n}$,
250: $G_N(r)=\GN{x}{y}/n$, $G_A(r)=\GA{x}{y}/n$, and
251: $\Delta=\epsilon/g\sqrt{n}$ gives the complete system of equations in
252: dimensionless form
253: \begin{eqnarray}
254: i{d\phi_a\over d\tau}&=&\phi_a^*\phi_m\nonumber\\
255: i{d\phi_m\over d\tau}&=&-\phi_m\Delta
256: +{1\over2}\left(\phi_a^2+G_A(r)|_{r=0}\right)
257: \nonumber\\
258: i{dG_N(r)\over d\tau}&=&\phi_mG_A^*(r)-\phi^*_mG_A(r)\nonumber\\
259: i{dG_A(r)\over d\tau}&=&-\nabla_r^2G_A(r)
260: +\phi_m\Bigl(2G_N(r)+\eta\,\delta^{(3)}(r)\Bigr)\,,
261: \label{uniform}
262: \end{eqnarray}
263: where $\delta^{(3)}(r)$ is the isotropic three-dimensional Dirac delta
264: function and we have defined $\eta=1/n\zeta^3$ as the inverse
265: diluteness parameter of the coupled atomic and molecular gas. There
266: are explicitly conserved quantities in these equations corresponding
267: to normalization (density of atoms plus twice the density of
268: molecules) i.e.\
269: \begin{equation}
270: |\phi_a|^2+G_N(r)|_{r=0}+2|\phi_m|^2=1\,,
271: \end{equation}
272: and energy density $u$ (energy due to detuning and coupling plus the
273: kinetic energy of the normal gas)
274: \begin{eqnarray}
275: u&=&-\Delta|\phi_m|^2+{\rm Re}
276: \left[\phi_m^*(\phi_a^2+G_A(r)|_{r=0})\right]\nonumber\\
277: &&{}-{1\over2}\left.\nabla_r^2G_N(r)\right|_{r=0}\,.
278: \end{eqnarray}
279:
280: Interestingly, Eqns.~(\ref{uniform}) cannot be directly integrated as
281: written. The delta function describes the {\em spontaneous} breakup of
282: molecules into atom pairs. This process is reminiscent of the decay
283: of an excited atom leading to the spontaneous emission of a photon. In
284: that case, the interaction with the unoccupied vacuum modes of the
285: radiation field gives both an energy width and an energy shift (Lamb
286: shift) to the decaying state. Analogously in the atom-molecule
287: system, there is an energy shift of the molecular state containing
288: contributions from all diagrams representing the breakup of molecules
289: into {\em virtual} atom pairs. The virtual atom pairs may form an
290: intermediate state off the energy shell. Summing over the full
291: spectrum of wave numbers for the pair gives an infinite shift to the
292: molecular level.
293:
294: This divergence is reconciled by noting that we have been inconsistent
295: in not including the self-energy of the molecular quantum state in the
296: original definition of the chemical potential $\mu_m$. We carry out
297: the renormalization in the following manner. We place an artificial
298: bound on the momenta of the atom pair by replacing the delta function
299: by a three dimensional Gaussian of standard width $\sigma_r$
300: normalized to have unit volume
301: \begin{equation}
302: \delta^{(3)}(r)\rightarrow(2\pi\sigma_r^2)^{-{3\over2}}
303: e^{-r^2/2\sigma_r^2}\,.
304: \label{deltasub}
305: \end{equation}
306: Physically, this accounts for the fact that the real potential is not
307: exactly a contact potential. The resulting energy shift is then finite
308: and we derive its value according to the following prescription. We
309: define a three dimensional Fourier transform
310: \begin{equation}
311: G_A(k)=\int
312: d^3\boldvec{r}\,G_A(r)e^{-i\smallvec{r}\cdot\smallvec{k}} \,,
313: \end{equation}
314: which for the isotropic case is simplified to
315: \begin{equation}
316: G_A(k)={4\pi\over k}\int_0^{\infty}r\,G_A(r)\sin(kr)\,dr\,.
317: \end{equation}
318: For sufficiently high wave numbers, an approximate equation may then
319: be written for $G_A(k)$ using Eq.~(\ref{uniform})
320: \begin{equation}
321: i{dG_A(k)\over d\tau}=k^2G_A(k)+\eta(k)\phi_m\,,
322: \end{equation}
323: where $\eta(k)=\eta\exp(-\sigma_r^2k^2/2)$. Taking $G_A(k)|_{\tau=0}=0$,
324: the solution of this equation is
325: \begin{eqnarray}
326: G_A(k)&=&-i\eta(k)\int_0^{\tau}\phi_m(\tau')e^{ik^2(\tau'-\tau)}
327: d\tau'\nonumber\\
328: &\approx&-{1\over k^2}\eta(k)\phi_m(\tau)\,.
329: \end{eqnarray}
330: Calculating the energy shift of the molecular level requires
331: substituting this result into the evolution equation for
332: $d\phi_m/d\tau$ in Eq.~(\ref{uniform}). Using the Fourier integral
333: $G_A(r)|_{r=0}=(2\pi^2)^{-1}\int_0^{\infty}k^2G_A(k)\,dk$, the
334: resulting shift may be incorporated into a renormalized detuning
335: $\Delta$
336: \begin{equation}
337: \Delta\rightarrow\Delta-{\eta\sqrt{2\pi}\over8\pi^2\sigma_r}\,.
338: \label{musub}
339: \end{equation}
340: The end result of this analysis is that Eqns.~(\ref{uniform}) are
341: satisfactorily renormalized by making the two substitutions,
342: Eq.~(\ref{deltasub}) and Eq.~(\ref{musub}). The evolution of any
343: relevant observable will then be independent of the choice of
344: $\sigma_r$, providing $\sigma_r$ is chosen to be sufficiently small.
345:
346: As a numerical example, we examine the $^{23}$Na Feshbach resonance at
347: 907~G~\cite{mit}. For this system $\kappa=2\pi\hbar\times44$~MHz, and
348: we consider the bound state to be on-resonance with the colliding atom
349: pairs. Other parameters are the scattering length $a=60\,a_0$, where
350: $a_0$ is the Bohr radius, and we take a density of $10^{15}\,{\rm
351: cm}^{-3}$. This gives an inverse diluteness parameter of
352: $\eta=16.5$. In the simulation in Fig.~\ref{condensate} we show the
353: evolution of the atomic and molecular condensate fields, comparing the
354: Hartree-Fock-Bogoliubov theory we have derived with the predictions of
355: mean-field theory. The initial condition is taken as 95\% of the
356: population in the atomic condensate and 5\% in the molecular
357: condensate. The persistent large scale oscillations of the population
358: of the atomic and molecular condensates as seen in a solely mean-field
359: theory dampen out when the coupling to the normal gas is included.
360: This is partly due to the fact that when molecules are formed from
361: atom pairs, the pair correlation function at that point is depleted,
362: and partly due to the possibility for spontaneous breakup of the
363: generated molecules into the normal component of the gas.
364:
365: In Fig.~\ref{gn}, we illustrate the behavior of the normal component
366: during this simulation. The density of the normal gas is
367: $G_N(r)|_{r=0}$. The temperature of the normal gas is found from the
368: spatial scale over which this correlation function decays. For
369: reference, for an equilibrium classical gas, $G_N$ is Gaussian with a
370: standard deviation given by the thermal de~Broglie wavelength.
371:
372: \begin{figure}
373: \begin{center}\
374: \epsfysize=70mm \epsfbox{condensate.eps}
375: \end{center}
376: \caption{
377: Time evolution of the atomic condensate $|\phi_a|^2$ and molecular
378: condensate $2|\phi_m|^2$ for solely mean-field theory (dashed) and
379: for the Hartree-Fock-Bogoliubov theory (solid).}
380: \label{condensate}
381: \end{figure}
382:
383: \begin{figure}
384: \begin{center}\
385: \epsfysize=60mm \epsfbox{gn.eps}
386: \end{center}
387: \caption{Evolution of the normal density.}
388: \label{gn}
389: \end{figure}
390:
391: \begin{figure}
392: \begin{center}\
393: \epsfysize=60mm \epsfbox{ga.eps}
394: \end{center}
395: \caption{Evolution of the anomalous density.}
396: \label{ga}
397: \end{figure}
398:
399: In Fig.~\ref{ga}, we show the evolution of the anomalous fluctuations
400: for this example. For a classical gas, the anomalous fluctuations are
401: explicitly zero, so clearly the quantum statistics of the normal
402: component generated here are not the usual classical thermodynamic
403: equilibrium. One of the reasons for this is that there can never be an
404: odd number of atoms in the thermal cloud since atoms are spontaneously
405: generated from molecules in pairs. This situation is similar to the
406: formation of a squeezed vacuum in optics using a laser pump to drive a
407: parametric amplifier in order to produce photon pairs. Note the value
408: of $|G_A(r)|$ near $r=0$ depends on $\sigma_r^{-1}$ and is not an
409: observable.
410:
411: In conclusion, the damping of the coherent atom-molecule oscillations
412: is accompanied by an increase in density of the normal component. The
413: behavior shown in Fig.~\ref{condensate} resolves the conceptual
414: difficulty associated with the rapid atom-molecule oscillations
415: predicted by mean-field theory---how can pairs of atoms find each
416: other rapidly enough to form molecules at the predicted rate? The
417: proper treatment of correlations shows that the oscillations are
418: slower in frequency and moreover rapidly damped.
419:
420: We would like to thank J. Cooper, E. Cornell, C. Wieman, and A.
421: Andreev for helpful discussions. This work was supported by the
422: Department of Energy.
423:
424: \begin{references}
425: \vspace*{-2pc}
426:
427: \bibitem{stringari} F. Dalfovo, S. Giorgini, L. P. Pitaevskii, and S.
428: Stringari, Rev.\ Mod.\ Phys.\ {\bf 71}, 463 (1999).
429:
430: \bibitem{collective} M. Edwards, P. A. Ruprecht, K. Burnett, R. J.
431: Dodd, and C. W. Clark, Phys.\ Rev.\ Lett.\ {\bf 77}, 1671 (1996).
432:
433: \bibitem{vortex} J. E. Williams and M. J. Holland, Nature {\bf 401},
434: 568 (1999); M. R. Matthews {\em et al.}, Phys.\ Rev.\ Lett.\ {\bf
435: 83}, 2498 (1999).
436:
437: \bibitem{walls} D. F. Walls, Nature {\bf 324}, 210 (1986).
438:
439: \bibitem{tim} E. Timmermans, P. Tommasini, R. C\^ot\'e, M. Hussein,
440: and A. Kerman, Phys.\ Rev.\ Lett.\ {\bf 83}, 2691 (1999); E.
441: Timmermans {\em et al.}, Phys.\ Rep.\ {\bf 315} 199 (1999);
442:
443: \bibitem{verhaar} F. A. Abeelen and B. J. Verhaar, Phys.\ Rev.\ Lett.,
444: {\bf 83} 1550 (1999).
445:
446: \bibitem{nist} V. A. Yurovsky, A. Ben-Reuven, P. S. Julienne, and C.
447: J. Williams, cond-mat/9903414.
448:
449: \bibitem{drummond} P. D. Drummond, K. V. Kheruntsyan, H. He, Phys.\
450: Rev.\ Lett.\ {\bf 81}, 3055 (1998).
451:
452: \bibitem{mit} S. Inouye {\em et al.}, Nature {\bf 392}, 151 (1998); J.
453: Stenger {\em et al.}, Phys.\ Rev.\ Lett. {\bf 82}, 2422 (1999).
454:
455: \bibitem{rb85} Ph. Courteille {\em et al.}. Phys.\ Rev.\ Lett.\ {\bf
456: 81}, 69 (1998); J. L. Roberts {\em et al.}, {\em ibid.} {\bf 81},
457: 5109 (1998).
458:
459: \bibitem{photo} P. S. Julienne {\em et al.}, Phys.\ Rev.\ A {\bf 58},
460: R797 (1998).
461:
462: \bibitem{heinzen} R. Wynar, R. S. Freeland, D. J. Han, C. Ryu, D. J.
463: Heinzen {\em et al.}, Science {\bf 287}, 1016 (2000).
464:
465: \end{references}
466:
467: \end{document}
468:
469: % LocalWords:
470: superfluidity Pitaevskii Schr dinger superconductivity Feshbach %
471: LocalWords: photoassociation Rabi eps Wick's Bogoliubov Eq Hartree
472: Fock Eqns % LocalWords: renormalization Na gn ga blah Wieman Andreev
473:
474: