betaFeedback

WorksPhysics and Astronomy

Inelastic confinement-induced resonances in multiwell potentials

Abstract

Recently, resonance features associated with multiwell effects have been reported in a fascinating experiment conducted with an ultracold-atom quantum simulator [Y. K. Lee, H. Lin, and W. Ketterle, Phys. Rev. Lett. 131, 213001 (2023)]. The results were explained as a consequence of superexchange, which determines the spin dynamics in a singly-occupied Mott insulator, by allowing for the virtual population of two particles on the same site. In this work, a complementary interpretation of the resonance features is provided based on inelastic confinement-induced resonances, manifested as avoided crossings in the two-particle spectrum, showing excellent agreement with the experimental results. The comparison of the results of the single- and double-well descriptions of the optical trapping potential confirms the multiwell origin of the experimental features. However, the double-well model is already shown to be sufficient for providing an explanation of the apparent symmetry violation that occurs within single-well models. Furthermore, certain limitations of the harmonic approximation are discussed.

Introduction

The study of ultracold atomic gases has noticeably extended over the last years [2, 3]. The reason is twofold, and relies on the excellent degree of control that can be achieved over the interactions and the confinement [4].

On the one hand, the atom-atom interaction can be tuned almost at will thanks to the magnetic Feshbach resonances (FRs) [5, 6, 7, 8]. These resonances occur when the energy of a molecular bound state is shifted under an external magnetic field thanks to the hyperfine structure of the atoms, becoming resonant with the energy of two colliding atoms in free space [9, 10]. A similar coupling can be also optically induced [11, 12, 13, 14]. Remarkably, within the ultracold regime, s-wave scattering dominates (except for indistinguishable Fermions). Thus the atom-atom interaction can be characterized by a single parameter, the s-wave scattering length asca_{\mathrm{sc}} [2, 15], which diverges at FRs. As a consequence, the effective interaction between two particles can be tuned from strongly attractive, if asca_{\mathrm{sc}} presents large negative values, to repulsive, if asca_{\mathrm{sc}} has large positive values, or vice versa.

On the other hand, the confinement of the atoms has also added an exquisite degree of control [16]. Atoms can be distributed at arbitrary positions in space with optical lattices created with counter propagating laser fields [17, 18, 19, 20]. Furthermore, when the laser intensities in one or two dimensions are substantially lowered, a three-dimensional (3D) lattice [21, 22, 23, 24] effectively reduces to a quasi-one dimensional (Q1D) elongated trap [1, 25, 26, 27, 28, 29, 30, 31] or to a quasi-two dimensional (Q2D) planar trap [31, 32, 33], where the motion of the atoms is effectively restricted to take place over tubes or planes, respectively. Indeed, optical lattices have played a central role in the creation of quantum-gas microscopes [4, 34, 35], where the positions of single atoms are accurately determined and monitored. These extraordinary tools are also complemented with optical tweezers, also known as microtraps [36, 37], which are very tiny traps, in which few atoms are strongly confined. The combination of several optical tweezers has also permitted new methods to synthesize ultracold molecules [38, 39, 40, 41] with unforeseen precision. Likewise, the so-called tweezer arrays [42] have opened new routes to study many-particle correlated matter, such as quantum many-body scars [43]. Moreover, the number of atoms can be also arbitrarily controlled in experiments all over the world, ranging from solely a few particles [38, 40, 44, 45] to large atomic samples constituted by thousands of atoms [24, 46]. Likewise, different atomic species [40, 41, 47] or atoms with different internal states [1, 38] can be used with high flexibility. In summary, the high degree of control that is nowadays achieved in ultracold atoms has transformed them into one of the most promising platforms for quantum-simulation and even quantum-computation applications [17, 48, 49, 50, 51]. The combination of tunable interactions and external optical confinements is, nevertheless, far from trivial as it changes the system behavior dramatically [5, 21, 31, 46, 47, 52, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64].

In a seminal work, the scattering process under external confinement was analyzed by Olshanii for two interacting particles in a harmonic Q1D wave guide [65], resulting in the introduction of an effective one-dimensional interaction parameter that is a unique function of the s-wave scattering length. In a subsequent work, Bergeman et al. [66] discussed the so-called (elastic) confinement-induced resonances (CIRs) which occur if the ratio of the scattering length and the confinement strength acquires a specific value for which the effective 1D interaction parameter diverges. Similarly, also under Q2D confinement (elastic) CIRs were predicted [56] and thereafter experimentally observed [32]. However, the theory of elastic CIRs could not explain the Q1D results reported by Haller et al. in a groundbreaking experiment where the resonant loss features of an ultracold atomic sample split under anisotropic Q1D confinement. Nowadays, the results of the experiment in Ref. [46] are recognized to be a consequence of a different type of resonances [64]: the inelastic confinement-induced resonances (ICIRs). ICIRs are ubiquitous in traps as they are caused by the coupling between the relative-motion (RM) and the center-of-mass (CM) degrees of freedom by, e. g., possible anharmonicities of the external confinement [64, 67, 68, 69], which explains why they were named anharmonicity-induced resonances by Kestner and Duan [61, 62].

Notably, the CM-RM coupling can also arise under harmonic confinement [70] if different atomic species are considered [71], or if the same species in different hyperfine states is trapped with different frequencies. Moreover, two-particle systems under strictly harmonic confinement can also exhibit trap-induced resonances [72, 73, 74] if the traps are not equally centered in space. Also in this type of ICIRs, resonances appear as avoided crossings in the two-particle spectrum when a molecular bound state becomes resonant with a trap eigenstate, though in this case the resonance is controlled by the relative separation of the trapping potentials.

The existence of ICIRs is nowadays accepted within the ultracold community, as they have been observed in systems of different shape and dimension [24, 46], including different atomic species [75]. In fact, ICIRs have also enabled the coherent molecule formation of two Fermionic atoms [38], a process that cannot be induced by elastic CIRs. Furthermore, their occurrence in solid-state devices has been also predicted [76]. In this case, the interaction is not of van-der-Waals kind but Coulombic. Their existence under 3D confinement, as recently demonstrated by Capecchi et al. [24], clearly distinguishes them from elastic CIRs that can only occur in Q1D or Q2D systems.

In this work, we revisit the experiment recently performed by Lee et al. [1]. By considering a simple theoretical model comprising two ultracold atoms that interact via a realistic ab-initio potential, we provide an alternative explanation of the experimental results that does not rely on a solid-state inspired interpretation but on a purely few-body molecular perspective. The agreement between the loss features detected in the experiment, which were associated in Ref. [1] with lattice-induced resonances [63], and the predictions rendered by our numerical simulations is very good. Remarkably, by considering a double-well (DW) potential the apparent symmetry violation that is discussed in the context of the adopted harmonic model in the Ref. [1] is also lifted. In addition, we provide a simple and transparent explanation for the experimental observation of ICIRs under Q1D confinement and its non-detection under isotropic 3D confinement. Noticeably, the reasonable performance of the single-well (SW) description of the system is also discussed, and its accuracy is detailed in comparison with the DW results. Both, harmonic and anharmonic potentials are considered within the SW descriptions of the optical trap. In this way, we show that the so called lattice-induced resonances that are discussed as a consequence of tunneling between neighboring sites have, indeed, the same origin as the ICIRs [64]: the RM-CM coupling.

The outline of this work is as follows. First, Sec. 2 presents the studied physical system, outlining the key features of the experimental setup, the theoretical models employed, and the numerical methods implemented. Second, Sec. 3 contains the results of our study along with the corresponding discussion. Finally, we conclude this paper in § 4 with a brief summary and outlook.

System and methods

This section describes the system and the methods considered for its theoretical description. Section 2.1 briefly summarizes the main characteristics and the current interpretation of the experiment reported in the Ref. [1]. § 2.2 is devoted to the two-particle models developed for this work. Finally, § 2.3 briefly summarizes the numerical scheme that has been used.

Summary of the experiment in the Ref. [1]

In the experiment presented in the Ref. [1], a Mott insulator [77, 78] of 7Li{}^{7}\mathrm{Li} ultracold atoms trapped in an optical lattice with single-site occupancy is first created, as sketched in Figure 1(a). The atoms are in the two lowest hyperfine states (labeled as aa and bb), mimicking a two-component Bose-Hubbard model [3]. Under these conditions, three-body processes are very rare because of single-site occupancy and the short-range behavior of the interparticle van-der-Waals interaction.

Ultracold atoms trapped in an optical lattice and the considered models

Figure 1. Ultracold atoms trapped in an optical lattice and the considered models. (a) Sketch of a Mott insulator state with a single-site occupancy trapped in an optical lattice [cf. Eqs. (2) and (5)]. (b) Single-well potential along the xx direction for a large trap depth of Vx=35ERV_x = 35 E_R, comparing the harmonic approximation of Eq. (3) (dashed brown) with the sextic potential of Eq. (4) (dash-dotted cyan). (c) Same as (b) for the yy direction. (d) Same as (b) along the zz direction, along with the double-well description given by Eq. (6) (solid olive). (e) Same as (d) for a small trap depth of Vz=10.7ERV_z = 10.7 E_R.

In the first experimental setting, the atoms are confined in an isotropic 3D optical lattice with potential depths equal to Vx=Vy=Vz=35ERV_x = V_y = V_z = 35 E_R, with ER=h2/(2mλ2)≈25E_R = h^2/(2m\lambda^2) \approx25 kHz the recoil energy, and λ=1064\lambda= 1064 nm the wave length; these parameters render trapping frequencies equal to ωx=ωy=ωz=2π×297.2\omega_x = \omega_y = \omega_z = 2\pi\times297.2 kHz.

In the second setting, by lowering the laser intensity in the zz direction, the optical lattice becomes strongly elongated, creating a Q1D confinement with potential depths equal to Vx=Vy=35ERV_x = V_y = 35E_R, and Vz=10.7ERV_z = 10.7 E_R. In this case, the corresponding trapping frequencies are related as ωx=ωy=1.8ωz\omega_x = \omega_y = 1.8\omega_z.

The system symmetry enables the realization of the spin-1/2 XXZ Heisenberg model [3], where the effective spin-spin interactions are a combination of superexchange processes and off-site interactions. In order to characterize these interactions, the researchers in the Ref. [1] determined the spin-anisotropy parameter by exploiting the enhanced lifetime of spin helix states [79, 80] at a specific wave vector. The spin-anisotropy parameter is influenced by interactions among all three spin-state combinations (aaaa, abab, and bbbb). This parameter was experimentally measured by tuning the bbbb interaction around one narrow and one wide magnetic FR [5, 6, 7, 8] associated with the bbbb atoms, which, remarkably, does not affect the aaaa and abab spin-spin interactions.

Within the wide FR, three dispersive features are observed at 892.249, 893.182, and 895.537 G. An additional dispersive feature more difficult to probe is found at 845.459 G. The corresponding scattering lengths are equal to 4632 a0a_0, 2287 a0a_0, −2662a0-2662 a_0, and −2593a0-2593 a_0 [81], respectively, a0a_0 being the Bohr radius. No dispersive feature is observed in the 3D experimental setup. As reported in the Ref. [1], none of those results can be associated with already known FRs. Conversely, the comparison with the results predicted by a simplified model of two particles confined in an (isotropic or anisotropic) harmonic trap and interacting through a contact potential, the authors relate them to lattice-induced resonances [63]. Despite using a harmonic model, the authors attribute the observed features to multiwell, i. e., to lattice effects, by searching for crossings between the molecular states formed under harmonic confinement and the zero-energy point related to the initial Mott-insulator state. Within this interpretation, the key point is a resonant directional tunneling of one atom from one well to the neighboring site in order to create a bound molecule, in a process that breaks the parity symmetry of the system. From the energy point of view, the bond formation is possible because in the final molecular states the (negative) binding energy is balanced by the (positive) CM energy of the created molecule, which enables the crossing with the initial Mott-insulator state of two atoms at neighbor sites. Although the importance of the CM-RM coupling is emphasized, the authors in [1] conclude that the resonances cannot be IGRs. Notably, the authors in Ref. [1] emphasize the simplicity of their harmonic-trap model, as in lattices many of the energy level crossings become avoided and the bands attain finite widths.

Theoretical models

We model the system of interest using the two-particle Hamiltonian

H^(r1,r2)=∑i=12[−ℏ22mP^i2+V^opt(ri)]+U^(r1,r2),(1)\hat{H}(\mathbf{r}_1,\mathbf{r}_2) = \sum_{i=1}^{2} \left[-\frac{\hbar^2}{2m}\hat{P}_i^2 + \hat{V}_{\mathrm{opt}}(\mathbf{r}_i)\right] + \hat{U}(\mathbf{r}_1,\mathbf{r}_2), \tag*{(1)}

where the sum accounts for the single-particle terms related to the kinetic-energy operator and to the potential-energy operator exerted by the optical lattice for each of the particles, while the last term is associated with the interparticle interaction. In the present study, this interaction is mostly described using a realistic ab-initio Born-Oppenheimer curve for a Li-Li dimer. As detailed below, although the interatomic potentials are very short-ranged, the particles located at different trap wells can still interact due to the small overlaps between the corresponding SW wave functions Figure 1(a), i. e., the non-perfect localization of the Wannier functions in most realistic lattice systems. This finite overlap facilitates the control of atom-atom interactions via FRs of atoms at different lattice sites, as previously observed, e. g., in the Mott-to-superfluid transition [82, 83].

Figure 1 also sketches some of the models for the trapping potentials that we have considered when modeling the experiment. Figures 1(b) and 1(c) show the SW description obtained by Taylor expanding a sin⁡2\sin^2-like optical potential around the origin as

Vopt(qi)=Vqsin⁡2(kqi)(2)V_{\mathrm{opt}}(q_i) = V_q\sin^2(kq_i) \tag*{(2)}
≈Vq(kqi)2+O((kqi)4),(3)\approx V_q(kq_i)^2 + \mathcal{O}\left((kq_i)^4\right), \tag*{(3)}
≈Vq((kqi)2−(kqi)43+2(kqi)645)+O((kqi)8),(4)\approx V_q\left((kq_i)^2-\frac{(kq_i)^4}{3}+\frac{2(kq_i)^6}{45}\right)+\mathcal{O}\left((kq_i)^8\right), \tag*{(4)}

with qi=xi,yi,q_i=x_i,y_i, and ziz_i the position coordinates of particle i=1,2i=1,2, k=2π/λk=2\pi/\lambda the wave number, and VqV_q the potential depth (see discussion below). As can be seen, the expansion of Eq. (2) up to order 2 yields the harmonic approximation provided by Eq. (3), while its expansion up to order 6 renders the sextic potential provided by Eq. (4).

Furthermore, in order to provide a more realistic description of the optical lattice, which also facilitates the study of off-site phenomena [84, 85], we also consider in the zz direction a DW potential. This model, shown in Figure 1(d) and Figure 1(e), is obtained by Taylor expanding a cos⁡2\cos^2-like optical potential around the origin up to order 12 as

Vopt(zi)=Vzcos⁡2(kzi)(5)V_{\mathrm{opt}}(z_i) = V_z\cos^2(kz_i) \tag*{(5)}
≈  Vz(1−(kzi)2+(kzi)43−2(kzi)645+(kzi)8315−2(kzi)1014175+2(kzi)12467775)+O((kzi)14).(6)\begin{aligned} \approx\;& V_z\left(1-(kz_i)^2+\frac{(kz_i)^4}{3}-\frac{2(kz_i)^6}{45}+\frac{(kz_i)^8}{315}-\frac{2(kz_i)^{10}}{14175}+\frac{2(kz_i)^{12}}{467775}\right)\\ &+\mathcal{O}\left((kz_i)^{14}\right). \tag*{(6)} \end{aligned}

To conclude this section, let us summarize the models that have been considered here:

  1. Single-well model of order 2 (SWM2). It constitutes a perfectly harmonic (SW) approximation to the trapping potential in all spatial directions [cf. Eq. (3)].

  1. Single-well model of order 6 (SWM6). It is an anharmonic sextic description of the lattice in all directions, which also provides a more realistic SW description of a lattice [cf. Eq. (4)].

  1. Double-well model (DWM). It combines a DW potential in the zz direction [cf. Eq. (6)] and two sextic potentials in the transversal directions [cf. Eq. (4)].

Notice that, for simplicity, we refer to model (iii) as DWM, despite its transversal confinement being described by two sextic (and not harmonic) potentials. Additional calculations were also performed using a model combining double wells in two of the directions, and the sextic potential in the remaining one, as well as a triple-well model (by Taylor expanding Eq. (2) up to order 22, see [86], in one direction combined with two sextic expansions in the remaining directions). For the remainder of this work, we omit these additional multiwell models since their predictions are already captured by the model (iii) (DWM), which is carefully discussed, along with its relation to models (i) and (ii).

Besides the different models for the optical lattice, different geometries (Q1D and 3D) and interaction descriptions (abab-initio curves and δ\delta regularized pseudopotential [87]) are considered. All the parameters in our simulations (potential depths, wave length, atomic properties, etc.) correspond to the values as reported in the experiment [1], i. e., none of them have been fitted.

Numerical approach

The study of Eq. (1) is very challenging, even in the stationary case, where one is essentially interested in the computation of the system’s eigenenergies and eigenstates. The reason is twofold. First, if the full dimensionality of the lattice (3D) is considered, six-dimensional interparticle-interaction integrals must be solved. Second, the calculation of the matrix elements is also very demanding due to the dramatically different scales of the short-range interatomic potential, of the order of a few atomic units, and the trapping potential, which typically exceeds hundreds or even thousands of atomic units.

These challenges are mostly circumvented by changing from absolute to CM and RM coordinates. Then, the interaction potential solely depends on the RM coordinate r=r1−r2\mathbf{r} = \mathbf{r}_1 - \mathbf{r}_2. However, in RM and CM coordinates, the optical-trap potential depends, in general, on both coordinates, the CM one defined as R=μ1r1+μ2r2\mathbf{R} = \mu_1\mathbf{r}_1+\mu_2\mathbf{r}_2, with μi=mi/(m1+m2)\mu_i = m_i/(m_1+m_2) being the reduced mass of particle ii. Nevertheless, if the trapping potential is expanded as, e. g., in Eqs. (3), (4), and (6), or it is fitted using other polynomial sets, the total potential can be split as

V^opt(r1)+V^opt(r2)+U^(r1,r2)=V^CM(R)+V^RM(r)+W^CM−RM(R,r),(7)\hat{V}_{\mathrm{opt}}(\mathbf{r}_1)+\hat{V}_{\mathrm{opt}}(\mathbf{r}_2)+\hat{U}(\mathbf{r}_1,\mathbf{r}_2)=\hat{V}_{\mathrm{CM}}(\mathbf{R})+\hat{V}_{\mathrm{RM}}(\mathbf{r})+\hat{W}_{\mathrm{CM-RM}}(\mathbf{R},\mathbf{r}), \tag*{(7)}

where V^CM(R)\hat{V}_{\mathrm{CM}}(\mathbf{R}) and V^RM(r)\hat{V}_{\mathrm{RM}}(\mathbf{r}) are the CM and RM potential-energy operators, and W^CM−RM(R,r)\hat{W}_{\mathrm{CM-RM}}(\mathbf{R},\mathbf{r}) is a term that accounts for the CM-RM coupling due to, e. g., anharmonicities of the potential or a possible mass imbalance. Then, Eq. (1) can be readily recast as

H^(r,R)=h^CM(R)+h^RM(r)+W^CM−RM(R,r),(8)\hat{H}(\mathbf{r},\mathbf{R})=\hat{h}_{\mathrm{CM}}(\mathbf{R})+\hat{h}_{\mathrm{RM}}(\mathbf{r})+\hat{W}_{\mathrm{CM-RM}}(\mathbf{R},\mathbf{r}), \tag*{(8)}

where

h^CM(R)=−ℏ22MP^CM2+V^CM(R)(9)\hat{h}_{\mathrm{CM}}(\mathbf{R})=-\frac{\hbar^2}{2M}\hat{\mathbf{P}}_{\mathrm{CM}}^2+\hat{V}_{\mathrm{CM}}(\mathbf{R}) \tag*{(9)}

is the CM Hamiltonian, and

h^rm(r)=−ℏ22μp^RM2+V^RM(r)(10)\hat{h}_{\mathrm{rm}}(\mathbf{r})=-\frac{\hbar^2}{2\mu}\hat{\mathbf{p}}_{\mathrm{RM}}^2+\hat{V}_{\mathrm{RM}}(\mathbf{r}) \tag*{(10)}

is the RM Hamiltonian, with the reduced mass μ=m1m2/(m1+m2)\mu= m_1m_2/(m_1 + m_2). For systems of equal masses under perfectly harmonic confinement, there is no coupling between the CM and the RM (WCM−RM=0W_{\mathrm{CM-RM}} = 0). In this idealized scenario, an analytical expression for the eigenenergies of Eq. (1) exists if the interatomic interaction is modeled using a contact δ\delta pseudopotential [87, 88].

However, numerical approaches are essential to accurately compute the system’s eigenenergies for other types of trapping potentials — such as those considered in this work. Similarly, if more general interatomic interactions are introduced, including Lennard-Jones, Morse, or realistic ab-initio potentials (as employed here), the problem becomes significantly more challenging, so that it can be only numerically solved since analytical solutions are no longer feasible.

In this work, we calculate the eigenenergies and the corresponding eigenfunctions for the Hamiltonian in Eq. (8) using the method introduced in the Ref. [70], which has been successfully applied to study ultracold atomic systems trapped in a large variety of sinusoidal-like optical lattices [24, 38, 64, 89]. If the trapping potential is symmetric with respect to the origin and separable in the different directions, then the Hamiltonian (1) belongs to the D2hD_{2h} point-symmetry group [90]. As a consequence, the system eigenfunctions satisfy the symmetries of the corresponding eight irreducible representations (IRs). The method first constructs the Hamiltonian matrices associated with Eqs. (9) and (10) in a basis set formed by BB splines and spherical harmonics YmlY_m^l (or combinations of them in order to account for the symmetry of a particular IR of interest). The diagonalization of these matrices yields the eigenenergies and eigenfunctions for the CM ψiCM(R)\psi_i^{\mathrm{CM}}(\mathbf{R}) and RM φjRM(r)\varphi_j^{\mathrm{RM}}(\mathbf{r}). Subsequently, a novel basis set formed of configurations constructed as products Ψi,j(R,r)=ψiCM(R)φjRM(r)\Psi_{i,j}(\mathbf{R},\mathbf{r})=\psi^{\mathrm{CM}}_i(\mathbf{R})\varphi^{\mathrm{RM}}_j(\mathbf{r}) of the previous single-coordinate wavefunctions is created. Finally, the Hamiltonian matrix (8) is computed in the Ψi,j\Psi_{i,j} full configuration-interaction basis set, and eventually diagonalized, this rendering the eigenenergies and eigenfunctions for the complete two-particle system.1

Results and discussion

In this section, we present the main findings of our study along with the corresponding discussion. This section is divided into two parts. § 3.1 contains the results for the DWM introduced in § 2.2. § 3.2 discusses the performance of SW models (both harmonic, SWM2, and anharmonic, SWM6), paying special attention to an apparent symmetry violation, which does not appear in the DWM.

Multiwell effects on inelastic confinement-induced resonances

Quasi-one dimensional confinement

For the case of a Q1D trap, Figure 2 shows the energy spectra of the totally symmetric eigenstates of Eq. (1), i.e., the ones belonging to the AgA_g IR [90]. The trapping potential is described by the DWM, i.e., by a DW in the zz direction and the SW as defined in Eq. (4) (sextic trap) in the xx and yy directions. As can be seen, the energy spectrum comprises two series of branches with two different slopes as a function of the az/asca_z/a_{\mathrm{sc}} ratio, depending on their underlying RM contributions, with az=2ℏ/(mLiωz)a_z=\sqrt{2\hbar/(m_{\mathrm{Li}}\omega_z)} the characteristic trap size. The states of interest for this work are highlighted with the same colors as those considered in the Ref. [1], see Table 1.

Energy spectra for totally symmetric eigenstates of two ultracold $^{7}\mathrm{Li}$ atoms

Figure 2. Energy spectra for totally symmetric eigenstates (AgA_g irreducible representation) of two ultracold 7Li^{7}\mathrm{Li} atoms confined in an optical lattice modeled with the sextic trap (4) in the xx and yy directions, and a double well (6) in the zz direction [model (iii)–DWM] as a function of the ratio between the characteristic trap size az=2ℏ/(mLiωz)a_z = \sqrt{2\hbar/(m_{\mathrm{Li}}\omega_z)}, and the scattering length asca_{\mathrm{sc}}. The trap depth equals 35ER35E_{\mathrm{R}} in xx and yy directions, and 10.7ER10.7E_{\mathrm{R}}, in zz direction. Highlighted branches correspond to the states listed in the Table 1. Vertical dashed lines indicate the positions of the dispersive features reported in the experiment in Ref. [1]. Insets: enlarged view of some of the avoided crossings.

RM kindDWM RM stateDWM CM stateFig.SWM2 & SWM6 RM stateSWM2 & SWM6 CM stateFig.Color
MolecularφDW,agRM,LBS\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}ψDW,agCM\psi_{\mathrm{DW},a_g}^{\mathrm{CM}} (0,0,0)(0,0,0)3(a)φSW,agRM,LBS\varphi_{\mathrm{SW},a_g}^{\mathrm{RM,LBS}}ψSW,agCM\psi_{\mathrm{SW},a_g}^{\mathrm{CM}} (0,0,0)(0,0,0)3(a')Blue
φDW,agRM,LBS\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}ψDW,agCM\psi_{\mathrm{DW},a_g}^{\mathrm{CM}} (0,0,2)(0,0,2)3(b)φSW,agRM,LBS\varphi_{\mathrm{SW},a_g}^{\mathrm{RM,LBS}}ψSW,b1uCM\psi_{\mathrm{SW},b_{1u}}^{\mathrm{CM}} (0,0,1)(0,0,1)3(b')Magenta
φDW,agRM,LBS\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}ψDW,agCM\psi_{\mathrm{DW},a_g}^{\mathrm{CM}} (0,0,4)(0,0,4)3(c)φSW,agRM,LBS\varphi_{\mathrm{SW},a_g}^{\mathrm{RM,LBS}}ψSW,agCM\psi_{\mathrm{SW},a_g}^{\mathrm{CM}} (0,0,2)(0,0,2)3(c')Green
φDW,agRM,LBS\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}ψDW,agCM\psi_{\mathrm{DW},a_g}^{\mathrm{CM}} (0,0,6)(0,0,6)3(d)φSW,agRM,LBS\varphi_{\mathrm{SW},a_g}^{\mathrm{RM,LBS}}ψSW,b1uCM\psi_{\mathrm{SW},b_{1u}}^{\mathrm{CM}} (0,0,3)(0,0,3)3(d')Yellow
———φSW,agRM,LBS\varphi_{\mathrm{SW},a_g}^{\mathrm{RM,LBS}}ψSW,b1uCM\psi_{\mathrm{SW},b_{1u}}^{\mathrm{CM}} (2,0,1)(2,0,1)—Brown
———φSW,agRM,LBS\varphi_{\mathrm{SW},a_g}^{\mathrm{RM,LBS}}ψSW,b1uCM\psi_{\mathrm{SW},b_{1u}}^{\mathrm{CM}} (0,2,1)(0,2,1)—Brown
TrapφDW,agRM,TS1\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,TS1}}ψDW,agCM\psi_{\mathrm{DW},a_g}^{\mathrm{CM}} (0,0,0)(0,0,0)4(a)×××Black
φDW,agRM,TS2\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,TS2}}ψDW,agCM\psi_{\mathrm{DW},a_g}^{\mathrm{CM}} (0,0,0)(0,0,0)4(b)φSW,agRM,TS1\varphi_{\mathrm{SW},a_g}^{\mathrm{RM,TS1}}ψSW,agCM\psi_{\mathrm{SW},a_g}^{\mathrm{CM}} (0,0,0)(0,0,0)4(a')Dark blue
———φSW,agRM,TS1\varphi_{\mathrm{SW},a_g}^{\mathrm{RM,TS1}}ψSW,b1uCM\psi_{\mathrm{SW},b_{1u}}^{\mathrm{CM}} (0,0,1)(0,0,1)4(b')Dark blue

Table 1. Colors associated with the eigenstates of the double-well model DWM (iii) and the corresponding ones for the single-well models, both harmonic SWM2 (i) and sextic SWM6 (ii), together with the corresponding relative-motion (RM) kind of state (molecular or trap). The state marked with “×” does not exist, while those marked with “−” are not shown or highlighted. These conventions are used throughout Figure 2–Figure 9 (in Figure 6 and Figure 7, the results for the DWM are superimposed as solid olive lines).

On the one hand, the branches with the most pronounced dependence of the energy on the az/asca_z/a_{\mathrm{sc}} ratio correspond to molecular states showing a strong influence of the interatomic interaction and thus on asca_{\mathrm{sc}}. The corresponding total wavefunctions combine the RM least bound state φDW,agRM,LBS\varphi^{\mathrm{RM,LBS}}_{\mathrm{DW},a_g}, with different CM states ψDW,agCM(nxCM,nyCM,nzCM)\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(n_x^{\mathrm{CM}},n_y^{\mathrm{CM}},n_z^{\mathrm{CM}}). Cuts through the probability densities of the lowest-lying states are shown in Figure 3. As can be seen, in all cases the probability density in this plane is mostly localized along the diagonal z2=z1z_2=z_1, which also corresponds to the CM axis (where the RM coordinate is z=0z=0), as expected for a molecular state. Notice, also, that the number of nodes along that diagonal increases from the bottom to the top if the CM excitation in the zz direction, nzCMn_z^{\mathrm{CM}}, changes from 0 [Figure 3(a)], to 2 [Figure 3(b)], 4 [Figure 3(c)], and 6 [Figure 3(d)]; while, in these cases, nxCM=nyCM=0n_x^{\mathrm{CM}}=n_y^{\mathrm{CM}}=0 holds.

Cuts along the $z_1,z_2$ plane of the probability densities through the least-bound molecular states

Figure 3. Cuts along the z1,z2z_1,z_2 plane of the probability densities through the least-bound molecular states (gray elongated lobes) for a quasi-one-dimensional optical trap (Vx=Vy=35ERV_x = V_y = 35 E_R, Vz=10.7ERV_z = 10.7 E_R) (a) to (d) for the double-well potential and thus the states ϕDW,agRM,LBSψDW,agCM(0,0,nDWCM)\phi^{\mathrm{RM,LBS}}_{\mathrm{DW},a_g}\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(0,0,n^{\mathrm{CM}}_{\mathrm{DW}}) and (a’) to (d’) for the sextic single-well potential and the states ϕSW6,agRM,LBSψSW6,ag/b1uCM(0,0,nSW6CM)\phi^{\mathrm{RM,LBS}}_{\mathrm{SW6},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW6},a_g/b_{1u}}(0,0,n^{\mathrm{CM}}_{\mathrm{SW6}}), with nDWCMn^{\mathrm{CM}}_{\mathrm{DW}} and nSW6CMn^{\mathrm{CM}}_{\mathrm{SW6}} specified in the lower right corner of each panel. The lines mark the equipotential lines for V(z1,z2)/VzV(z_1,z_2)/V_z proportional to 0.5 using the same colors as listed in the Table 1.

On the other hand, the almost horizontal curves in Figure 2, whose energies are almost unaffected by the az/asca_z/a_{\mathrm{sc}} ratio, are associated with RM trap states. In the absence of a trap they

would correspond to two free (unbound) atoms. As in the case of the molecular states, the CM part ψCM(nxCM,nyCM,nzCM)\psi^{\mathrm{CM}}(n_x^{\mathrm{CM}},n_y^{\mathrm{CM}},n_z^{\mathrm{CM}}) allows for different numbers of CM excitations. In the figure, the first-trap state φDW,agRM,TS1ψDW,agCM(0,0,0)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,TS1}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,0,0) and the second trap state φDW,agRM,TS2ψDW,agCM(0,0,0)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,TS2}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,0,0) have been marked in black and dark blue colors, respectively. The probability densities of these two states are presented in Figure 4. As can be seen, the (lowest-lying) first trap state shown in Figure 4(a) has most of its probability density along the off-diagonal z2=−z1z_2=-z_1 and thus at different wells, which is associated with a large RM coordinate, i. e., a configuration where both atoms are located at different wells. The probability density along the diagonal z2=z1z_2=z_1 is, on the contrary, comparatively negligible. Conversely, the maxima of the probability density of the second trap state is primarily concentrated close to the diagonal z2=z1z_2=z_1, which corresponds to a small value of the RM coordinate, i. e., a configuration where both atoms are located at the same well. However, in contrast to the bound molecular states, this state is with respect to RM an antibonding one, as can be seen from the node along the diagonal z1=z2z_1=z_2. This different configuration also manifests in the slope of the corresponding branches. Note, while the second trap state is moderately influenced by variations in the az/asca_z/a_{\mathrm{sc}} ratio (a behavior that is also observed in the SW model, see discussion in § 3.2), the energy of the first trap state remains almost totally constant within the considered range (in fact, even in the non-interacting limits, where az/asc→+∞a_z/a_{\mathrm{sc}}\to+\infty, too). As a consequence, the first trap state of the DW potential plays the role of a perfect Mott insulator with practically no interactions [77, 78, 82, 83] and single-site occupancy in an (ideal) infinite optical lattice [see sketch in Figure 1(a)].

Four contour plots for the double-well and sextic single-well models

Figure 4. Same as Figure 3, but for (a) the first and (b) the second trap states of the double-well model, as well as for (a’) the first AgA_g and (b’) B1uB_{1u} trap states of the sextic single-well model (cf. § 2.2).

Due to symmetry conservation, if the interaction, i. e., asca_{\mathrm{sc}}, or the trap size, i. e., aza_z, are tuned and the energy of a molecular state reaches that of a trap state with the same symmetry (in this case the AgA_g IR), an avoided crossing emerges in the spectrum [91]. This is associated with an ICIR, because at such an avoided crossing an unbound (trapped) atom pair can be transferred into a molecular bound state and vice versa. The right inset of Figure 2 shows enlarged views of the avoided crossings that involve the first trap state, a Mott insulator, and the lowest-lying molecular states of interest. All these avoided crossings are quite broad with widths of order O(10−2ℏωz)\mathcal{O}(10^{-2}\hbar\omega_z). The molecular states in Figure 2 shown in different colors are φDW,agRM,LBSψDW,agCM(0,0,2)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,0,2) (magenta), φDW,agRM,LBSψDW,agCM(0,0,4)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,0,4) (green), and φDW,agRM,LBSψDW,agCM(0,0,6)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,0,6) (yellow). As they are non-degenerate, each one is responsible for the emergence of a separate ICIR.

The vertical colored lines in Figure 2 mark the positions of the dispersive features observed in the experiment described by Lee et al. [1]. Remarkably, the positions of the avoided crossings (clearly visible in the insets), i. e., of the ICIRs in the DWM, show a very good agreement with the experimental results. Recall that the energy spectrum has been obtained using only the given experimental parameters, such as the laser intensity and an ab-initio atom–atom interaction, without introducing any fitting procedure or additional approximations beyond those present in the experiment itself.

Finally, notice the existence of a very wide avoided crossing (magenta star), with an az/asca_z/a_{\mathrm{sc}} ratio relatively close to the dispersive feature related to the narrow FR (leftmost thin dashed line) (see Sec. 2.1). As can be well seen in the enlarged view of the left inset of the figure, this avoided crossing is substantially wider (∼7\sim7 times wider) than the other ones discussed. This avoided crossing corresponds to an ICIR due to the interaction between the molecular state φDW,agRM,LBSψDW,agRM,LBS(0,0,4)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}(0,0,4) and the second trap state φDW,agRM,TS2ψDW,agCM(0,0,0)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,TS2}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,0,0). While the large width should, in principle, lead to a good experimental visibility of this state, such a detection would certainly require initial population of the second trap state in which, as discussed above, both atoms are predominantly located in the same well which is also responsible for the large width. Such a population of the second trap state may occur due to thermal excitations, an imperfect preparation of the Mott state [24], or diabatic effects due to a fast ramping of the magnetic field with which the scattering length is tuned. The study of this avoided crossing, i. e., ICIR, could be performed in more detail in an experiment starting with a Mott insulator with double occupancy on each site [78], or, even better, with a tweezer array that would enable a more exquisite control of the atoms’ positions [40, 92, 93].

Three-dimensional confinement

Figure 5 shows the results for a deep 3D optical trap (1st experimental setting, Vx=Vy=VzV_x = V_y = V_z, see Sec. 2.1) described by the DWM considered in § 3.1.1, where the positions of three avoided crossings have been highlighted.

Energy spectrum for a trap with a depth equal to 35 $E_{\mathrm{R}}$ in all directions

Figure 5. Same as Figure 2, but for a trap with a depth equal to 35 ERE_{\mathrm{R}} in all directions.

As in the Q1D setup, the branches which show a strong dependence on the az/asca_z/a_{\mathrm{sc}} ratio correspond to molecular states, while those with a smaller dependence are associated with trap states (see Table 1). However, unlike in the Q1D setup, several states shown in Figure 5 are nearly degenerate. This arises because, relative to the xx and yy directions, the symmetry along the zz axis is only slightly broken by the presence of the DW. For example, while the state φDW,agRM,LBSψDW,agCM(0,0,2)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,0,2) (magenta) is non-degenerate, the states φDW,agRM,LBSψDW,agCM(2,0,0)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(2,0,0) and φDW,agRM,LBSψDW,agCM(0,2,0)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,2,0) (green) are degenerate, and their energies are very similar to those of the state φDW,agRM,LBSψDW,agCM(0,0,4)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,0,4) (also green). Similarly, the state φDW,agRM,LBSψDW,agCM(0,0,6)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,0,6) (yellow), has an energy that is only slightly smaller than those of the degenerate states φDW,agRM,LBSψDW,agCM(2,0,2)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(2,0,2) and φDW,agRM,LBSψDW,agCM(0,2,2)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,2,2) (brown).

As can be seen in the enlarged view of the avoided crossings, the values of their widths are several orders of magnitude smaller than those for the Q1D setup (note the different scales of the insets in Figure 2 compared to those in the inset in Figure 5). These avoided crossings involve the first trap state (Mott insulator) φDW,agRM,TS1ψDW,agCM(0,0,0)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,TS1}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,0,0) and the molecular states φDW,agRM,LBSψDW,agCM(0,0,2)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,0,2) (magenta), φDW,agRM,LBSψDW,agCM(2,0,0)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(2,0,0), and φDW,agRM,LBSψDW,agCM(0,2,0)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,2,0) (green), φDW,agRM,LBSψDW,agCM(0,0,4)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,0,4) (green), and φDW,agRM,LBSψDW,agCM(0,0,6)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,0,6) (yellow). We believe that the dramatic difference in the widths of the avoided crossings is responsible of the absence of dispersive features in the 3D experiment, in agreement with previous related predictions by Bolda et al. [67] and Schneider et al. [86]. Thus, experimentally observing these ICIRs would require a very fine tuning of the interparticle interactions at a much slower pace to follow the corresponding adiabatic curves, according to the Landau-Zener theory [94, 95].

Validity and limits of single-well models

In this section, we discuss the appropriateness of a SW description of the system to predict the dispersive features reported in the experiment in Ref. [1]. For this purpose, the energy spectrum is shown for the anisotropic Q1D setups SWM2 (harmonic) and SWM6 (sextic) in Figure 6(a) and Figure 6(b), respectively. The corresponding results for the isotropic 3D trap are presented in Figure 7(a) and Figure 7(b). For comparison, the DWM eigenenergies of interest [cf. the colored curves in Figure 2 and Figure 5] have been also superimposed. For the sake of clarity, we show only those SW eigenstates that are totally symmetric with respect to all Cartesian planes, i. e., they belong to the AgA_g IR, and those that are symmetric with respect to the xzxz and yzyz Cartesian planes, and antisymmetric with respect to the xyxy plane, i. e., they belong to the B1uB_{1u} IR [90].

Energy spectra for the $A_g$ and $B_{1u}$ irreducible representations

Figure 6. Same as Figure 2 but for the eigenstates of the AgA_g and B1uB_{1u} irreducible representations and an isotropic single-well model using the Taylor expansion given by Eq. (4) up to order (a) 2 (harmonic approximation), with some of the results rendered by the regularized δ\delta pseudopotential superimposed (dotted lines), and (b) 6 (sextic approximation). The solid olive lines show some of the energies of the AgA_g double-well states of Figure 2.

Plots of energy spectra for an isotropic trap

Figure 7. Same as Figure 6, but for an isotropic trap with Vx=Vy=Vz=35ERV_x = V_y = V_z = 35 E_{\mathrm{R}}.

The discussion, and thus this section, is divided in five parts. First, the reproduction of a Mott insulator state with SW models is briefly discussed in Sec. 3.2.1. Second, Sec. 3.2.2 is devoted to the SW predictions, as well as their implications on the symmetry conservation. Third, Sec. 3.2.3 details how the SW limitations can be overcome to deal with multiwell phenomena. Fourth, Sec. 3.2.4 summarizes the performance of a regularized δ\delta interatomic potential. Fifth, Sec. 3.2.5 contains the main findings of this work by comparing the ICIR positions of all considered models, and their relation to the experimental features.

Presence and absence of trap states

The energy branch of the first trap state obtained within the DWM (black lines in Figure 2 and Figure 5) and that plays the role of a Mott insulator in an infinite optical lattice (without short-range interatomic interactions) is absent in both SW models, as can be seen in Figure 6 and Figure 7. Instead, the first trap state of the SW, i. e., φSW,agRM,TS1ψSW,agCM(0,0,0)\varphi_{\mathrm{SW},a_g}^{\mathrm{RM,TS1}}\psi_{\mathrm{SW},a_g}^{\mathrm{CM}}(0,0,0) (lower dark blue curve) which belongs to the AgA_g IR [88], is similar to the second AgA_g trap state φDW,agRM,TS2ψDW,agCM(0,0,0)\varphi_{\mathrm{DW},a_g}^{\mathrm{RM,TS2}}\psi_{\mathrm{DW},a_g}^{\mathrm{CM}}(0,0,0) of the DWM. In fact, the probability densities of both states are also very similar, as is seen from the comparison of the cuts through the probability densities shown in Figure 4(a’) and Figure 4(a). The next trap state

in ascending order of energy is φSW,agRM,TS1ψSW,b1uCM(0,0,1)\varphi^{\mathrm{RM,TS1}}_{\mathrm{SW},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW},b_{1u}}(0,0,1) (upper dark blue curve) and belongs to the B1uB_{1u} IR. The corresponding probability density shown in Figure 4(b′) has a node at the origin. The probability density for both SWM6 states is mostly concentrated along the antidiagonal z2=−z1z_2=-z_1 (in the case of the B1uB_{1u} trap state slightly off-set due to the existence of a node along that line), which translates into a large value of the RM coordinate and, thus, to a large separation. However, the effects of the short-range interaction are noticeable and responsible for the probability density to be elongated along the diagonal z2=z1z_2=z_1.

Diabatic and adiabatic crossings, and symmetry considerations

Noteworthy, in the Q1D setup shown in Figure 6 the only crossing between the lowest-lying trap state φDW,agRM,TS1ψDW,agCM(0,0,0)\varphi^{\mathrm{RM,TS1}}_{\mathrm{DW},a_g}\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(0,0,0) and the colored molecular branches involves the molecular state φSW,agRM,LBSψSW,b1uCM(0,0,3)\varphi^{\mathrm{RM,LBS}}_{\mathrm{SW},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW},b_{1u}}(0,0,3) (yellow). Remind, however, that the DWM exhibits more crossings [cf. Figure 2]. As can be seen in the insets of Figure 6 and Figure 7, the mentioned crossing is a true (diabatic) crossing due to the different symmetries of the interacting states (B1uB_{1u} for the molecular state and AgA_g for the trap state). This diabatic crossing also appears in the 3D case (see right insets of Figure 7(a) and 7(b)). Contrarily, the corresponding crossing for the DWM is avoided [cf. the olive lines in the inset of Figure 6(b)]. Recall that under such an isotropic trapping potential, the molecular state φSW,agRM,LBSψSW,b1uCM(0,0,3)\varphi^{\mathrm{RM,LBS}}_{\mathrm{SW},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW},b_{1u}}(0,0,3) is degenerate with the states φSW,agRM,LBSψSW,b2uCM(0,3,0)\varphi^{\mathrm{RM,LBS}}_{\mathrm{SW},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW},b_{2u}}(0,3,0) and φSW,agRM,LBSψSW,b3uCM(3,0,0)\varphi^{\mathrm{RM,LBS}}_{\mathrm{SW},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW},b_{3u}}(3,0,0). In all cases, the mentioned crossings cannot be associated with an ICIR, since the crossings are true ones, i. e., not avoided crossings. However, in the 3D case more crossings with the lowest-lying trap state φSW,agRM,TS1ψSW,agCM(0,0,0)\varphi^{\mathrm{RM,TS1}}_{\mathrm{SW},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW},a_g}(0,0,0) are present, as can be seen in Figure 7. For example, the molecular states φSW,agRM,LBSψSW,agCM(0,0,2)\varphi^{\mathrm{RM,LBS}}_{\mathrm{SW},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW},a_g}(0,0,2) (green) and φSW,agRM,LBSψSW,b1uCM(0,0,3)\varphi^{\mathrm{RM,LBS}}_{\mathrm{SW},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW},b_{1u}}(0,0,3) (yellow) cross the trap states φSW,agRM,TS1ψSW,agCM(0,0,0)\varphi^{\mathrm{RM,TS1}}_{\mathrm{SW},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW},a_g}(0,0,0) (lower dark blue curve), and φSW,agRM,TS1ψSW,b1uCM(0,0,1)\varphi^{\mathrm{RM,TS1}}_{\mathrm{SW},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW},b_{1u}}(0,0,1) (top dark blue), respectively. Furthermore, the left insets of Figure 7 show a dramatic difference between the previous crossings for SWM2 and SWM6: while within the harmonic approximation the crossings are true ones, the anharmonicities present in the SWM6 force the crossings between states with the same symmetry to be avoided, as they concern states with the same symmetry [91]. Notice in Figure 7(b) that the energy of the (yellow) state φSW6,agRM,LBSψSW6,b1uCM(0,0,3)\varphi^{\mathrm{RM,LBS}}_{\mathrm{SW6},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW6},b_{1u}}(0,0,3),2 which is non-degenerate and very similar to that of the state φDW,agRM,LBSψDW,agCM(0,0,6)\varphi^{\mathrm{RM,LBS}}_{\mathrm{DW},a_g}\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(0,0,6), is only slightly smaller than that of the degenerate states (brown) φSW6,agRM,LBSψSW6,b1uCM(2,0,1)\varphi^{\mathrm{RM,LBS}}_{\mathrm{SW6},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW6},b_{1u}}(2,0,1) and φSW6,agRM,LBSψSW6,b1uCM(0,2,1)\varphi^{\mathrm{RM,LBS}}_{\mathrm{SW6},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW6},b_{1u}}(0,2,1), and also very similar to the energies of the states φDW,agRM,LBSψDW,agCM(2,0,2)\varphi^{\mathrm{RM,LBS}}_{\mathrm{DW},a_g}\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(2,0,2) and φDW,agRM,LBSψDW,agCM(0,2,2)\varphi^{\mathrm{RM,LBS}}_{\mathrm{DW},a_g}\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(0,2,2). These results stem from the structural similarities among the constituent orbitals. On the one hand, the RM orbitals φSW6,agRM,LBS\varphi^{\mathrm{RM,LBS}}_{\mathrm{SW6},a_g} and φDW,agRM,LBS\varphi^{\mathrm{RM,LBS}}_{\mathrm{DW},a_g} are closely matched, both displaying wild oscillations near the origin. On the other hand, their CM orbitals also share similar spatial profiles: each features two excitations in either the xx or yy direction and a single excitation per well along zz. Specifically, the SWM6 states ψSW6,b1uCM(2,0,1)\psi^{\mathrm{CM}}_{\mathrm{SW6},b_{1u}}(2,0,1) and ψSW6,b1uCM(0,2,1)\psi^{\mathrm{CM}}_{\mathrm{SW6},b_{1u}}(0,2,1) contain a single zz excitation overall within the (single) well, whereas the DWM states ψDW,agCM(2,0,2)\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(2,0,2) and ψDW,agCM(0,2,2)\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(0,2,2) possess two zz excitations, where one is localized in each well.

Multiwell effects described within single-well approximations

The SW models can be used to analyze multiwell effects [1], if the zero-energy point related to the Mott insulator is included ad hoc in the energy spectrum. This energy can be approximated by the ground-state energy of the non-interacting case (az/asc→∞a_z/a_{\mathrm{sc}} \to\infty). Then, the (artificially introduced) crossings between the energy of this manually added state and the ones of the molecular branches, which have been marked with empty symbols in Figure 6 and Figure 7, can be effectively used to predict ICIRs involving the Mott insulator state. The reason is twofold. On the one hand, the zero-crossing energy for the SW, especially for the SWM6, lies very close to the energy of the state φDW,agRM,TS1ψDW,agCM(0,0,0)\varphi^{\mathrm{RM,TS1}}_{\mathrm{DW},a_g}\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(0,0,0). On the other hand, the energies of the SW molecular states, especially in the SWM6 case, are very similar to the ones of the DWM molecular states. Concretely, the energies of the states φDW,agRM,LBSψDW,agCM(0,0,nDWCM)\varphi^{\mathrm{RM,LBS}}_{\mathrm{DW},a_g}\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(0,0,n^{\mathrm{CM}}_{\mathrm{DW}}) are very similar to those of the states φSW6,agRM,LBSψSW6,agCM(0,0,nSWCM)\varphi^{\mathrm{RM,LBS}}_{\mathrm{SW6},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW6},a_g}(0,0,n^{\mathrm{CM}}_{\mathrm{SW}}), and φSW,agRM,LBSψSW,b1uCM(0,0,nSWCM)\varphi^{\mathrm{RM,LBS}}_{\mathrm{SW},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW},b_{1u}}(0,0,n^{\mathrm{CM}}_{\mathrm{SW}}), depending on whether nSWCM=nDWCM/2n^{\mathrm{CM}}_{\mathrm{SW}}=n^{\mathrm{CM}}_{\mathrm{DW}}/2 is even or odd, respectively 3. The agreement between the SW6 and DW eigenenergies is also manifested in the remarkable similarities between their molecular states shown in Figure 3 within one of the wells, especially for those that are less affected by the anharmonicities in the optical potential due to their smaller number of CM excitations. In the 3D case, more states are degenerate due to the higher symmetry.

To unravel the similarities and the differences between the SW (SW2 and SW6) and DW models in more detail, Figure 8 shows cuts along the zz direction of the lowest-lying CM eigenstates that form those total eigenstates (see Table 1) that are responsible for the previously discussed ICIRs. Here, both the symmetric (left column) and the antisymmetric (right column) CM eigenstates for the DWM are plotted. As can be seen, the overlaps confirm that the DWM eigenstates agree more closely with the SWM6 eigenstates than with those of SWM2. This better agreement is also observed for the corresponding eigenenergies. Notice, also, that this agreement is particularly remarkable between the SWM6 and the DWM lowest-lying eigenstates. These results indicate the importance of the anharmonicities in the optical trap, and the consequent limitations of the harmonic approximation. Similarly, note that the SWM eigenstates show a better agreement with the antisymmetric than with the symmetric DWM eigenstates. This fact can be explained due to the existence of a central node in the antisymmetric eigenstates of the DWM, which nullifies the probability density of the wave function, mimicking better the situation of a SW where the density is by construction decaying exponentially outside the well.

Figure 8. Cuts along the zz direction of the lowest-lying center-of-mass eigenstates of the Hamiltonian (9) for a quasi-one-dimensional optical lattice with depth Vz=10.7ERV_z = 10.7 E_R. The lattice is modeled using three different potentials: a single-well harmonic approximation (3) (dash-dotted brown, left wells), a single-well sextic potential (4) (dashed cyan, right wells), and a double-well potential (6) (thin solid olive). The eigenstate profiles are plotted using line styles matching those of their respective model potentials and colored according to the total eigenstates shown in Figure 3. The number of excitations for each eigenstate is indicated in the lower right corner of each panel, and the overlaps of the single-well states with the double-well states on the top. The eigenenergies are shown (with the same color and line style) as horizontal lines. The (arbitrary phases) of the SW eigenstates are chosen to match the corresponding DW eigenstates.

Finally, notice that the DW states shown in the left column of Figure 8 can (approximately) be obtained by the addition and subtraction of the SW states (right column), a common feature of DW potentials, such as the one experienced by the electron of the hydrogen molecular ion, there explaining the occurrence of bonding and antibonding orbitals. This concept of superposition of single-well eigenfunctions connects the present results also with the lattice-induced-resonance model by Stecher et al. in Ref. [63], where symmetric and antisymmetric Wannier wavefunctions are considered to explain the formation of molecular dimers by the association of two atoms confined in neighboring sites of optical lattices.

Remarkably, although totally symmetric (AgA_g) two-particle states confined in a 3D optical lattice can be constructed with combinations of RM and CM orbitals with other symmetries, e.g., with RM and CM orbitals where both belong to the b1gb_{1g} IR, only the fully symmetric RM and CM orbitals comprise a molecular bound state that can potentially couple to the trap states. Other symmetries yield either deeply bound states that do not enable this coupling or trap states where the two atoms are far apart. Thus, for the occurrence of ICIRs involving the first-trap state that plays the role of a Mott insulator state in an infinite optical lattice, only the aga_g CM symmetric states (see cuts along the zz axis in the left column of Figure 8) are relevant. Contrarily, the two-particle wave function for the CM antisymmetric states (right column of Figure 8) do not present any avoided crossing with the mentioned first-trap state due to their strong energetic difference in the RM part.

δ\delta contact pseudopotential

Another very popular approximation in describing ultracold atoms is to model the interatomic interaction by a (regularized) δ\delta contact potential. This was also used in the modeling of the experiment in [1]. As can be seen in Figure 6(a) and Figure 7(a), within the attractive regime (asc<0a_{\mathrm{sc}} < 0) the energy levels obtained with the δ\delta potential are very similar to the ones of the ab-initio results. Conversely, the finite-size effects of the interaction become noticeable within the repulsive regime (asc>0a_{\mathrm{sc}} > 0), especially for small interactions, i. e., small absolute values of asca_{\mathrm{sc}}, where the pseudopotential pronouncedly underestimates the ab-initio results, see also [96]. Still, in the Q1D setup, the crossing points of the first-trap state of the DWM with the molecular states calculated using the δ\delta pseudopotential (and a harmonic trap) are in surprisingly good agreement with both the DWM and the experimental results for nSW2CM=2n_{\mathrm{SW2}}^{\mathrm{CM}} = 2 (green) and 3 (yellow). This is due to error cancellation, as the smaller molecular energy for the pseudopotential compensates the larger energy of the harmonic trap with respect to SWM6 and DWM. (Note the increasing energy difference in Figure 8 as the number of CM excitations rises.) As a consequence, the positions of the avoided crossings, and thus of the corresponding ICIRs, are accidentally agreeing better with experiment in the case of the pseudopotential than the ab-initio results. The positions of the crossings are also substantially different for the 3D trap, but here no experimental results exist for comparison. Notice, however, that the results rendered by the pseudopotential would be worse than those of the ab-initio interaction for other models, such as SWM6 or DWM, which provide, nonetheless, a more appropriate description of the optical lattice.

Positions of the avoided crossing for the double- and single-well models

Figure 9 shows the positions of the crossings observed in the SWM and DWM spectra. This figure summarizes the main results of this work. Recall that while the SW models (SW2: filled yellow triangle, SW6: filled yellow diamond) present only one true crossing involving the first trap state φSW,agRM,TS1ψSW,agCM(0,0,0)\varphi_{\mathrm{SW},a_g}^{\mathrm{RM,TS1}}\psi_{\mathrm{SW},a_g}^{\mathrm{CM}}(0,0,0) and φSW,agRM,LBSψSW,b1uCM(0,0,3)\varphi_{\mathrm{SW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{SW},b_{1u}}^{\mathrm{CM}}(0,0,3), the DWM shows three avoided crossing (filled circles) in agreement with the experimental features (dashed lines). However, the three crossings of the zero-energy point of the SWM2 (empty triangles) show a moderate agreement with the experimental and with the DWM results. This agreement is substantially better for the SWM6 (empty diamonds and squares), a result that highlights the importance of anharmonicities in the optical lattice. Note, also, that due to the already discussed cancellation of errors the δ\delta pseudopotential (thin triangles) yields more accurate results than the ab-initio interacting potential (thick triangles) for φSW,agRM,LBSψSW,agCM(0,0,2)\varphi_{\mathrm{SW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{SW},a_g}^{\mathrm{CM}}(0,0,2) (green) and φSW,agRM,LBSψSW,b1uCM(0,0,3)\varphi_{\mathrm{SW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{SW},b_{1u}}^{\mathrm{CM}}(0,0,3) (yellow), but a less accurate result for φSW,agRM,LBSψSW,b1uCM(0,0,1)\varphi_{\mathrm{SW},a_g}^{\mathrm{RM,LBS}}\psi_{\mathrm{SW},b_{1u}}^{\mathrm{CM}}(0,0,1) (magenta). Of course, only the avoided crossings (absent in the SW2 approximation) reveal also the coupling strengths and thus the widths of the ICIRs.

Figure 9. Positions of the avoided crossings for the optical trap with a depth of Vx=Vy=35ERV_x=V_y=35E_{\mathrm{R}} and (a) Vz=10.7ERV_z=10.7E_{\mathrm{R}} or (b) Vz=35ERV_z=35E_{\mathrm{R}}, with their the number of center-of-mass excitations for the double-well of Figure 2 and Figure 5 (left axis), shown as circles, and the single-well models (right axis) of Figure 6 and Figure 7, shown as diamonds and squares (sextic potential) and triangles (harmonic approximation), respectively. The filled symbols correspond to true avoided crossings between the molecular states φDW/SW,agRM,LBSψDW/SW,agCM(0,0,nDW/SWCM)\varphi^{\mathrm{RM,LBS}}_{\mathrm{DW/SW},a_g}\psi^{\mathrm{CM}}_{\mathrm{DW/SW},a_g}(0,0,n^{\mathrm{CM}}_{\mathrm{DW/SW}}), with nSWCM=nDWCM/2=1,2,3n^{\mathrm{CM}}_{\mathrm{SW}}=n^{\mathrm{CM}}_{\mathrm{DW}}/2=1,2,3, and the first trap state φDW/SW,agRM,TS1ψDW/SW,agCM(0,0,0)\varphi^{\mathrm{RM,TS1}}_{\mathrm{DW/SW},a_g}\psi^{\mathrm{CM}}_{\mathrm{DW/SW},a_g}(0,0,0) as they appear in the energy spectrum. The empty symbols correspond to the values of az/asca_z/a_{\mathrm{sc}} where the energy of the corresponding molecular state agrees with the one of non-interacting atoms (which emulates the perfect Mott-insulator ground state.) The interatomic potential is described using ab-initio curves, except for the (thin) triangles, which correspond to the δ\delta contact pseudopotential.

The comparison between the positions of the avoided crossings (filled symbols), i. e., of the ICIRs, for the Q1D and 3D setups shown in Figure 9(a) and Figure 9(b), respectively, also renders inter-

esting results. On the one hand, the ICIR that involves the molecular state φDW,agRM,LBSψDW,agCM(0,0,2)\varphi^{\mathrm{RM,LBS}}_{\mathrm{DW},a_g}\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(0,0,2), whose energy is very close to that of φSW6,agRM,LBSψSW6,b1uCM(0,0,1)\varphi^{\mathrm{RM,LBS}}_{\mathrm{SW6},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW6},b_{1u}}(0,0,1), and the trap state φDW,agRM,TS1ψDW,agCM(0,0,0)\varphi^{\mathrm{RM,TS1}}_{\mathrm{DW},a_g}\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(0,0,0) moves from the attractive branch of the spectrum (asc=−2651 a0a_{\mathrm{sc}}=-2651\,a_0, az/asc=−0.944a_z/a_{\mathrm{sc}}=-0.944) to a position close to the unitary limit within the repulsive branch of the spectrum (magenta symbols). The corresponding scattering length is one order of magnitude larger than in the Q1D case. On the other hand, due to the stronger confinement in the 3D case, the other two ICIRs associated with the molecular states φDW,agRM,LBSψDW,agCM(0,0,4)\varphi^{\mathrm{RM,LBS}}_{\mathrm{DW},a_g}\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(0,0,4) (green circle) and φDW,agRM,LBSψDW,agCM(0,0,6)\varphi^{\mathrm{RM,LBS}}_{\mathrm{DW},a_g}\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(0,0,6) (yellow circle), whose energies are similar to those of the states φSW6,agRM,LBSψSW6,agCM(0,0,2)\varphi^{\mathrm{RM,LBS}}_{\mathrm{SW6},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW6},a_g}(0,0,2) and φSW6,agRM,LBSψSW6,b1uCM(0,0,3)\varphi^{\mathrm{RM,LBS}}_{\mathrm{SW6},a_g}\psi^{\mathrm{CM}}_{\mathrm{SW6},b_{1u}}(0,0,3), respectively, appear for substantially larger interactions in the Q1D setup than in the 3D case, as the values of the corresponding scattering lengths are much larger (asc=5411 a0a_{\mathrm{sc}}=5411\,a_0 vs asc=1724 a0a_{\mathrm{sc}}=1724\,a_0, i.e., az/asc=0.462a_z/a_{\mathrm{sc}}=0.462 vs 1.079, for φDW,agRM,LBSψDW,agCM(0,0,4)\varphi^{\mathrm{RM,LBS}}_{\mathrm{DW},a_g}\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(0,0,4), and asc=2328 a0a_{\mathrm{sc}}=2328\,a_0 vs asc=1145 a0a_{\mathrm{sc}}=1145\,a_0, i.e., az/asc=1.075a_z/a_{\mathrm{sc}}=1.075 vs 1.625, for φDW,agRM,LBSψDW,agCM(0,0,6)\varphi^{\mathrm{RM,LBS}}_{\mathrm{DW},a_g}\psi^{\mathrm{CM}}_{\mathrm{DW},a_g}(0,0,6)).

Alternative pictures for explaining the experimentally observed resonance features

As already mentioned, the authors of Ref. [1] relate their experimental results to lattice-induced resonances [63] and emphasize that they cannot be confinement-induced resonances (see the introduction and, especially, footnote [25] in Ref. [1]). The main argument for this claim is based on symmetry, namely that the experimental features are, within the adopted model, ascribed to odd-parity excitations that should be symmetry forbidden. The symmetry breaking in lattice-induced resonances is explained by interaction-driven tunneling. Since the present work claims ICIRs to be responsible for the resonances observed in Ref. [1] and also in view of earlier confusion regarding confinement-induced resonances, some comments appear to be necessary.

First of all, the name inelastic confinement-induced resonances (ICIR) was introduced in Ref. [64] in order to distinguish those confinement-induced resonances that are based on CM-RM coupling between a bound molecular state and an unbound trap state from the ones that involve only the RM and may be understood as a divergence in the effective interaction (scattering length) due to dimensional confinement to Q1D [65, 66] or Q2D [56] geometry which were originally termed confinement-induced resonances and later renamed into elastic confinement-induced resonances [64]. The most striking difference between the two types of CIRs is that only in the case of ICIRs molecule formation or breaking is possible, since the binding energy can be transferred in between RM and CM degrees of freedom, as was predicted in Ref. [64] and experimentally confirmed in Ref. [38] where overlapping elastic and inelastic CIRs were observed and distinguished. With the given definition, CIR include elastic and inelastic CIRs, while inelastic ICIRs include among others anharmonic CIRs as well as lattice-induced resonances as special cases.

Second, as discussed in § 3.2.2, the claimed symmetry violation of an ICIR-based explanation in Ref. [1] is due to a limitation of the adopted model where the crossing of the eigenstates of a single-well potential with the non-interacting state is adopted. As is demonstrated, already a DWM for describing two neighbour sites of a lattice does not encounter symmetry issues. Noteworthy, in the here adopted model the resonant features occur as avoided crossings in the ab-initio energy spectrum, confirming the coupling between the involved molecular bound state and the trap state. If the coupling would vanish due to symmetry, a true instead of an avoided crossing would occur. The physical picture behind is that, quantum mechanically, there is the same probability for the molecule to form either in the left or in the right well. In fact, the obtained state is an (equal) superposition of the molecule being left or right. (It may be noted that this is also directly visible in the 2nd quantized formulation of the Hamiltonian given in Ref. [63].)

Third, it may be noted that the common solid-state physics formulation and language (hopping, tunneling) as is also adopted in Ref. [1] infer a time-dependent picture, even for the description of stationary states. This is in contrast to the stationary approach adopted in this work for the analysis of the energy spectra. This leads to different pictures describing, however, the same physical process. In the experiment the atoms are initially prepared in a singly-occupied Mott insulator state. Then, the magnetic field is varied resulting in a time-dependent variation of the two-particle interaction. The formation of a molecule requires the two atoms to be sufficiently close to each other, in the present case to be on the same lattice site. In one picture, this requires (active) tunneling of one of the atoms from its original site to the neighbor site, denoted as interaction-induced tunneling in Ref. [1]. (In 2nd quantization language, the atom is annihilated on one site and created on its neighbor site.) In the complementary picture, the molecule formation is possible even in a singly-occupied Mott state if there is a non-zero probability for the atom to occupy the neighbor site. Mathematically, the delocalization (imperfect localization of the ground-state Wannier function) and the hopping (tunneling) matrix elements are directly related, thus leading to the same molecule-formation probability. It may be noted that active tunneling from one site to the other would occur, e. g., if the lattice would be time-dependently tilted or the barrier between the wells would otherwise be varied [97].

Fourth, the present findings fully support the claim in Ref. [1] that in the experiment (also to the authors’ knowledge for the first time) the molecule formation of atoms involving different lattice sites has been observed. Furthermore, the present work shows that a quantitative modeling of the experimental results is possible already adopting a double-well model. (Remind that it was checked that a triple-well model did effectively not change the results.) Again, this finding is not surprising in view of the fact that the experiment started with a Mott-insulator state. This implies relatively localized Wannier functions and narrow bands. The hopping amplitudes are thus small relative to the on-site terms. As a consequence, the tight-binding approximation considering only next-neighbor interactions is a good approximation. Consequently, a double-well model should also be a reasonable approximation. While it may be very interesting to observe also "true" lattice resonances where atoms delocalized over the whole lattice may be turned into molecules, the corresponding coupling is expected to be very small [67].

Summary and outlook

In this work, we provide a comprehensive alternative explanation for the novel dispersive features experimentally observed by Lee et al. in Ref. [1] in ultracold atoms trapped in an optical lattice. Based on the analysis of various optical-lattice models, from idealized harmonic to complex multiwell configurations, we demonstrate that the experimental results are accurately explained by a trapping potential modeled with anharmonic single-wells in two directions and a double-well in the third. Specifically, the resonance features observed experimentally under quasi-one-dimensional trapping show very good agreement with the avoided crossings in the calculated two-particle spectrum. These avoided crossings involve molecular states (combining the relative-motion least-bound state with lowest-lying center-of-mass excitations) and a unique first-trap state of the double-well. This first-trap state naturally represents a perfect Mott insulator with single-site occupancy, a crucial aspect not captured by single-well descriptions of the optical lattice. Furthermore, Lee et al. attributed this phenomenon to lattice-induced resonances facilitated by tunneling. Our framework reinterprets the resonant features as inelastic confinement-induced resonances, which manifest as avoided crossings in the two-particle energy spectrum. We demonstrate that these avoided crossings, even when involving states delocalized across multiple wells, arise from the same fundamental mechanism: the coupling between the relative-motion and center-of-mass degrees of freedom due to the trapping potential’s anharmonicities. Therefore, we provide a unified view of both anharmonicity-induced resonances [61, 62], lattice-induced [63], and inelastic confinement-induced resonances [64], as we conclude that they share the same underlying physical origin. The ICIRs that were defined as resonances associated with avoided crossings between a molecular state and a trap state caused by the coupling between center-of-mass and relative motion may be seen as the general phenomenon, including anharmonicity-induced as well as lattice-induced resonances as special cases.

Furthermore, our model provides a possible explanation for the absence of similar resonance features under three-dimensional trapping for the discussed experiment, attributing it to the significantly narrower width of the corresponding avoided crossings in the two-particle spectrum [67, 86]. This suggests that observing multiwell inelastic confinement-induced resonances under three-dimensional confinement4 would require much finer tuning of interparticle interactions.

While the experimental Mott-insulator state is not fully captured by single-well descriptions, we also investigated the feasibility of such models in predicting inelastic confinement induced resonances positions. We found that the double-well spectrum’s avoided crossings align remarkably well with the crossings of the single-well molecular states (formed by the least-bound state and the lowest-lying center-of-mass one with the energy of the non-interacting state, which effectively resembles the zero-point energy associated with a perfect Mott insulator state. Our analysis further underscores the critical role of anharmonicities in the trapping potential, as evidenced by the improved accuracy of sixth-order Taylor series expansions over harmonic approximations. Notably, the double-well description resolves an apparent symmetry violation present in single-well models, providing a more complete and accurate representation of the system.

In summary, this work significantly advances the understanding of multiwell effects in optical lattices, which certainly opens new routes to control and manipulate novel correlated many-body states.

Acknowledgments

We thank E. Lee, H. Lin, and W. Ketterle for useful discussions and for providing the experimental data [1] prior to publication. We also extend our gratitude to Manuele Landini and Christoph Nägerl for insightful discussions. The authors acknowledge computing resources at the Magerit Supercomputer of the Universidad Politécnica de Madrid.

Generative AI tools were used to improve the clarity and readability of the manuscript. All content generated using these tools was critically reviewed, verified, and edited by the authors, who accept full responsibility for the final version of the article.

F.R. conceived the study and performed the numerical simulations. F.R. and A.S. analyzed the numerical results and wrote the manuscript.

Funding information This work has been supported by the Grant PID2021-122711NB-C21 funded by MCIN/AEI/10.13039/501100011033. AS gratefully acknowledges support by Berlin Quantum.

References

References

  1. [1]Y. K. Lee, H. Lin and W. Ketterle, Spin dynamics dominated by resonant tunneling into molecular states, Phys. Rev. Lett. 131, 213001 (2023), doi:10.1103/PhysRevLett.131.213001.DOI
  2. [2]C. J. Pethick and H. Smith, Bose-Einstein condensation in dilute gases, Cambridge University Press, Cambridge, UK (2002).
  3. [3]M. Lewenstein, A. Sanpera and V. Ahufinger, Ultracold Atoms in Optical Lattices: Simulating Quantum Many-Body Systems, Oxford University Press, Oxford, UK, 1st edn. (2012).DOI
  4. [4]R. Veyron, J.-B. Gerent, G. Baclet, V. Mancois, P. Bouyer and S. Bernon, In situ subwavelength microscopy of ultracold atoms using dressed excited states, PRX Quantum 5, 030349 (2024), doi:10.1103/PRXQuantum.5.030349.DOI
  5. [5]M. L. Wall and L. D. Carr, Microscopic model for Feshbach interacting Fermions in an optical lattice with arbitrary scattering length and resonance width, Phys. Rev. Lett. 109, 055302 (2012), doi:10.1103/PhysRevLett.109.055302.DOI
  6. [6]P. S. Julienne, E. Tiesinga and T. Köhler, Making cold molecules by time-dependent Feshbach resonances, Journal of Modern Optics 51, 1787 (2004), doi:10.1080/09500340408232491.DOI
  7. [7]F. H. Mies, E. Tiesinga and P. S. Julienne, Manipulation of Feshbach resonances in ultracold atomic collisions using time-dependent magnetic fields, Phys. Rev. A 61, 022721 (2000), doi:10.1103/PhysRevA.61.022721.DOI
  8. [8]C. H. Schunck, M. W. Zwierlein, C. A. Stan, S. M. F. Raupach, W. Ketterle, A. Simoni, E. Tiesinga, C. J. Williams and P. S. Julienne, Feshbach resonances in fermionic 6Li, Phys. Rev. A 71, 045601 (2005), doi:10.1103/PhysRevA.71.045601.DOI
  9. [9]A. Simoni, J.-M. Launay and P. Soldán, Feshbach resonances in ultracold atom-molecule collisions, Phys. Rev. A 79, 032701 (2009), doi:10.1103/PhysRevA.79.032701.DOI
  10. [10]C. Chin, R. Grimm, P. Julienne and E. Tiesinga, Feshbach resonances in ultracold gases, Rev. Mod. Phys. 82, 1225 (2010), doi:10.1103/RevModPhys.82.1225.
  11. [11]P. O. Fedichev, Y. Kagan, G. V. Shlyapnikov and J. T. M. Walraven, Influence of nearly resonant light on the scattering length in low-temperature atomic gases, Phys. Rev. Lett. 77, 2913 (1996), doi:10.1103/PhysRevLett.77.2913.DOI
  12. [12]M. Theis, G. Thalhammer, K. Winkler, M. Hellwig, G. Ruff, R. Grimm and J. H. Denschlag, Tuning the scattering length with an optically induced Feshbach resonance, Phys. Rev. Lett. 93, 123001 (2004), doi:10.1103/PhysRevLett.93.123001.DOI
  13. [13]G. Thalhammer, M. Theis, K. Winkler, R. Grimm and J. H. Denschlag, Inducing an optical Feshbach resonance via stimulated Raman coupling, Phys. Rev. A 71, 033403 (2005), doi:10.1103/PhysRevA.71.033403.DOI
  14. [14]N. Sándor, R. González-Férez, P. S. Julienne and G. Pupillo, Rydberg optical Feshbach resonances in cold gases, Phys. Rev. A 96, 032719 (2017), doi:10.1103/PhysRevA.96.032719.DOI
  15. [15]E. L. Boldo, E. Tiesinga and P. S. Julienne, Effective-scattering-length model of ultracold atomic collisions and Feshbach resonances in tight harmonic traps, Phys. Rev. A 66, 013403 (2002), doi:10.1103/PhysRevA.66.013403.DOI
  16. [16]M. Takamoto, F.-L. Hong, R. Higashi and H. Katori, An optical lattice clock, Nature 435, 321 (2005), doi:10.1038/nature03541.DOI
  17. [17]D. Jaksch, C. Bruder, J. I. Cirac, C. W. Gardiner and P. Zoller, Cold Bosonic atoms in optical lattices, Phys. Rev. Lett. 81, 3108 (1998), doi:10.1103/PhysRevLett.81.3108.DOI
  18. [18]I. Bloch, Ultracold quantum gases in optical lattices, Nat. Phys. 1, 23 (2005), doi:10.1038/nphys138.DOI
  19. [19]M. Lewenstein, A. Sanpera, V. Ahufinger, B. Damski, A. Sen and U. Sen, Ultracold atomic gases in optical lattices: mimicking condensed matter physics and beyond, Adv. Phys. 56, 243 (2007), doi:10.1080/00018730701223200.DOI
  20. [20]D. R. Meacher, Optical lattices – crystalline structure bound by light, Contemp. Phys. 39, 329 (1998), doi:10.1080/001075198181847.DOI
  21. [21]M. Köhl, H. Moritz, T. Stöferle, K. Günter and T. Esslinger, Fermionic atoms in a three dimensional optical lattice: Observing Fermi surfaces, dynamics, and interactions, Phys. Rev. Lett. 94, 080403 (2005), doi:10.1103/PhysRevLett.94.080403.DOI
  22. [22]C. Ospelkaus, S. Ospelkaus, L. Humbert, P. Ernst, K. Sengstock and K. Bongs, Ultracold heteronuclear molecules in a 3D optical lattice, Phys. Rev. Lett. 97, 120402 (2006), doi:10.1103/PhysRevLett.97.120402.DOI
  23. [23]U. Schneider, I. Hackermüller, S. Will, T. Best, I. Bloch, T. A. Costi, R. W. Helmes, D. Rasch and A. Rosch, Metallic and insulating phases of repulsively interacting Fermions in a 3D optical lattice, Science 322, 1520 (2008), doi:10.1126/science.1165449.DOI
  24. [24]D. Capecchi, G. Cantillano, M. J. Mark, F. Meinert, A. Schindewolf, M. Landini, A. Saenz, F. Revuelta and H.-C. Nägerl, Observation of confinement-induced resonances in a 3D lattice, Phys. Rev. Lett. 131, 213002 (2023), doi:10.1103/PhysRevLett.131.213002.DOI
  25. [25]M. Cristiani, O. Morsch, J. H. Müller, D. Ciampini and E. Arimondo, Experimental properties of Bose-Einstein condensates in one-dimensional optical lattices: Bloch oscillations, Landau-Zener tunneling, and mean-field effects, Phys. Rev. A 65, 063612 (2002), doi:10.1103/PhysRevA.65.063612.DOI
  26. [26]B. E. Granger and D. Blume, Tuning the interactions of spin-polarized Fermions using quasi-one-dimensional confinement, Phys. Rev. Lett. 92, 133202 (2004), doi:10.1103/PhysRevLett.92.133202.DOI
  27. [27]T. Kinoshita, T. Wenger and D. S. Weiss, Local pair correlations in one-dimensional Bose gases, Phys. Rev. Lett. 95, 190406 (2005), doi:10.1103/PhysRevLett.95.190406.DOI
  28. [28]N. N. Klausen, J. L. Bohn and C. H. Greene, Nature of spinor Bose-Einstein condensates in rubidium, Phys. Rev. A 64, 053602 (2001), doi:10.1103/PhysRevA.64.053602.DOI
  29. [29]C. Mora, R. Egger and A. O. Gogolin, Three-body problem for ultracold atoms in quasi-one-dimensional traps, Phys. Rev. A 71, 052705 (2005), doi:10.1103/PhysRevA.71.052705.DOI
  30. [30]M. Rigol, V. Rousseau, R. T. Scalettar and R. R. P. Singh, Collective oscillations of strongly correlated one-dimensional Bosons on a lattice, Phys. Rev. Lett. 95, 110402 (2005), doi:10.1103/PhysRevLett.95.110402.DOI
  31. [31]P. Naidon and P. S. Julienne, Optical Feshbach resonances of alkaline-earth-metal atoms in a one- or two-dimensional optical lattice, Phys. Rev. A 74, 062713 (2006), doi:10.1103/PhysRevA.74.062713.DOI
  32. [32]B. Fröhlich, M. Feld, E. Vogt, M. Koschorreck, W. Zwerger and M. Köhl, Radio-frequency spectroscopy of a strongly interacting two-dimensional Fermi gas, Phys. Rev. Lett. 106, 105301 (2011), doi:10.1103/PhysRevLett.106.105301.DOI
  33. [33]M. Wouters, J. Tempere and J. T. Devreese, Feshbach resonances in a quasi-two-dimensional atomic gas, Phys. Rev. A 68, 053603 (2003), doi:10.1103/PhysRevA.68.053603.DOI
  34. [34]C. Weitenberg, M. Endres, J. F. Sherson, M. Cheneau, P. Schausz, T. Fukuhara, I. Bloch and S. Kuhr, Single-spin addressing in an atomic Mott insulator, Nature 471, 319 (2011), doi:10.1038/nature09827.DOI
  35. [35]F. Schäfer, T. Fukuhara, S. Sugawa, Y. Takasu and Y. Takahashi, Tools for quantum simulation with ultracold atoms in optical lattices, Nature Reviews Physics 2, 411 (2020), doi:10.1038/s42254-020-0195-3.DOI
  36. [36]F. Serwane, G. Zürn, T. Lompe, T. B. Ottenstein, A. N. Wenz and S. Jochim, Deterministic preparation of a tunable few-Fermion system, Science 332(6027), 336 (2011), doi:10.1126/science.1201351.DOI
  37. [37]A. M. Kaufman, B. J. Lester and C. A. Regal, Cooling a single atom in an optical tweezer to its quantum ground state, Phys. Rev. X 2, 041014 (2012), doi:10.1103/PhysRevX.2.041014.DOI
  38. [38]S. Sala, G. Zürn, T. Lompe, A. Wenz, S. Murmann, F. Serwane, S. Jochim and A. Saenz, Coherent molecule formation in anharmonic potentials near confinement-induced resonances, Phys. Rev. Lett. 110, 203202 (2013), doi:10.1103/PhysRevLett.110.203202.DOI
  39. [39]Y. Yu, K. Wang, J. D. Hood, L. R. B. Picard, J. T. Zhang, W. B. Cairncross, J. M. Hutson, R. Gonzalez-Ferez, T. Rosenband and K.-K. Ni, Coherent optical creation of a single molecule, Phys. Rev. X 11, 031061 (2021), doi:10.1103/PhysRevX.11.031061.DOI
  40. [40]D. K. Ruttley, A. Guttridge, S. Spence, R. C. Bird, C. R. Le Sueur, J. M. Hutson and S. L. Cornish, Formation of ultracold molecules by merging optical tweezers, Phys. Rev. Lett. 130, 223401 (2023), doi:10.1103/PhysRevLett.130.223401.DOI
  41. [41]A. Guttridge, T. R. Hepworth, D. K. Ruttley, A. A. T. Durst, M. T. Eiles and S. L. Cornish, Individual assembly of two-species Rydberg molecules using optical tweezers, Phys. Rev. Lett. 134, 133401 (2025), doi:10.1103/PhysRevLett.134.133401.DOI
  42. [42]D. Barredo, V. Lienhard, S. de Léséleuc, T. Lahaye and A. Browaeys, Synthetic three-dimensional atomic structures assembled atom by atom, Nature 561, 79 (2018), doi:10.1038/s41586-018-0450-2.DOI
  43. [43]C. J. Turner, A. A. Michailidis, D. A. Abanin, M. Serbyn and Z. Papić, Weak ergodicity breaking from quantum many-body scars, Nat. Phys. 14, 745 (2018), doi:10.1038/s41567-018-0137-5.DOI
  44. [44]A. N. Wenz, G. Zürn, S. Murmann, I. Brouzos, T. Lompe and S. Jochim, From few to many: Observing the formation of a Fermi sea one atom at a time, Science 342, 457 (2013), doi:10.1126/science.1240516.DOI
  45. [45]L. Bayha, M. Holten, R. Klemt, K. Subramanian, J. Bierlin, S. M. Reimann, G. M. Bruun, P. M. Preiss and S. Jochim, Observing the emergence of a quantum phase transition shell by shell, Nature 587, 583 (2020), doi:10.1038/s41586-020-2936-y.DOI
  46. [46]E. Haller, M. J. Mark, R. Hart, J. G. Danzl, L. Reichsöllner, V. Melzikh, P. Schmelcher and H.-C. Nägerl, Confinement-induced resonances in low-dimensional quantum systems, Phys. Rev. Lett. 104, 153203 (2010).DOI
  47. [47]T. Sowiński and M. A. García-March, One-dimensional mixtures of several ultracold atoms: a review, Rep. Prog. Phys. 82, 104401 (2019), doi:10.1088/1361-6633/ab3a80.DOI
  48. [48]M. Greiner, O. Mandel, T. Esslinger, T. Hänsch and I. Bloch, Quantum phase transition from a superfluid to a Mott insulator in a gas of ultracold atoms, Nature 415, 39 (2002), doi:10.1038/415039a.DOI
  49. [49]P.-I. Schneider and A. Saenz, Quantum computation with ultracold atoms in a driven optical lattice, Phys. Rev. A 85, 050304 (2012), doi:10.1103/PhysRevA.85.050304.DOI
  50. [50]H. Bernien, S. Schwartz, A. Keesling, H. Levine, A. Omran, H. Pichler, S. Choi, A. S. Zibrov, M. Endres, M. Greiner, V. Vuletić and M. D. Lukin, Probing many-body dynamics on a 51-atom quantum simulator, Nature 551, 579 (2017), doi:10.1038/nature24622.DOI
  51. [51]D. Bluvstein, A. A. Geim, S. H. Li, S. J. Evered, J. P. Bonilla Ataides, G. Baranes, A. Gu, T. Manovitz, M. Xu, M. Kalinowski, S. Majidy, C. Kokail et al., A fault-tolerant neutral-atom architecture for universal quantum computation, Nature 649, 39 (2026), doi:10.1038/s41586-025-09848-5.DOI
  52. [52]D. S. Petrov and G. V. Shlyapnikov, Interatomic collisions in a tightly confined Bose gas, Phys. Rev. A 64, 012706 (2001), doi:10.1103/PhysRevA.64.012706.DOI
  53. [53]Z. Idziaszek and T. Calarco, Analytical solutions for the dynamics of two trapped interacting ultracold atoms, Phys. Rev. A 74, 022712 (2006), doi:10.1103/PhysRevA.74.022712.DOI
  54. [54]J.-J. Liang and C. Zhang, Two ultracold atoms in a completely anisotropic trap, Phys. Scr. 77, 025302 (2008), doi:10.1088/0031-8949/77/02/025302.DOI
  55. [55]Y. Chen, D.-W. Xiao, R. Zhang and P. Zhang, Analytical solution for the spectrum of two ultracold atoms in a completely anisotropic confinement, Phys. Rev. A 101, 053624 (2020), doi:10.1103/PhysRevA.101.053624.DOI
  56. [56]D. S. Petrov, M. Holzmann and G. V. Shlyapnikov, Bose-Einstein condensation in quasi-2D trapped gases, Phys. Rev. Lett. 84, 2551 (2000), doi:10.1103/PhysRevLett.84.2551.DOI
  57. [57]D. S. Petrov, G. V. Shlyapnikov and J. T. M. Walraven, Regimes of quantum degeneracy in trapped 1D gases, Phys. Rev. Lett. 85, 3745 (2000), doi:10.1103/PhysRevLett.85.3745.DOI
  58. [58]P. O. Fedichev, M. J. Bijlsma and P. Zoller, Extended molecules and geometric scattering resonances in optical lattices, Phys. Rev. Lett. 92, 080401 (2004), doi:10.1103/PhysRevLett.92.080401.DOI
  59. [59]T. Stöferle, H. Moritz, K. Günter, M. Köhl and T. Esslinger, Molecules of Fermionic atoms in an optical lattice, Phys. Rev. Lett. 96, 030401 (2006), doi:10.1103/PhysRevLett.96.030401.DOI
  60. [60]R. B. Diener and T.-L. Ho, Fermions in optical lattices swept across Feshbach resonances, Phys. Rev. Lett. 96, 010402 (2006), doi:10.1103/PhysRevLett.96.010402.DOI
  61. [61]J. P. Kestner and L.-M. Duan, Anharmonicity-induced resonances for ultracold atoms and their detection, New J. Phys. 12, 053016 (2010), doi:10.1088/1367-2630/12/5/053016.DOI
  62. [62]J. P. Kestner and L.-M. Duan, Effective single-band models for strongly interacting fermions in an optical lattice, Phys. Rev. A 81, 043618 (2010), doi:10.1103/PhysRevA.81.043618.DOI
  63. [63]J. von Stecher, V. Gurarie, L. Radzihovsky and A. M. Rey, Lattice-induced resonances in one-dimensional bosonic systems, Phys. Rev. Lett. 106, 235301 (2011), doi:10.1103/PhysRevLett.106.235301.DOI
  64. [64]S. Sala, P.-I. Schneider and A. Saenz, Inelastic confinement-induced resonances in low-dimensional quantum systems, Phys. Rev. Lett. 109, 073201 (2012), doi:10.1103/PhysRevLett.109.073201.DOI
  65. [65]M. Olshanii, Atomic scattering in the presence of an external confinement and gas of impenetrable Bosons, Phys. Rev. Lett. 81, 938 (1998), doi:10.1103/PhysRevLett.81.938.
  66. [66]T. Bergeman, M. G. Moore and M. Olshanii, Atom-atom scattering under cylindrical harmonic confinement: Numerical and analytic studies of the confinement induced resonance, Phys. Rev. Lett. 91, 163201 (2003), doi:10.1103/PhysRevLett.91.163201.DOI
  67. [67]E. L. Bolda, E. Tiesinga and P. S. Julienne, Ultracold dimer association induced by a far-off-resonance optical lattice, Phys. Rev. A 71, 033404 (2005), doi:10.1103/PhysRevA.71.033404.DOI
  68. [68]V. Peano, M. Thorwart, C. Mora and R. Egger, Confinement-induced resonances for a two-component ultracold atom gas in arbitrary quasi-one-dimensional traps, New J. Phys. 7, 192 (2005), doi:10.1088/1367-2630/7/1/192.DOI
  69. [69]S. Grishkevich and A. Saenz, Influence of a tight isotropic harmonic trap on photoassociation in ultracold homonuclear alkali-metal gases, Phys. Rev. A 76, 022704 (2007), doi:10.1103/PhysRevA.76.022704.DOI
  70. [70]S. Grishkevich and A. Saenz, Theoretical description of two ultracold atoms in a single site of a three-dimensional optical lattice using realistic interatomic interaction potentials, Phys. Rev. A 80, 013403 (2009), doi:10.1103/PhysRevA.80.013403.DOI
  71. [71]C. Baroni, G. Lamporesi and M. Zaccanti, Quantum mixtures of ultracold gases of neutral atoms, Nature Reviews Physics 6(12), 736 (2024), doi:10.1038/s42254-024-00773-6.DOI
  72. [72]R. Stock, I. H. Deutsch and E. L. Bolda, Quantum state control via trap-induced shape resonance in ultracold atomic collisions, Phys. Rev. Lett. 91, 183201 (2003), doi:10.1103/PhysRevLett.91.183201.DOI
  73. [73]M. Krych and Z. Idziaszek, Controlled collisions of two ultracold atoms in separate harmonic traps, Phys. Rev. A 80, 022710 (2009), doi:10.1103/PhysRevA.80.022710.DOI
  74. [74]S. O. Onyango, F. Revuelta and A. Saenz, Numerical treatment of a pair of particles in spatially displaced traps, In preparation.
  75. [75]G. Lamporesi, J. Catani, G. Barontini, Y. Nishida, M. Inguscio and F. Minardi, Scattering in mixed dimensions with ultracold gases, Phys. Rev. Lett. 104, 153202 (2010), doi:10.1103/PhysRevLett.104.153202.DOI
  76. [76]M. Troppenz, S. Sala, P.-I. Schneider and A. Saenz, Inelastic confinement-induced resonances in quantum dots, doi:10.48550/arXiv.1509.01159, ArXiv:1509.01159 (2015).DOI
  77. [77]N. Syassen, D. M. Bauer, M. Lettner, D. Dietze, T. Volz, S. Dürr and G. Rempe, Atom-molecule Rabi oscillations in a Mott insulator, Phys. Rev. Lett. 99, 033201 (2007), doi:10.1103/PhysRevLett.99.033201.DOI
  78. [78]R. Jördens, N. Strohmaier, K. Günter, H. Moritz and T. Esslinger, A Mott insulator of fermionic atoms in an optical lattice, Nature 455, 204 (2008), doi:10.1038/nature07244.DOI
  79. [79]V. Popkov, X. Zhang and A. Klümper, Phantom Bethe excitations and spin helix eigenstates in integrable periodic and open spin chains, Phys. Rev. B 104, L081410 (2021), doi:10.1103/PhysRevB.104.L081410.DOI
  80. [80]P. N. Jepsen, Y. K. E. Lee, H. Lin, I. Dimitrova, Y. Margalit, W. W. Ho and W. Ketterle, Long-lived phantom helix states in Heisenberg quantum magnets, Nat. Phys. 18, 899 (2022), doi:10.1038/s41567-022-01651-7.DOI
  81. [81]E. Lee, H. Lin and W. Ketterle, Private communication.
  82. [82]M. Greiner, O. Mandel, T. Esslinger, T. Hänsch and I. Bloch, Quantum phase transition from a superfluid to a Mott insulator in a gas of ultracold atoms, Nature 415, 39 (2002), doi:10.1038/415039a.DOI
  83. [83]W. S. Bakr, A. Peng, M. E. Tai, R. Ma, J. Simon, J. I. Gillen, S. Fölling, L. Pollet and M. Greiner, Probing the superfluid-to-Mott insulator transition at the single-atom level, Science 329, 547 (2010), doi:10.1126/science.1192368.DOI
  84. [84]M. A. Garcia-March, D. R. Dounas-Frazer and L. D. Carr, Macroscopic superposition states of ultracold bosons in a double-well potential, Frontiers of Physics 7(1), 131 (2012), doi:10.1007/s11467-011-0236-6.DOI
  85. [85]J. Gillet, M. A. Garcia-March, T. Busch and F. Sols, Tunneling, self-trapping, and manipulation of higher modes of a Bose-Einstein condensate in a double well, Phys. Rev. A 89, 023614 (2014), doi:10.1103/PhysRevA.89.023614.DOI
  86. [86]P.-I. Schneider, S. Grishkevich and A. Saenz, Ab initio determination of Bose-Hubbard parameters for two ultracold atoms in an optical lattice using a three-well potential, Phys. Rev. A 80, 013404 (2009), doi:10.1103/PhysRevA.80.013404.DOI
  87. [87]D. Blume and C. H. Greene, Fermi pseudopotential approximation: Two particles under external confinement, Phys. Rev. A 65, 043613 (2002), doi:10.1103/physreva.65.043613.DOI
  88. [88]T. Busch, B.-G. Englert, K. Rzażewski and M. Wilkens, Two cold atoms in a harmonic trap, Foundations of Physics 28, 549 (1998), doi:10.1023/A:1018705520999.DOI
  89. [89]S. Grishkevich, S. Sala and A. Saenz, Theoretical description of two ultracold atoms in finite three-dimensional optical lattices using realistic interatomic interaction potentials, Phys. Rev. A 84, 062710 (2011), doi:10.1103/PhysRevA.84.062710.DOI
  90. [90]J. J. Rotman, An Introduction to the Theory of Groups, vol. 148 of Graduate Texts in Mathematics, Springer, New York, 4 edn. (1995).DOI
  91. [91]J. von Neumann and E. P. Wigner, On the behavior of eigenvalues in adiabatic processes, Physikalische Zeitschrift 30, 467 (1929), English translation in Quantum Chemistry: Classic Scientific Papers, edited by H. Hettema, World Scientific, Singapore, 2000, p. 25.
  92. [92]S. Murmann, A. Bergschneider, V. M. Klinkhamer, G. Zürn, T. Lompe and S. Jochim, Two fermions in a double well: Exploring a fundamental building block of the hubbard model, Phys. Rev. Lett. 114, 080402 (2015), doi:10.1103/PhysRevLett.114.080402.DOI
  93. [93]J. D. Hood, Y. Yu, Y.-W. Lin, J. T. Zhang, K. Wang, L. R. Liu, B. Gao and K.-K. Ni, Multi-channel interactions of two atoms in an optical tweezer, Phys. Rev. Res. 2, 023108 (2020), doi:10.1103/PhysRevResearch.2.023108.DOI
  94. [94]L. Landau, On the theory of transfer of energy at collisions II, Phys. Z. Sowjetunion 2, 46 (1932).
  95. [95]C. Zener, Non-adiabatic crossing of energy levels, Proceedings of the Royal Society of London. Series A, Containing Papers of a Mathematical and Physical Character 137(833), 696 (1932), doi:10.1098/rspa.1932.0165.DOI
  96. [96]J. Wulff, M. u. Rehman and A. Saenz, Impact of the confinement strength on the energy spectra of two atoms interacting via zero-range or realistic potentials, Few-Body Systems 67(4), 61 (2026), doi:10.1007/s00601-026-02081-4.DOI
  97. [97]F. Meinert, M. J. Mark, E. Kirilov, K. Lauber, P. Weinmann, M. Gröbner, A. J. Daley and H.-C. Nägerl, Observation of many-body dynamics in long-range tunneling after a quantum quench, Science 344, 1259 (2014), doi:10.1126/science.1248402, https://www.science.org/doi/pdf/10.1126/science.1248402.DOI

Paper details

vlnc:W_4ft8z6at
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Generative AI tools

View statement
Looking for related papers…
Contents