Pseudogap regime of the unitary Fermi gas with lattice auxiliary-field quantum Monte Carlo in the continuum limit
Abstract
The unitary Fermi gas (UFG) is a strongly correlated system of two-species (spin-1/2) fermions with a short-range attractive interaction modeled by a contact interaction and has attracted much interest across different disciplines. The UFG is considered a paradigm for strongly correlated superfluids and has been investigated extensively, with generally good agreement found between theory and experiment. However, the extent of a pseudogap regime above the critical temperature for superfluidity, in which pairing correlations persist, is still debated both theoretically and experimentally. Here we study thermodynamic properties of the UFG across the superfluid phase transition using finite-temperature lattice auxiliary-field quantum Monte Carlo (AFMC) methods in the canonical ensemble of fixed particle numbers. We present results for the condensate fraction, a model-independent energy-staggering pairing gap, the spin susceptibility, and a free energy-staggering pairing gap. We extrapolate our lattice AFMC results to the continuous time and continuum limits, thus removing the systematic error associated with the finite filling factor of previous AFMC studies. Applying a finite-size scaling analysis to the condensate fraction results, we determine the critical temperature to be . For the largest particle number studied , the energy-staggering pairing gap, which provides a direct signature of a pseudogap regime, is suppressed above a pairing scale temperature of . The spin susceptibility displays moderate suppression above with a spin gap temperature of . We also calculate a free energy-staggering pairing gap, which shows substantially reduced statistical errors when compared with the energy-staggering gap, allowing for a clear signature of pairing correlations in the finite-size system. All results indicate that the pseudogap regime is narrow, with pseudogap signatures emerging at temperatures below . The reduced statistical errors of the free energy gap enable an extrapolation at low temperatures, allowing an estimate of the zero-temperature pairing gap .
I Introduction
The unitary Fermi gas (UFG), with diverging s-wave scattering length, is of great interest in diverse areas of physics as it provides a well-defined paradigm for strongly correlated systems relevant to the study of nuclei, quark matter, neutron stars, and atomic Fermi gases. It undergoes a phase transition from a normal strongly correlated gas to an s-wave superfluid Leggett (2006); Bloch et al. (2008); Giorgini et al. (2008); Inguscio et al. (2008); Zwerger (2012); Levin et al. (2012); Svistunov et al. (2015); Pitaevskii and Stringari (2016) below a critical temperature . The UFG is realized in the middle of a crossover from a weakly interacting Fermi gas described by the Bardeen-Cooper-Schrieffer (BCS) Bardeen et al. (1957) theory to a weakly repulsive Bose-Einstein condensate (BEC) Zwerger (2012); Eagles (1969); Leggett (1980); Nozières and Schmitt-Rink (1985) (for in the superfluid phase). The UFG has remarkable properties as one of the highest-temperature superfluids in units of the Fermi temperature Randeria (2010). It exhibits a nearly perfect fluid behavior Enss (2012); Bluhm et al. (2017); Nishida and Abuki (2005), and can be realized in ultracold atomic Fermi gas experiments with and Ketterle and Zwierlein (2008); Hu et al. (2022).
There is no known controlled expansion in a small parameter (around some nearby exact solution) to approach the superfluid transition of the UFG. It is therefore difficult to theoretically address this system on a quantitative level. The UFG, due in part to its simplicity and broad interest, has become a testing ground for strongly correlated many-body methods. Successful approaches in advancing our understanding of the UFG include the epsilon expansion Nishida and Son (2006), the Luttinger-Ward approach Haussmann et al. (2007, 2009); Zwerger (2016); Frank et al. (2018), conformal field theory methods Nishida and Son (2007a, b), quantum critical approaches Nikolić and Sachdev (2007); Enss (2012), non-self consistent diagrammatic resummation methods Perali et al. (2002a); Ohashi and Griffin (2002); Chen et al. (2005), and various quantum Monte Carlo methods Gilbreth and Alhassid (2013); Bulgac et al. (2006, 2008); Drut et al. (2012); Akkineni et al. (2007); Burovski et al. (2006a, b, 2008); Goulko and Wingate (2010); Van Houcke et al. (2011); Hu et al. (2010); Lee and Schäfer (2006); Endres et al. (2013).
The UFG has been extensively investigated experimentally in recent decades using ultracold Fermi gases Regal et al. (2004); Zwierlein et al. (2004); Kinast et al. (2004); Chin et al. (2004); Zwierlein et al. (2005); Nascimbene et al. (2010); Nascimbène et al. (2011); Ku et al. (2012); Zwerger (2012); Mukherjee et al. (2017), but important open problems remain. A topic that has generated considerable interest is the existence and extent of a pseudogap regime above the critical temperature for superfluidity, in which there is no off-diagonal-long-range order (ODLRO) but pairing correlations persist with a gapped single-particle spectrum near the Fermi surface.
The extent of pairing correlations in the UFG has been debated in recent years. For low-temperatures (), studies of pairing properties demonstrated good agreement between theory and experiment Shin et al. (2007); Gezerlis and Carlson (2008); Carlson et al. (2003); Chang et al. (2004); Carlson and Reddy (2005, 2008); Carlson et al. (2011). However, at finite temperature, and in particular across the superfluid critical temperature , our understanding is more limited. While there has been substantial progress in our understanding of the normal phase both theoretically Kashimura et al. (2012a, b); van Wyk et al. (2016); Perali et al. (2002b, 2011a); Palestini et al. (2012a); Tajima et al. (2016); Pini et al. (2019); Perali et al. (2002a, 2011b); Palestini et al. (2012b); Pantel et al. (2014); Tajima et al. (2014); Santos (1994); Jankó et al. (1997); Sewer et al. (2002); Chien et al. (2010); Chien and Levin (2010); Wlazłowski et al. (2013a); Magierski et al. (2009, 2011); Enss and Haussmann (2012); Jensen et al. (2020a) and experimentally Gaebler et al. (2010); Ku et al. (2012); Nascimbene et al. (2010); Nascimbène et al. (2011); Sommer et al. (2011); Wulin et al. (2011), our understanding of pairing correlations (even qualitatively) above is still incomplete. For reviews of the pseudogap regime in the UFG, see Refs. Chen et al. (2005); Chen and Wang (2014); Mueller (2017); Jensen et al. (2019).
In this work, we calculate several thermodynamic properties of the spin-balanced homogeneous UFG using lattice auxiliary-field quantum Monte Carlo (AFMC) in the canonical ensemble of fixed particle numbers. In particular, we calculate the condensate fraction, a model-independent energy-staggering pairing gap, the spin susceptibility, and the free energy-staggering pairing gap for and particles. All properties are calculated in the continuum limit in both imaginary time and lattice spacing, thus removing systematic errors of a finite-filling factor of our previous work Jensen et al. (2020a) and allowing direct comparison with experiments.
Applying finite-size scaling to the condensate fraction data, we determine the superfluid critical temperature of the UFG to be . The canonical ensemble framework enables the calculation of an energy-staggering pairing gap and a free energy-staggering pairing gap without the need for a numerically challenging analytic continuation. In particular, the free energy-staggering gap has remarkably small statistical errors, allowing for an extrapolation from finite temperature to estimate accurately the energy-staggering pairing gap (where is the Fermi energy). From the pairing gaps for our largest particle number simulated, , we estimate an upper bound on the pairing temperature scale associated with these observables to be (this pairing temperature decreases with increasing particle number ). We find a similar value for the spin gap temperature of the spin susceptibility (for ), . These results for and indicate a narrow temperature pseudogap regime for the UFG.
The outline of this paper is as follows. In Sec. II we provide background material for the UFG and discuss progress toward understanding the pseudogap regime, with a focus on recent lattice Monte Carlo studies. In Sec. III we describe our canonical-ensemble AFMC method, and the recent optimization algorithms that allow for large lattice simulations. In Sec. IV we present the AFMC results of this work, and in Sec. V we provide concluding remarks.
II BCS-BEC crossover and the UFG
The BCS-BEC crossover in three spatial dimensions is a continuous crossover (at temperature ) as a function of (where is the s-wave scattering length and is the Fermi wavenumber) from a weakly attractive Fermi gas described by BCS theory to a system of weakly interacting bosons described by BEC theory Zwerger (2012); Giorgini et al. (2008). As temperature is increased, the gas exhibits a (3D XY) superfluid phase transition into a normal disordered phase. The BCS regime corresponds to where the normal phase can be described by Fermi liquid theory. The BEC regime, corresponding to , is a weakly interacting Bose gas of tightly bound dimers with binding energy where is the mass of a single fermion. In the BCS regime, for there are Cooper pairs of spatial extent much greater than the interparticle separation where is the particle density for fermions in a spatial volume . In the BEC regime, the pairs are tightly bound dimers of size much smaller than the interparticle separation. In the unitary limit, defined by (i.e., infinite scattering length), the two-particle bound state is a zero-energy resonance. In this regime, the size of the Cooper pairs is comparable to the interparticle separation and the system is strongly correlated.
II.1 Scattering amplitude and the continuum model
We consider a dilute Fermi gas at low temperature. The scattering between two particles with wavevectors and is well described by the low-momentum s-wave expansion of the inverse of the scattering amplitude ,
(1) |
where is the relative momentum wavevector , the effective range, and refers to higher-order corrections in . This expansion holds for 6Li and 40K, where the low-energy interaction at large distance is well described with the van der Waals potential with a microscopic interparticle interaction range of ( is Planck’s constant) in the limits . Here is the thermal de Broglie wavelength ( is the Boltzmann constant), and the Fermi wavenumber is given by . See Ref. Khuri et al. (2009) for a detailed discussion of the low-momentum scattering expansion. Though the effective range parameter is often of the order (for a broad Feshbach resonance ) this is not guaranteed to hold Castin and Werner (2012); Werner and Castin (2012). The unitary limit is reached when the modulus of the scattering amplitude is maximized, saturating the constraint imposed by unitarity, from the optical theorem, , and leading to . We thus define the UFG by the conditions: , , , and .
These conditions are met with cold atomic Fermi gases using Feshbach resonance techniques to tune an external magnetic field so that a closed channel bound state becomes resonant with open channel scattering (see Refs. Chin et al. (2004, 2010); Giorgini et al. (2008)). Near a broad Feshbach resonance, many properties of the gas can be understood with a single-channel model using a contact interaction between spin-up and spin-down fermions (corresponding to an atomic hyperfine state manifold with two relevant states). More specifically, we model a system of spin-up, and spin-down, fermions in a three-dimensional cubic box of side length with periodic boundary conditions. The continuum Hamiltonian of this system is given by
(2) |
where and are, respectively, creation and annihilation operators for a fermion at position with spin projection . The contact interaction model is ill-defined without regularization. The model can be regularized with an ultra-violet (UV) cutoff in momentum space.
II.2 The pseudogap regime
A pseudogap regime is known to exist in high- superconductors. For instance, the specific heat is strongly suppressed above in contrast to conventional superconductors which show Fermi liquid behavior for low temperatures above Dagotto (1994); Timusk and Statt (1999). The mechanism leading to the pseudogap regime of the cuprates is still not fully understood Randeria (2010); Dagotto (1994); Timusk and Statt (1999) as the complexity, competing orders, and strongly correlated nature of the problem are difficult to address with theoretical approaches. The BCS-BEC crossover is also strongly correlated for but the ordering is well understood to be a superfluid phase transition with no competing orderings and the appropriate microscopic model is simple. Pseudogap phenomena in this system can be attributed to precursor pairing correlations. One can define a temperature scale as the temperature at which signatures of pairing first appear as the temperature is lowered from the normal phase. There are different definitions of that depend on the choice of observable used to determine the signatures of pairing correlations Mueller (2017); Jensen et al. (2019). Precursor pairing develops in the BCS-BEC crossover for as the coupling changes from the BCS regime where Fetter and Walecka (1971) to the BEC regime where and is the dissociation temperature scale for breaking the two-body bound dimers due to thermal fluctuations. However, it is not clear how this precursor pairing emerges. It is a challenging many-body problem to calculate the extent of pairing signatures above for .
Many different techniques have been applied to the UFG and the calculation of observables that are expected to show pairing structure. The AFMC studies of Refs. Magierski et al. (2009, 2011); Wlazłowski et al. (2013a), where a spherical cutoff was used in the lattice model, found signatures of pairing above the critical temperature. Refs. Chien et al. (2010); Chien and Levin (2010); Perali et al. (2011b); Kashimura et al. (2012b); Palestini et al. (2012c, b); Tajima et al. (2014); Wulin et al. (2011); Pini et al. (2019, 2020) applied non-self-consistent diagrammatic approaches and found varying degrees of pairing above the critical temperature and gapped signatures in support of a pronounced pseudogap regime for the UFG. In contrast to these works, the self-consistent Luttinger-Ward diagrammatic approach of Refs. Haussmann et al. (2009); Enss and Haussmann (2012) did not find a pronounced pseudogap regime and found that and are very close for the UFG (see Refs. Pini et al. (2019, 2020) for comparisons between self-consistent and non-self-consistent approaches). On the experimental side, the results also show disagreement. The works of Refs. Wulin et al. (2011); Gaebler et al. (2010) found pseudogap signatures whereas Refs. Ku et al. (2012); Nascimbene et al. (2010); Nascimbène et al. (2011); Sommer et al. (2011) saw no gapped signatures.
In the UFG, the value of () is still debated in the literature. Discrepancies exist even when the same pairing observables are used to determine , as the temperature dependence of these observables differ substantially between various works. In Sec. IV we present controlled AFMC results for two pseudogap signatures: the temperature dependence of the spin susceptibility and the energy-staggering pairing gap across the superfluid phase transition.
II.3 Lattice Model
The continuum model described by the Hamiltonian in Eq. (2) can be replaced by a discrete lattice model and the continuum limit is then recovered by extrapolating to zero filling factor.
To construct the lattice model, we discretize a cubic box of volume with lattice sites forming a three-dimensional cubic lattice with lattice spacing . The lattice of linear size has lattice sites at positions when is odd. The lattice Hamiltonian for a contact on-site attractive interaction is given by
(3) |
where and are, respectively, creation and annihilation operators for discrete quasi-momentum and spin projection . In this work we use the quadratic single-particle dispersion relation though other dispersion relations have also been used (see Ref. Werner and Castin (2012) for a discussion of dispersion relations). The lattice operators obey the anticommutation relations . The number operator at lattice site and spin projection is with the lattice field operators obeying . The constant in Eq.(3) is the rescaled coupling constant. The bare coupling constant is determined by the requirement that the physical -wave scattering length ( for the UFG) is reproduced on the lattice. This leads to
(4) |
where the integral in (4) is over the first Brillouin zone of the lattice. Relation (4) can be derived by solving the Lippmann-Schwinger equation on an infinite lattice.
The UFG is recovered by solving the lattice model for and then taking the continuum limit when the filling factor on the lattice for given particle number , followed by the thermodynamic limit . The approach to the continuum limit was first addressed in the works of Refs. Burovski et al. (2006b); Pricoupenko and Castin (2007). Following these works, we expect the many-body energies to be linear in for a low filling factor . In the present work (see Sec. IV), we take the continuum limit for various thermodynamic observables using a linear fit in for a given particle number .
II.4 Lattice AFMC studies
Lattice AFMC methods are among the most accurate, but the main challenges have been (i) taking the continuum limit (or equivalently ), and (ii) taking the thermodynamic limit of large particle number . There have been several works using large-scale lattice AFMC methods to study signatures of pairing correlations in the UFG at finite temperature. These studies have improved the understanding of finite-temperature properties of the UFG. However, satisfactory control of systematic errors has not been achieved in these studies. We summarize the current status and recent works in this section.
Spherical cutoff studies.— In Refs. Bulgac et al. (2006, 2008); Burovski et al. (2008); Magierski et al. (2009, 2011); Wlazłowski et al. (2012, 2013a, 2013b, 2015) the UFG was studied using a spherical cutoff in the single-particle quasi-momentum space. Ref. Burovski et al. (2008) studied a continuum model where the single-particle quasi-momentum states in the lab frame were restricted to have magnitude less than a cutoff with a renormalization condition similar to Eq. (4) but where the integration region is determined by a spherical cutoff in quasi-momentum space (i.e., for and for ). This model does not allow scattering to lab frame quasi-momenta states with . Refs. Bulgac et al. (2006, 2008); Magierski et al. (2009, 2011); Wlazłowski et al. (2012, 2013a, 2013b, 2015) used a similar model but on a periodic lattice. Pseudogap signatures at unitarity were observed with this model in the simulations of Refs. Magierski et al. (2009, 2011); Wlazłowski et al. (2013a) for temperatures below .
Ref. Werner and Castin (2012) calculated the two-particle scattering in the continuum limit using a spherical cutoff model and found that the spherical cutoff does not produce the desired two-body scattering amplitude for finite center-of-mass momentum . The low-energy inverse scattering amplitude was found to have the form where (typical and are of order for scattering of the degenerate UFG). Comparing this with Eq. (1), the extra center-of-mass term, , shifts the scattering length to an effective value, away from unitarity () towards the BCS side. This argument suggests the spherical cutoff model underestimates pseudogap effects. However, collision effects for varying center-of-mass momentum in the spherical cutoff model lead to enhanced pairing effects Jensen et al. (2019, 2020a).
Ref. Jensen et al. (2020a) performed a many-body calculation with and without the spherical cutoff. The comparison presented there is not sufficient to address differences at the many-body level in the continuum limit. The effective range parameter and the center-of-mass parameter are different when using a spherical cutoff, as opposed to including the full first Brillouin zone (i.e., they approach the continuum limit differently). To address this, simulations were performed in Ref. Jensen et al. (2019) for the spin susceptibility for particles with multiple lattice sizes to approach the continuum limit. These studies, however, do not resolve the magnitude of the discrepancy in the thermodynamic limit, which has not been addressed.
The spherical cutoff model reduces the single-particle model space and allows for larger lattice simulations within the AFMC framework to more closely approach the continuum and thermodynamic limits (for given computational resources). The early works utilizing this truncation provided the first lattice AFMC investigation of finite temperature pairing signatures. However, recent algorithm advances Drut et al. (2011); Jensen et al. (2019); Richie-Halford et al. (2020); Gilbreth et al. (2021) and increased availability of computational resources have allowed for large-lattice simulations using the full first Brillouin zone for the single-particle model space.
Full first Brillouin zone studies.— There have been a number of studies with finite-temperature AFMC using the full first Brillouin zone with the quadratic single-particle dispersion relation. The early work of Ref. Lee and Schäfer (2006) performed finite filling factor calculations to find a number of observables including the thermal energy and spin susceptibility. This study was performed for smaller lattice sizes up to and down to temperatures of providing the first lattice AFMC estimate of the Bertsch parameter . Ref. Drut et al. (2011) produced the first large-scale () finite-temperature AFMC calculations for the UFG using the full first Brillouin zone (see Ref. Drut and Nicholson (2013) for details on algorithm advances). However, this work presented results for the contact and did not focus on pseudogap properties.
Recently, Ref. Richie-Halford et al. (2020) calculated pairing observables for for multiple values of the interaction parameter along the BCS-BEC crossover and obtained estimates for as a function of . This work reached much smaller filling factors compared with the simulations of Ref. Jensen et al. (2020a) and provided estimates throughout the entire strongly correlated regime of the BCS-BEC crossover. Ref. Richie-Halford et al. (2020) restricted the study to a moderate particle number and did not take the lattice continuum limit.
In Refs. Jensen et al. (2020a, 2019), the current authors used the canonical ensemble algorithm discussed in Sec. III, to produce a large-scale study at finite filling factor for pairing observables using the full first Brillouin zone. These works presented results for the spin susceptibility, condensate fraction, and the energy-staggering pairing gap. Results were presented for multiple particle numbers up to . This work found a negligible pseudogap regime based on the analysis of the pairing gap and spin susceptibility for large particle number. However, these results had sizable lattice effects with an effective range parameter and cannot be directly compared with experiment or other UFG results. This presents a large systematic error relative to the UFG with vanishing effective range parameters as defined in Sec. II.1. The current work (see results in Sec. IV) addresses the systematic errors of these previous studies.
III Finite-temperature lattice AFMC
We calculate thermodynamic properties of the system with the Hamiltonian of Eq. (2) using a discrete lattice of size and total particle number with controllable systematic errors as discussed in Sec. II.3. In particular, we have implemented finite-temperature auxiliary-field quantum Monte Carlo (AFMC) methods in the canonical ensemble of fixed particle numbers and .
III.1 Hubbard-Stratonovich transformation
The Gibbs operator at inverse temperature can be thought of as the propagator in imaginary-time . Decomposing the Hamiltonian into , where is the kinetic energy operator and is the two-body interaction, we use a symmetric Trotter decomposition
(5) |
where we have divided the imaginary time into discrete time slices of length . The number operator, , at site is given by . For fermions, each interaction term in Eq. (2) can be written as . Completing the squares, we obtain the Hubbard-Stratonovich transformation Hubbard (1959); Stratonovich (1957),
(6) |
where for and for . This is repeated for each of the time slices and lattice sites, and leads, in the limit , to the functional integral form
(7) |
The integration in Eq. (7) is over auxiliary fields defined for each lattice site and imaginary time slice (), and the integration measure . is a time-ordered product for the propagator of non-interacting fermions in the imaginary-time dependent fields , given explicitly by
(8) |
where and is a Gaussian weight. The operator describes the many-particle propagator of the non-interacting spin-balanced system of fermions in external time-dependent one-body auxiliary fields .
III.2 Thermal expectation values of observables and Gaussian quadratures
The thermal expectation value of an observable is given by
(9) |
where is a positive-definite weight function, is the sign function, and is the expectation value for a single configuration of the auxiliary fields. At each site and time slice, we approximate the integral over by Gaussian quadrature
(10) |
where and . The weights are given by and . We sample the discrete auxiliary fields according to the distribution using the Metropolis algorithm Metropolis et al. (1953).
III.3 Particle-number projection
The particle-number projection operator for particles and single-particle states can be written as a discrete Fourier transform Ormand et al. (1994)
(11) |
where () are quadrature points, and is a chemical potential introduced to ensure the numerical stability of the Fourier sum. is chosen so as to reproduce the correct number of particles in a grand-canonical average in the ensemble , i.e., . We can calculate grand-canonical traces over the many-particle Fock space by using the matrix representation of the many-particle propagator in the single-particle space. For example
(12) |
where is the identity matrix in the single-particle space. Thus the canonical trace of at fixed particle number is given by
(13) |
In our AFMC calculations, we choose the single-particle basis to be the single-particle states with good quasi-momentum and spin projection . The single-particle propagator matrices are then evaluated in momentum space. When calculating the short-time propagator with the linearized interaction term, we obtain a speedup for the local interaction in coordinate space using a fast Fourier transform approach Batrouni et al. (1986); Davies et al. (1988); Bulgac et al. (2006); Drut and Nicholson (2013). All observables are calculated with two number projections for fixed and fixed with the exception of the spin susceptibility. The spin susceptibility uses only one number projection for total particle number as discussed in Section IV.4.
III.4 Monte Carlo sign
The spin-balanced Fermi gas with an attractive contact interaction has a good Monte Carlo sign, i.e., the canonical trace of is positive for each sample of the auxiliary fields. To show this, we note that in general
(14) |
For an attractive interaction, the propagators and are invariant under time reversal and both particle-number projected traces on the right-hand-side of Eq. (14) are real. For the spin-balanced system , we have and thus the canonical trace is positive for any sample .
III.5 Numerical stabilization
The calculation of the propagator matrix requires the multiplication of matrices. In this section we use the simplified notation where denotes the imaginary-time step propagator. The condition number of is defined by , where is a matrix norm Meyer (2000); Golub and Van Loan (2013) such as the largest eigenvalue of or the Frobenius norm,
(15) |
As becomes large and we approach low temperatures (large ), the condition number can become very large and the product is ill-conditioned, i.e., we lose relevant information regarding the separation of large and small scales.
This problem is resolved in AFMC calculations by decomposing the propagator into a singular-value decomposition (SVD) Loh et al. (1989); Loh Jr and Gubernatis (1992); Loh et al. (2005); Gubernatis et al. (2016), or a QDR decomposition, in order to keep the separation of scales in a diagonal matrix and avoid the loss of information from mixing scales. In both cases, the matrix is decomposed into a product , where is a diagonal matrix with the different scales and and are well-conditioned matrices Loh Jr and Gubernatis (1992); Koonin (1997); Gilbreth and Alhassid (2015); Gubernatis et al. (2016). This can be described schematically by
(16) |
where the size of the characters represents the numerical scales.
In our implementation, we use a QDR decomposition, i.e., , where is a unitary matrix, is upper right triangular, and is diagonal. To diagonalize , one cannot multiply out the factors in this order without losing information contained in the smaller scales of . Using the standard method to calculate stabily, which involves a QDR decomposition for each Fourier sum component in the canonical trace, would lead to scaling with system size of (as opposed to for grand canonical calculations). This is costly as the number of single-particle states grows larger. Instead, one can diagonalize the matrix obtained from by a similarity transformation, and from this calculate number-projected observables. This procedure reduces the number of operations required for numerically stabilized canonical-ensemble calculations from (without diagonalization) to Gilbreth and Alhassid (2015). This method is used in our code. Recently, a new algorithm in the canonical ensemble was introduced in Ref. Shen et al. (2023) using a recursive method which was shown to be numerically stable. It will be interesting to compare both methods in future work.
III.6 Model space truncation
In our simulations we use the model space truncation method first introduced in Ref. Jensen et al. (2019) and presented in detail in Ref. Gilbreth et al. (2021). This allows for substantial savings in computational time.
The main idea of the method is to reduce the dimension of the one-body propagator in the decomposition from to , where , the number of significantly occupied single-particle states, is of the order of particle number . The entries of the diagonal matrix can be sorted in an appropriate descending order and decompose , where is obtained by zeroing out the diagonal elements for row (and the discarded terms are in ). This leads to
(17) |
where represents the occupied part of and is a small perturbation. All observables are then calculated using the reduced-rank matrix . The error made is represented by the matrix with norm . Our algorithm implementation uses the Frobenius norm and we choose such that the error matrix satisfies , where is a given small parameter. Ref. Gilbreth et al. (2021) provides an error analysis and further details of the truncation method.
The benefit of the truncation is to replace the original time complexity of computing the partition function for a given field configuration of Eq. 12 in the complete single-particle model space with complexity. This reduction in complexity becomes particularly large at low temperatures where is comparable to particle number . Further optimizations to reduce the complexity of the decompositions were introduced in the work of Ref. He et al. (2019) and applied in grand-canonical simulations of the two-dimensional BCS-BEC crossover He et al. (2022). These additional optimizations will be useful for reaching even larger lattice sizes and, in particular, lower temperatures in future studies of the UFG.
IV Results and analysis
In Ref. Jensen et al. (2020a) we performed simulations for and particles for a filling factor of . The finite filling factor introduces a systematic error as discussed in Sec. II.4. In Ref. Jensen et al. (2020b), we eliminated this systematic error when calculating the temperature dependence of Tan’s contact across the superfluid phase transition for and particles. This was accomplished by extrapolating to the continuum limit using results obtained for lattice sizes of and . Here we extend the simulations of Ref. Jensen et al. (2020b) to extrapolate to the continuum limit for several other thermodynamic observables including the condensate fraction, the energy-staggering pairing gap, the spin susceptibility, and the free energy-staggering pairing gap. We also performed simulations for particles for temperatures within the critical region to reduce the statistical error of our estimate for the critical temperature calculated by finite-size scaling.
In our calculations, for each particle number , temperature , and lattice size , we also carry out simulations with several (typically four) values of the time slice and extrapolate to the continuous imaginary-time limit .
In Appendix A, we discuss further details of the extrapolations to continuous time and to the continuum limit and present typical extrapolation plots for the condensate fraction and free energy staggering pairing gap.
IV.1 Condensate Fraction
We determine the superfluid critical temperature by analyzing the condensate fraction as a function of temperature and particle number in the continuum limit of the lattice model. We define the condensate fraction by
(18) |
where is the largest eigenvalue of the two-body density matrix , and is the maximal number of pairs. For fermions, can be shown to satisfy
(19) |
where is the number of lattice points. Having a large eigenvalue that scales with the system size is referred to as off-diagonal long-range order (ODLRO) Yang (1962).
In Fig. 1 we show the condensate fraction (after the continuous time and continuum extrapolations discussed in Appendix A.1) as a function of temperature for (solid blue squares), (solid orange circles), and (solid green down triangles) particles. We also show results for particles (solid red diamonds) in the vicinity of the critical temperature. There is a noticeable difference between the results of Ref. Jensen et al. (2020a) and our current continuum results. This is expected and demonstrates the importance of the continuum limit extrapolations in obtaining reliable results from lattice simulations. We also show the quantum Monte Carlo result of Ref. Astrakharchik et al. (2005) (open orange pentagon) where a diffusion Monte Carlo calculation was carried out for the full BCS to BEC crossover, and the quantum Monte Carlo (QMC) result of Ref. He et al. (2020) (purple cross) where a more complex lattice interaction is introduced to accelerate the convergence to the continuum limit. We compare with the recent low-temperature Josephson current experimental results of Ref. Kwon et al. (2020) (open up yellow triangle) and the finite temperature rapid ramp experimental result of Ref. Ku et al. (2012) (open black circles).
The grey band represents our estimate for the critical temperature from a finite-size scaling analysis (see Sec. IV.2). We observe that the condensate fraction decreases with particle number in the normal phase as expected when the thermodynamic limit is approached in a finite-size system, but it does not show a noticeable trend with below the critical temperature. Our results for the condensate fraction are larger than the rapid ramp experimental results of Ref. Ku et al. (2012) for all temperatures. Comparing our results with the results of other works, we find good agreement with the diffusion Monte Carlo result of Ref. Astrakharchik et al. (2005) (open orange pentagon) and with the experimental result of Ref. Kwon et al. (2020) (open up yellow triangle). Our low-temperature results are slightly higher than the improved lattice interaction result of Ref. He et al. (2020) (purple cross).
IV.2 Finite-size scaling and critical temperature
To determine the superfluid critical temperature we apply a finite-size scaling analysis to the condensate fraction for and (the simulations were performed only in the critical regime to improve our estimate of ). In Refs. Burovski et al. (2006a, b); Akkineni et al. (2007); Goulko and Wingate (2010) a finite-size scaling analysis, discussed in Ref. Binder (1981), was applied to the 3D XY phase transition of the UFG to determine the critical temperature. The critical behavior of the continuum limit condensate fraction is captured by the scaling ansatz
(20) |
where , the correlation length, is a universal scaling function, and is a non-universal coefficient of the term for the leading correction to scaling with an exponent . The exponents of the universality class have previously been calculated with , with for the exponent of the leading irrelevant field Guida and Zinn-Justin (1998); Campostrini et al. (2006); Pelissetto and Vicari (2002).
In Fig. 2 we plot the condensate fraction in the continuum limit scaled by as a function of for the different particle numbers. does not represent a simulated lattice size but the size of the physical system with particles. We arbitrarily choose for particles in Fig. 1(b) and the values for the remaining particle numbers are determined through the relation . If the corrections to the scaling ansatz are vanishing the crossings of the curves for different particle number scaled by will occur at the superfluid critical temperature . However, the corrections to the scaling relation can be significant for the finite particle numbers simulated in the current work. We follow Ref. Burovski et al. (2006a, b); Goulko and Wingate (2010) and expand to leading order ( is analytic at ), where the correlation length is in the critical regime where the reduced temperature is . We then obtain the relation between the crossing temperature for system sizes and , denoted by , and the thermodynamic critical temperature
(21) |
with
(22) |
Here absorbs constants in the expansion of the universal function . Refs. Burovski et al. (2006a, b) dropped the second term in the denominator of Eq. (22) to obtain a simplified relation
(23) |
In this work the data is not precise enough to perform the analysis of Ref Goulko and Wingate (2010) and instead we extract with the simplified crossings function . In the inset of Fig. 2 we show the crossing temperatures (open black circles) as a function of for and particles where the crossings are determined from straight line fits to the Monte Carlo data for each particle number . We have performed a generalized least square analysis to determine the critical temperature (solid black circle). This is larger than the estimate of Ref. Jensen et al. (2020a), where simulations were performed at a finite filling factor of . We observe that the critical temperature extracted from the AFMC data in the continuum limit is closer to the experimental value of Ref. Ku et al. (2012) and to the continuum limit diagrammatic Monte Carlo estimates of Ref. Burovski et al. (2006a) and of Ref. Goulko and Wingate (2010). The path integral Monte Carlo work of Ref. Akkineni et al. (2007) found a higher critical temperature . There are several other theoretical estimates for the critical temperature in Refs. Bulgac et al. (2006, 2008); Lee and Schäfer (2006); Nishida (2007); Haussmann et al. (2007); Abe and Seki (2009); Akkineni et al. (2007); Floerchinger et al. (2010); Boettcher et al. (2014). Experimental results for were obtained in the works of Refs. Ku et al. (2012); Nascimbene et al. (2010); Horikoshi et al. (2010). For reference we list the critical temperature estimates of various experiments and theoretical calculations in Table 1.
Method | error | |
---|---|---|
MIT experiment Ku et al. (2012) | 0.167 | 0.013 |
experiment Nascimbene et al. (2010) | 0.157 | 0.015 |
experiment Horikoshi et al. (2010) | 0.17 | 0.01 |
lattice DiagMC Burovski et al. (2006a) | 0.152 | 0.007 |
lattice DiagMC Goulko and Wingate (2010) | 0.173 | 0.006 |
lattice AFMC Richie-Halford et al. (2020) | 0.16 | 0.02 |
lattice AFMC Bulgac et al. (2008) | 0.15 | 0.01 |
lattice AFMC Bulgac et al. (2006) | 0.23 | 0.02 |
low-density neutron Abe and Seki (2009) | 0.189 | 0.012 |
RPIMC Akkineni et al. (2007) | ||
lattice Monte Carlo Lee and Schäfer (2006) | 0.14 | |
FRG Floerchinger et al. (2010) | 0.248 | |
-expansion matching from d=2,4 Nishida (2007) | 0.183 | 0.014 |
Luttinger-Ward Haussmann et al. (2007) | 0.183 | 0.014 |
lattice AFMC (this work) | 0.16 | 0.01 |
IV.3 Energy-staggering pairing gap
A signature of a pseudogap regime is the non-vanishing of a pairing gap above the critical temperature. We define a model-independent energy-staggering pairing gap at temperature for an even number of particles by
(24) |
where is the thermal energy for spin-up particles and spin-down particles. This observable was first calculated for the UFG at in Ref. Carlson et al. (2003) and at finite temperature in the works of Refs. Jensen et al. (2019, 2020a).
The calculation of at finite temperature requires the use of the canonical ensemble of fixed particle numbers. To this end, we use the particle-number reprojection method of Ref. Alhassid et al. (1999). The expectation value of the Hamiltonian for fixed particle numbers and is given by
(25) |
where
(26) |
is a positive definite weight function used for the Monte Carlo sampling, is the Monte Carlo sign, and
(27) |
is the expectation value of at fixed spin-up ad spin-down particle numbers for a given configuration of the auxiliary fields. We use exact particle-projection operators given by Eq. (11).
To calculate for different particle numbers , we use the original Monte Carlo sampling for particles but reproject on the new particle numbers Alhassid et al. (1999).
(28) |
where we have introduced the notation
(29) |
In Fig. 3(a) we show the continuum limit energy-staggering pairing gap (measured in units of the Fermi energy ) as a function of temperature for and particles. We observe that for the largest particle number (), the pairing gap vanishes above a temperature of . This suggests that the pseudogap regime, in which the pairing gap does not vanish above is considerably narrower than previously believed.
IV.4 Spin susceptibility
The spin susceptibility is suppressed by pairing correlations and is another signature of a pseudogap regime. It is given by
(30) |
where the expectation value on the r.h.s. is calculated for the spin-balanced system using only one particle projection on total number of particles in Eq. (30).
We calculated the spin susceptibility using the same Monte Carlo sampling at fixed numbers of spin-up and spin-down particles, i.e., according to defined in Eq. (26). We then have
(31) |
where is the expectation value of at fixed total particle number for a given field configuration .
In Fig. 3(b) we show the continuum limit spin susceptibility (in units of the free Fermi gas susceptibility ) as a function of for particle numbers and . We observe the strong suppression of below the critical temperature due to pairing correlations. However, we also observe moderate suppression above . While there is a clear particle number dependence and we have not reached the thermodynamic limit, we can put an upper bound on a spin gap temperature below which starts to become suppressed. We observe that the value of the spin gap temperature is similar to that of .
Based on our results in Fig. 3 for the energy-staggering pairing gap and spin susceptibility, we estimate the pseudogap regime above to be below .
IV.5 Free energy gap and extraction of the gap
The energy-staggering pairing gap is noisy, which makes the determination of an appropriate cutoff in filling factor for the continuum extrapolation rather difficult. A much less noisy quantity to investigate pairing correlations is the free energy staggering pairing gap, which is defined by a formula analogous to Eq. (24) but with the thermal energy replaced by the free energy
(32) |
Here and are, respectively, the free energy and partition function for spin-up particles and spin-down particles.
Within the AFMC framework, this observable is calculated using the particle-number reprojection method. We rewrite the free energy gap as
(33) |
Each partition function ratio can then be calculated using particle-number reprojection, e.g.,
(34) |
where is positive definite for .
In Fig. 4 we show the AFMC free energy gap (in units of the Femi gas energy ) as a function of temperature for and particles. The statistical errors on the free energy gap are small and this quantity can thus be determined more accurately than the energy-staggering pairing gap. We see that the free energy gap is suppressed above the critical temperature . To see more clearly the behavior of above , we plot in the inset the free energy gap on a log-linear scale. We observe that as we decrease the temperature from above , the free energy gap starts to increase below the pairing temperature scale that we identified in the energy-staggering pairing gap and the spin susceptibility in Fig. 3. Extrapolation to the thermodynamic limit will be important in future works to determine whether the pseudogap temperature scale is robust or will go down, reducing further the extent of the UFG pseudogap regime.
The free energy gap can be used to obtain a more precise value for the energy-staggering pairing gap as follows. We write as , where is the entropy gap. Both and saturate at low temperatures leading to a linear dependence of on at low temperatures. In Fig. 4 we show linear extrapolations of the free energy gap to for (blue dashed line) and (pink dashed line). We then estimate the energy-staggering pairing gap by taking the average of the (solid blue up triangle) and (solid red up triangle) extrapolations to find . We compare with the QMC results of Ref. Carlson and Reddy (2005) (open orange circle), Ref. Carlson and Reddy (2008) (open blue square), Ref. Lee et al. (2011) (solid yellow up triangle), and the experimental results of Ref. Schirotzek et al. (2008) (open down green triangle) and Ref. Hoinka et al. (2017) (open up black triangle). For reference, in Table 2 we list theoretical and experimental results for the pairing gap of the UFG.
Method | error | |
---|---|---|
MIT low-T experiment Schirotzek et al. (2008) | 0.44 | 0.03 |
Swinburne low-T experiment Hoinka et al. (2017) | 0.47 | 0.03 |
Ground-state fixed-node Monte Carlo Chang et al. (2004) | 0.594 | 0.024 |
Ground-state AFMC Carlson et al. (2003) | 0.55 | 0.05 |
Ground-state AFMC Carlson and Reddy (2005) | 0.50 | 0.03 |
QMC + MIT experiment Carlson and Reddy (2008) | 0.45 | 0.05 |
Luttinger-Ward Haussmann et al. (2007) | 0.46 | |
Functional renormalization group Bartosch et al. (2009) | 0.61 | |
Functional renormalization group Floerchinger et al. (2010) | 0.46 | |
Lattice quantum Monte Carlo Lee et al. (2011) | 0.52 | 0.01 |
AFMC (this work) | 0.576 | 0.024 |
V Conclusion and outlook
There has been substantial progress in recent years towards understanding the BCS-BEC crossover and the UFG in the two-species Fermi gas with short-range interactions . Experimental advances have provided an understanding of several observables for this strongly correlated system DeMarco and Jin (1999); Modugno et al. (2002); Truscott et al. (2001); Schreck et al. (2001); Granade et al. (2002); Hadzibabic et al. (2002); Jochim et al. (2003); Zwierlein et al. (2005); Navon et al. (2010); Ku et al. (2012); Carcy et al. (2019); Mukherjee et al. (2019). However, from a theoretical perspective, there is no controlled analytic approach to the UFG. The current theoretical approaches that are controllable (and scale to study large system sizes) are quantum Monte Carlo methods.
In this work, we calculated the first lattice continuum limit of the condensate fraction at finite temperature for the homogeneous UFG. The continuum limit extrapolations require calculations for large lattices and they were made possible through the development of canonical-ensemble AFMC optimizations in Refs. Gilbreth et al. (2021); Gilbreth and Alhassid (2015). Our low-temperature results for the condensate fraction show remarkable agreement with the QMC result of Ref. Astrakharchik et al. (2005). Using our finite-temperature results for the condensate fraction for multiple particle numbers , we used finite-size scaling to determine the superfluid critical temperature. Our result is in agreement with experiment Ku et al. (2012) and with the QMC results of Refs. Burovski et al. (2006a); Goulko and Wingate (2010).
To address open questions regarding the existence of a pseudogap regime and its extent in the UFG, we calculated a model-independent energy-staggering pairing gap and the spin susceptibility in the continuum limit. The calculation of the energy-staggering pairing gap was made possible by the use of a canonical-ensemble AFMC formulation of fixed particle numbers. The pairing gap and spin susceptibility can be used to identify signatures of a pseudogap regime above and below the pairing temperature scale . From the largest particle number simulations in the continuum limit we find an upper bound of . Thus we conclude that the pseudogap regime in the UFG is narrow. Compared with our previous finite filling factor study in Ref. Jensen et al. (2020a), we observe that in the continuum limit the pairing signatures increased for fixed temperature and particle number . This is consistent with the results of Ref. Gezerlis and Carlson (2008) where it was shown at that a finite effective range interaction suppresses attractive pairing correlations.
We also calculated the free energy staggering pairing gap and found it to have significantly reduced statistical errors compared with the energy-staggering pairing gap. Our results for this observable supports a pairing temperature scale of . In addition, extrapolating the low-temperature results for the free energy gap to zero temperature, we obtained an accurate estimate for the zero-temperature pairing gap of . Such extrapolation is not possible with the energy-staggering pairing gap because of its much larger statistical errors.
While we have reached the continuum limit for the first time for several thermodynamic observables that are useful signatures of pairing correlations for particle number as large as , we have not quite reached the thermodynamic limit of large particle number for the pairing gap and the spin susceptibility. It would be interesting to investigate the thermodynamic limit of these observables in future works.
Acknowledgements.
This work of was supported in part by the U. S. DOE grants Nos. DE-SC0019521, DE-SC0020177, and DE-FG02-00ER41132. The calculations presented here used resources of the National Energy Research Scientific Computing Center (NERSC), a U.S. Department of Energy Office of Science User Facility operated under Contract No. DE-AC02-05CH11231. We also thank the Yale Center for Research Computing for guidance and use of the research computing infrastructure.Appendix A Continuous time and continuum extrapolations
We demonstrate the continuous time and continuum extrapolations for the condensate fraction in Sec. A.1. Typical continuum extrapolations for the free energy gap are shown in Sec. A.2.
A.1 Condensate fraction
Extrapolations to continuous time.— The finite imaginary time slice introduces a systematic error in the AFMC calculations. As discussed in Sec. III.1, using a symmetric Trotter decomposition and a three-point Gaussian quadrature leads to a discretization error of . We extrapolate by a linear fit in for sufficiently small values of .
In Fig. 5 we show representative extrapolations of the condensate fraction for particles and lattice size of at three temperatures (a) , (b) , and (c) . The slope of the fit increases dramatically in magnitude as temperature is reduced, so the extrapolations become necessary at low temperatures (see for and ). We performed these extrapolations for all particle numbers and lattice sizes.
Continuum extrapolations.— After carrying out the extrapolations, we perform the continuum limit extrapolation using lattices of increasing size with for each temperature and particle number .
In Fig. 6 we show the condensate fraction as a function of for particle numbers and for three temperatures in the critical regime near . We perform linear fits in using the range . We use this range in the continuum extrapolations of the condensate fraction for all temperatures and particle numbers. We note that the behavior of for larger filling factors is significantly different for lower temperatures. At the lower temperature (upper panels of Fig. 6), the largest filling factor results are below the continued straight line fit (dashed line) whereas for higher temperatures (middle and lower panels), the larger filling factor results are above the continued fit. The continuum extrapolation provides the final estimate for the condensate fraction for a given particle number and temperature .
A.2 Free energy gap
In this section, we demonstrate the continuum limit () extrapolations for the free energy gap . These continuum extrapolations are carried out after the continuous time extrapolations .
References
- Leggett (2006) A. J. Leggett, Quantum liquids: Bose condensation and Cooper pairing in condensed-matter systems (Oxford University Press, 2006).
- Bloch et al. (2008) I. Bloch, J. Dalibard, and W. Zwerger, Rev. Mod. Phys. 80, 885 (2008).
- Giorgini et al. (2008) S. Giorgini, L. P. Pitaevskii, and S. Stringari, Rev. Mod. Phys. 80, 1215 (2008).
- Inguscio et al. (2008) M. Inguscio, W. Ketterle, and C. Salomon, Ultra-cold Fermi gases, Vol. 164 (IOS press, 2008).
- Zwerger (2012) W. Zwerger, The BCS-BEC Crossover and the Unitary Fermi Gas (Springer-Verlag, Heidelberg, 2012).
- Levin et al. (2012) K. Levin, A. Fetter, and D. Stamper-Kurn, Ultracold bosonic and fermionic gases (Elsevier, 2012).
- Svistunov et al. (2015) B. V. Svistunov, E. S. Babaev, and N. V. Prokof’ev, Superfluid states of matter (Crc Press, 2015).
- Pitaevskii and Stringari (2016) L. Pitaevskii and S. Stringari, Bose-Einstein condensation and superfluidity, Vol. 164 (Oxford University Press, 2016).
- Bardeen et al. (1957) J. Bardeen, L. N. Cooper, and J. R. Schrieffer, Phys. Rev. 108, 1175 (1957).
- Eagles (1969) D. M. Eagles, Phys. Rev. 186, 456 (1969).
- Leggett (1980) A. J. Leggett, “Diatomic molecules and Cooper pairs,” in Modern Trends in the Theory of Condensed Matter: Proceedings of the XVI Karpacz Winter School of Theoretical Physics, February 19 – March 3, 1979 Karpacz, Poland, edited by A. Pekalski and J. A. Przystawa (Springer Berlin Heidelberg, Berlin, Heidelberg, 1980) pp. 13–27.
- Nozières and Schmitt-Rink (1985) P. Nozières and S. Schmitt-Rink, Journal of Low Temperature Physics 59, 195 (1985).
- Randeria (2010) M. Randeria, Nature Physics 6, 561 (2010).
- Enss (2012) T. Enss, Phys. Rev. A 86, 013616 (2012).
- Bluhm et al. (2017) M. Bluhm, J. Hou, and T. Schäfer, Phys. Rev. Lett. 119, 065302 (2017).
- Nishida and Abuki (2005) Y. Nishida and H. Abuki, Phys. Rev. D 72, 096004 (2005).
- Ketterle and Zwierlein (2008) W. Ketterle and M. W. Zwierlein, “Making, probing and understanding ultracold fermi gases,” in Proceedings of the International School of Physics “Enrico Fermi” - Course 164 “Ultra-cold Fermi Gases”, edited by M. Inguscio, W. Ketterle, and C. Salomon (IOS Press, Amsterdam, 2008) pp. 95–287.
- Hu et al. (2022) H. Hu, X.-C. Yao, and X.-J. Liu, AAPPS Bulletin 32, 26 (2022).
- Nishida and Son (2006) Y. Nishida and D. T. Son, Phys. Rev. Lett. 97, 050403 (2006).
- Haussmann et al. (2007) R. Haussmann, W. Rantner, S. Cerrito, and W. Zwerger, Phys. Rev. A 75, 023610 (2007).
- Haussmann et al. (2009) R. Haussmann, M. Punk, and W. Zwerger, Phys. Rev. A 80, 063612 (2009).
- Zwerger (2016) W. Zwerger, “Strongly interacting fermi gases,” in Proceedings of the International School of Physics “Enrico Fermi” - Course 191 “Quantum Matter at Ultralow Temperatures”, edited by M. Inguscio, W. Ketterle, S. Stringari, and G. Roati (IOS Press, Amsterdam, SIF Bologna, 2016) pp. 63–141.
- Frank et al. (2018) B. Frank, J. Lang, and W. Zwerger, JETP 127, 812 (2018).
- Nishida and Son (2007a) Y. Nishida and D. T. Son, Phys. Rev. D 76, 086004 (2007a).
- Nishida and Son (2007b) Y. Nishida and D. T. Son, Phys. Rev. A 75, 063617 (2007b).
- Nikolić and Sachdev (2007) P. Nikolić and S. Sachdev, Phys. Rev. A 75, 033608 (2007).
- Perali et al. (2002a) A. Perali, P. Pieri, G. C. Strinati, and C. Castellani, Phys. Rev. B 66, 024510 (2002a).
- Ohashi and Griffin (2002) Y. Ohashi and A. Griffin, Phys. Rev. Lett. 89, 130402 (2002).
- Chen et al. (2005) Q. Chen, J. Stajic, S. Tan, and K. Levin, Physics Reports 412, 1 (2005).
- Gilbreth and Alhassid (2013) C. N. Gilbreth and Y. Alhassid, Phys. Rev. A 88, 063643 (2013).
- Bulgac et al. (2006) A. Bulgac, J. E. Drut, and P. Magierski, Phys. Rev. Lett. 96, 090404 (2006).
- Bulgac et al. (2008) A. Bulgac, J. E. Drut, and P. Magierski, Phys. Rev. A 78, 023625 (2008).
- Drut et al. (2012) J. E. Drut, T. A. Lähde, G. Wlazłowski, and P. Magierski, Phys. Rev. A 85, 051601(R) (2012).
- Akkineni et al. (2007) V. K. Akkineni, D. M. Ceperley, and N. Trivedi, Phys. Rev. B 76, 165116 (2007).
- Burovski et al. (2006a) E. Burovski, N. Prokof’ev, B. Svistunov, and M. Troyer, Phys. Rev. Lett. 96, 160402 (2006a).
- Burovski et al. (2006b) E. Burovski, N. Prokof’ev, B. Svistunov, and M. Troyer, New Journal of Physics 8, 153 (2006b).
- Burovski et al. (2008) E. Burovski, E. Kozik, N. Prokof’ev, B. Svistunov, and M. Troyer, Phys. Rev. Lett. 101, 090402 (2008).
- Goulko and Wingate (2010) O. Goulko and M. Wingate, Phys. Rev. A 82, 053621 (2010).
- Van Houcke et al. (2011) K. Van Houcke, F. Werner, E. Kozik, N. Prokofev, B. Svistunov, M. Ku, A. Sommer, L. Cheuk, A. Schirotzek, and M. Zwierlein, arXiv preprint arXiv:1110.3747 (2011).
- Hu et al. (2010) H. Hu, X.-J. Liu, and P. D. Drummond, New Journal of Physics 12, 063038 (2010).
- Lee and Schäfer (2006) D. Lee and T. Schäfer, Phys. Rev. C 73, 015202 (2006).
- Endres et al. (2013) M. G. Endres, D. B. Kaplan, J.-W. Lee, and A. N. Nicholson, Phys. Rev. A 87, 023615 (2013).
- Regal et al. (2004) C. A. Regal, M. Greiner, and D. S. Jin, Phys. Rev. Lett. 92, 040403 (2004).
- Zwierlein et al. (2004) M. W. Zwierlein, C. A. Stan, C. H. Schunck, S. M. F. Raupach, A. J. Kerman, and W. Ketterle, Phys. Rev. Lett. 92, 120403 (2004).
- Kinast et al. (2004) J. Kinast, S. L. Hemmer, M. E. Gehm, A. Turlapov, and J. E. Thomas, Phys. Rev. Lett. 92, 150402 (2004).
- Chin et al. (2004) C. Chin, M. Bartenstein, A. Altmeyer, S. Riedl, S. Jochim, J. H. Denschlag, and R. Grimm, Science 305, 1128 (2004).
- Zwierlein et al. (2005) M. W. Zwierlein, C. H. Schunck, C. A. Stan, S. M. F. Raupach, and W. Ketterle, Phys. Rev. Lett. 94, 180401 (2005).
- Nascimbene et al. (2010) S. Nascimbene, N. Navon, K. Jiang, F. Chevy, and C. Salomon, Nature 463, 1057 (2010).
- Nascimbène et al. (2011) S. Nascimbène, N. Navon, S. Pilati, F. Chevy, S. Giorgini, A. Georges, and C. Salomon, Phys. Rev. Lett. 106, 215303 (2011).
- Ku et al. (2012) M. J. H. Ku, A. T. Sommer, L. W. Cheuk, and M. W. Zwierlein, Science 335, 563 (2012).
- Mukherjee et al. (2017) B. Mukherjee, Z. Yan, P. B. Patel, Z. Hadzibabic, T. Yefsah, J. Struck, and M. W. Zwierlein, Phys. Rev. Lett. 118, 123401 (2017).
- Shin et al. (2007) Y. Shin, C. H. Schunck, A. Schirotzek, and W. Ketterle, Phys. Rev. Lett. 99, 090403 (2007).
- Gezerlis and Carlson (2008) A. Gezerlis and J. Carlson, Phys. Rev. C 77, 032801(R) (2008).
- Carlson et al. (2003) J. Carlson, S.-Y. Chang, V. R. Pandharipande, and K. E. Schmidt, Phys. Rev. Lett. 91, 050401 (2003).
- Chang et al. (2004) S. Y. Chang, V. R. Pandharipande, J. Carlson, and K. E. Schmidt, Phys. Rev. A 70, 043602 (2004).
- Carlson and Reddy (2005) J. Carlson and S. Reddy, Phys. Rev. Lett. 95, 060401 (2005).
- Carlson and Reddy (2008) J. Carlson and S. Reddy, Phys. Rev. Lett. 100, 150403 (2008).
- Carlson et al. (2011) J. Carlson, S. Gandolfi, K. E. Schmidt, and S. Zhang, Phys. Rev. A 84, 061602(R) (2011).
- Kashimura et al. (2012a) T. Kashimura, R. Watanabe, and Y. Ohashi, Phys. Rev. A 86, 043622 (2012a).
- Kashimura et al. (2012b) T. Kashimura, R. Watanabe, and Y. Ohashi, Phys. Rev. A 86, 043622 (2012b).
- van Wyk et al. (2016) P. van Wyk, H. Tajima, R. Hanai, and Y. Ohashi, Phys. Rev. A 93, 013621 (2016).
- Perali et al. (2002b) A. Perali, P. Pieri, G. C. Strinati, and C. Castellani, Phys. Rev. B 66, 024510 (2002b).
- Perali et al. (2011a) A. Perali, F. Palestini, P. Pieri, G. C. Strinati, J. T. Stewart, J. P. Gaebler, T. E. Drake, and D. S. Jin, Phys. Rev. Lett. 106, 060402 (2011a).
- Palestini et al. (2012a) F. Palestini, P. Pieri, and G. C. Strinati, Phys. Rev. Lett. 108, 080401 (2012a).
- Tajima et al. (2016) H. Tajima, R. Hanai, and Y. Ohashi, Phys. Rev. A 93, 013610 (2016).
- Pini et al. (2019) M. Pini, P. Pieri, and G. C. Strinati, Phys. Rev. B 99, 094502 (2019).
- Perali et al. (2011b) A. Perali, F. Palestini, P. Pieri, G. C. Strinati, J. T. Stewart, J. P. Gaebler, T. E. Drake, and D. S. Jin, Phys. Rev. Lett. 106, 060402 (2011b).
- Palestini et al. (2012b) F. Palestini, P. Pieri, and G. C. Strinati, Phys. Rev. Lett. 108, 080401 (2012b).
- Pantel et al. (2014) P.-A. Pantel, D. Davesne, and M. Urban, Phys. Rev. A 90, 053629 (2014).
- Tajima et al. (2014) H. Tajima, T. Kashimura, R. Hanai, R. Watanabe, and Y. Ohashi, Phys. Rev. A 89, 033617 (2014).
- Santos (1994) R. R. dos Santos, Phys. Rev. B 50, 635 (1994).
- Jankó et al. (1997) B. Jankó, J. Maly, and K. Levin, Phys. Rev. B 56, R11407 (1997).
- Sewer et al. (2002) A. Sewer, X. Zotos, and H. Beck, Phys. Rev. B 66, 140504(R) (2002).
- Chien et al. (2010) C.-C. Chien, H. Guo, Y. He, and K. Levin, Phys. Rev. A 81, 023622 (2010).
- Chien and Levin (2010) C.-C. Chien and K. Levin, Phys. Rev. A 82, 013603 (2010).
- Wlazłowski et al. (2013a) G. Wlazłowski, P. Magierski, J. E. Drut, A. Bulgac, and K. J. Roche, Phys. Rev. Lett. 110, 090401 (2013a).
- Magierski et al. (2009) P. Magierski, G. Wlazłowski, A. Bulgac, and J. E. Drut, Phys. Rev. Lett. 103, 210403 (2009).
- Magierski et al. (2011) P. Magierski, G. Wlazłowski, and A. Bulgac, Phys. Rev. Lett. 107, 145304 (2011).
- Enss and Haussmann (2012) T. Enss and R. Haussmann, Phys. Rev. Lett. 109, 195303 (2012).
- Jensen et al. (2020a) S. Jensen, C. N. Gilbreth, and Y. Alhassid, Phys. Rev. Lett. 124, 090604 (2020a).
- Gaebler et al. (2010) J. P. Gaebler, J. T. Stewart, T. E. Drake, D. S. Jin, A. Perali, P. Pieri, and G. C. Strinati, Nature Physics 6, 569 (2010).
- Sommer et al. (2011) A. Sommer, M. Ku, G. Roati, and M. W. Zwierlein, Nature 472, 201 (2011).
- Wulin et al. (2011) D. Wulin, H. Guo, C.-C. Chien, and K. Levin, Phys. Rev. A 83, 061601(R) (2011).
- Chen and Wang (2014) Q. Chen and J. Wang, Frontiers of Physics 9, 539 (2014).
- Mueller (2017) E. J. Mueller, Rep. Prog. Phys. 80, 104401 (2017).
- Jensen et al. (2019) S. Jensen, C. N. Gilbreth, and Y. Alhassid, Eur. Phys. J. Spec. Top. 227, 2241 (2019).
- Khuri et al. (2009) N. N. Khuri, A. Martin, J.-M. Richard, and T. T. Wu, Journal of Mathematical Physics 50, 072105 (2009).
- Castin and Werner (2012) Y. Castin and F. Werner, “The unitary gas and its symmetry properties,” in The BCS-BEC Crossover and the Unitary Fermi Gas, edited by W. Zwerger (Springer Berlin Heidelberg, Berlin, Heidelberg, 2012) pp. 127–191.
- Werner and Castin (2012) F. Werner and Y. Castin, Phys. Rev. A 86, 013626 (2012).
- Chin et al. (2010) C. Chin, R. Grimm, P. Julienne, and E. Tiesinga, Rev. Mod. Phys. 82, 1225 (2010).
- Dagotto (1994) E. Dagotto, Rev. Mod. Phys. 66, 763 (1994).
- Timusk and Statt (1999) T. Timusk and B. Statt, Reports on Progress in Physics 62, 61 (1999).
- Fetter and Walecka (1971) A. Fetter and J. D. Walecka, Quantum theory of many-particle systems (McGraw-Hill, 1971).
- Palestini et al. (2012c) F. Palestini, A. Perali, P. Pieri, and G. C. Strinati, Phys. Rev. B 85, 024517 (2012c).
- Pini et al. (2020) M. Pini, P. Pieri, M. Jäger, J. H. Denschlag, and G. C. Strinati, New Journal of Physics 22, 083008 (2020).
- Pricoupenko and Castin (2007) L. Pricoupenko and Y. Castin, Journal of Physics A: Mathematical and Theoretical 40, 12863 (2007).
- Wlazłowski et al. (2012) G. Wlazłowski, P. Magierski, and J. E. Drut, Phys. Rev. Lett. 109, 020406 (2012).
- Wlazłowski et al. (2013b) G. Wlazłowski, P. Magierski, A. Bulgac, and K. J. Roche, Phys. Rev. A 88, 013639 (2013b).
- Wlazłowski et al. (2015) G. Wlazłowski, W. Quan, and A. Bulgac, Phys. Rev. A 92, 063628 (2015).
- Drut et al. (2011) J. E. Drut, T. A. Lähde, and T. Ten, Phys. Rev. Lett. 106, 205302 (2011).
- Richie-Halford et al. (2020) A. Richie-Halford, J. E. Drut, and A. Bulgac, Phys. Rev. Lett. 125, 060403 (2020).
- Gilbreth et al. (2021) C. Gilbreth, S. Jensen, and Y. Alhassid, Computer Physics Communications 264, 107952 (2021).
- Drut and Nicholson (2013) J. E. Drut and A. N. Nicholson, Journal of Physics G: Nuclear and Particle Physics 40, 043101 (2013).
- Hubbard (1959) J. Hubbard, Phys. Rev. Lett. 3, 77 (1959).
- Stratonovich (1957) R. L. Stratonovich, Dokl. Akad. Nauk SSSR [Sov. Phys. - Dokl.] 115, 1097 (1957).
- Metropolis et al. (1953) N. Metropolis, A. W. Rosenbluth, M. N. Rosenbluth, A. H. Teller, and E. Teller, The journal of chemical physics 21, 1087 (1953).
- Ormand et al. (1994) W. E. Ormand, D. J. Dean, C. W. Johnson, G. H. Lang, and S. E. Koonin, Phys. Rev. C 49, 1422 (1994).
- Batrouni et al. (1986) G. G. Batrouni, A. Hansen, and M. Nelkin, Phys. Rev. Lett. 57, 1336 (1986).
- Davies et al. (1988) C. T. H. Davies, G. G. Batrouni, G. R. Katz, A. S. Kronfeld, G. P. Lepage, K. G. Wilson, P. Rossi, and B. Svetitsky, Phys. Rev. D 37, 1581 (1988).
- Meyer (2000) C. D. Meyer, Matrix analysis and applied linear algebra, Vol. 71 (Siam, 2000).
- Golub and Van Loan (2013) G. H. Golub and C. F. Van Loan, Matrix computations (JHU press, 2013).
- Loh et al. (1989) E. Loh, J. Gubernatis, R. Scalettar, R. Sugar, and S. White, in Interacting Electrons in Reduced Dimensions (Springer, 1989) pp. 55–60.
- Loh Jr and Gubernatis (1992) E. Y. Loh Jr and J. E. Gubernatis, in Electronic phase transitions (Modern Problems in Condensed Matter Sciences), edited by W. Hanke and Y. Kopaev (North-Holland, 1992) pp. 177–235.
- Loh et al. (2005) E. Loh, J. Gubernatis, R. Scalettar, S. White, D. Scalapino, and R. Sugar, International Journal of Modern Physics C 16, 1319 (2005).
- Gubernatis et al. (2016) J. Gubernatis, N. Kawashima, and P. Werner, Quantum Monte Carlo Methods: Algorithms for Lattice Models (Cambridge University Press, 2016).
- Koonin (1997) S. E. Koonin, “Shell model monte carlo methods,” in Contemporary Nuclear Shell Models: Proceedings of an International Workshop Held in Philadelphia, PA, USA, 29–30 April 1996, edited by X.-W. Pan, D. H. Feng, and M. Vallières (Springer Berlin Heidelberg, Berlin, Heidelberg, 1997) pp. 132–132.
- Gilbreth and Alhassid (2015) C. N. Gilbreth and Y. Alhassid, Computer Physics Communications 188, 1 (2015).
- Shen et al. (2023) T. Shen, H. Barghathi, J. Yu, A. Del Maestro, and B. M. Rubenstein, Phys. Rev. E 107, 055302 (2023).
- He et al. (2019) Y.-Y. He, H. Shi, and S. Zhang, Phys. Rev. Lett. 123, 136402 (2019).
- He et al. (2022) Y.-Y. He, H. Shi, and S. Zhang, Phys. Rev. Lett. 129, 076403 (2022).
- Jensen et al. (2020b) S. Jensen, C. N. Gilbreth, and Y. Alhassid, Phys. Rev. Lett. 125, 043402 (2020b).
- Yang (1962) C. N. Yang, Rev. Mod. Phys. 34, 694 (1962).
- Astrakharchik et al. (2005) G. E. Astrakharchik, J. Boronat, J. Casulleras, and S. Giorgini, Phys. Rev. Lett. 95, 230405 (2005).
- He et al. (2020) R. He, N. Li, B.-N. Lu, and D. Lee, Phys. Rev. A 101, 063615 (2020).
- Kwon et al. (2020) W. J. Kwon, G. D. Pace, R. Panza, M. Inguscio, W. Zwerger, M. Zaccanti, F. Scazza, and G. Roati, Science 369, 84 (2020), https://www.science.org/doi/pdf/10.1126/science.aaz2463 .
- Binder (1981) K. Binder, Phys. Rev. Lett. 47, 693 (1981).
- Guida and Zinn-Justin (1998) R. Guida and J. Zinn-Justin, Journal of Physics A: Mathematical and General 31, 8103 (1998).
- Campostrini et al. (2006) M. Campostrini, M. Hasenbusch, A. Pelissetto, and E. Vicari, Phys. Rev. B 74, 144506 (2006).
- Pelissetto and Vicari (2002) A. Pelissetto and E. Vicari, Physics Reports 368, 549 (2002).
- Nishida (2007) Y. Nishida, Phys. Rev. A 75, 063618 (2007).
- Abe and Seki (2009) T. Abe and R. Seki, Phys. Rev. C 79, 054003 (2009).
- Floerchinger et al. (2010) S. Floerchinger, M. M. Scherer, and C. Wetterich, Phys. Rev. A 81, 063619 (2010).
- Boettcher et al. (2014) I. Boettcher, J. M. Pawlowski, and C. Wetterich, Phys. Rev. A 89, 053630 (2014).
- Horikoshi et al. (2010) M. Horikoshi, S. Nakajima, M. Ueda, and T. Mukaiyama, Science 327, 442 (2010).
- Schirotzek et al. (2008) A. Schirotzek, Y.-i. Shin, C. H. Schunck, and W. Ketterle, Phys. Rev. Lett. 101, 140403 (2008).
- Hoinka et al. (2017) S. Hoinka, P. Dyke, M. G. Lingham, J. J. Kinnunen, G. M. Bruun, and C. J. Vale, Nature Physics 13, 943 (2017).
- Lee et al. (2011) J.-W. Lee, M. Endres, D. B. Kaplan, and A. Nicholson, PoS Lattice 2010, 197 (2011).
- Alhassid et al. (1999) Y. Alhassid, S. Liu, and H. Nakada, Phys. Rev. Lett. 83, 4265 (1999).
- Bartosch et al. (2009) L. Bartosch, P. Kopietz, and A. Ferraz, Phys. Rev. B 80, 104514 (2009).
- DeMarco and Jin (1999) B. DeMarco and D. S. Jin, Science 285, 1703 (1999).
- Modugno et al. (2002) G. Modugno, M. Modugno, F. Riboli, G. Roati, and M. Inguscio, Phys. Rev. Lett. 89, 190404 (2002).
- Truscott et al. (2001) A. G. Truscott, K. E. Strecker, W. I. McAlexander, G. B. Partridge, and R. G. Hulet, Science 291, 2570 (2001).
- Schreck et al. (2001) F. Schreck, L. Khaykovich, K. L. Corwin, G. Ferrari, T. Bourdel, J. Cubizolles, and C. Salomon, Phys. Rev. Lett. 87, 080403 (2001).
- Granade et al. (2002) S. R. Granade, M. E. Gehm, K. M. O’Hara, and J. E. Thomas, Phys. Rev. Lett. 88, 120405 (2002).
- Hadzibabic et al. (2002) Z. Hadzibabic, C. A. Stan, K. Dieckmann, S. Gupta, M. W. Zwierlein, A. Görlitz, and W. Ketterle, Phys. Rev. Lett. 88, 160401 (2002).
- Jochim et al. (2003) S. Jochim, M. Bartenstein, A. Altmeyer, G. Hendl, S. Riedl, C. Chin, J. H. Denschlag, and R. Grimm, Science 302, 2101 (2003).
- Navon et al. (2010) N. Navon, S. Nascimbène, F. Chevy, and C. Salomon, Science 328, 729 (2010).
- Carcy et al. (2019) C. Carcy, S. Hoinka, M. G. Lingham, P. Dyke, C. C. N. Kuhn, H. Hu, and C. J. Vale, Phys. Rev. Lett. 122, 203401 (2019).
- Mukherjee et al. (2019) B. Mukherjee, P. B. Patel, Z. Yan, R. J. Fletcher, J. Struck, and M. W. Zwierlein, Phys. Rev. Lett. 122, 203402 (2019).