Rigorous foundation and extension of Leggett's upper bound on the condensate fraction of a weakly repulsive Bose gas at positive temperature

Leggett (New J. Phys. 3 (2001) 23) derived a conditional upper bound on the condensate fraction $f(T)$ of the three-dimensional dilute Bose gas with weak repulsion: at the ideal-gas critical temperature, $f\le(3\eta/\alpha_0)^{1/3}$ with $\eta=nV_0/kT_c(0)$, $V_0=\int V\,d^3r$, and $\alpha_0\simeq0.3$, and above $T_c(0)$ a bound vanishing with the interaction strength. The bound rests on an unproved Gaussian extensivity assumption on the condensate-number distribution — known to fail for the ideal gas below $T_c$ — and assumes an everywhere-nonnegative potential and a homogeneous free-space geometry. The follow-up literature: Leggett's own 2003 extension to harmonic traps remains conditional on the same assumption; the rigorous positive-temperature free-energy program (Seiringer 2008; Yin 2010; Basti et al. 2025) still assumes nonnegative potentials and proves nothing about the condensate fraction; a rigorous but non-sharp upper bound on the critical temperature exists (Seiringer–Ueltschi 2009); and a 2025 review reports that no exact results exist on the fluctuation statistics of weakly interacting gases. The open core is intact: prove or weaken the fluctuation assumption for interacting gases and derive the corresponding rigorous bound on $f(T)$, with extensions beyond $V\ge0$ and to trapped geometries.

Background

Bose–Einstein condensation (BEC) is the phenomenon, expected in dilute gases of bosons (particles of integer spin) at sufficiently low temperature, in which a macroscopic fraction of all the particles occupies a single quantum state. For a gas in a box in the thermodynamic limit — $N$ particles in volume $\Omega$ with $N,\Omega\to\infty$ at fixed density $n=N/\Omega$ — the natural order parameter is the condensate fraction $f=\langle\hat N_0\rangle/N$, where $\hat N_0$ counts particles in the zero-momentum single-particle state and $\langle\cdot\rangle$ denotes the equilibrium expectation in the canonical ensemble (fixed $N$); equivalently, $f$ is the largest eigenvalue of the one-particle density matrix (the kernel encoding one-body correlations) divided by $N$. The ideal, noninteracting Bose gas condenses below a critical temperature $T_c(0)$ fixed by $n$ and the particle mass: $f>0$ for $T<T_c(0)$ and $f=0$ for $T\ge T_c(0)$.

For interacting gases, rigorous knowledge at positive temperature is remarkably sparse. Hohenberg (1967), through an inequality on one-particle correlations, excluded BEC at finite temperature in free space in one and two dimensions, and showed that in three dimensions the condensate fraction obeys $f^{2/3}/(1-f)\le\gamma(T_c(0)/T)$ with $\gamma\simeq2.3$ — a bound independent of the interaction strength and weak near $T_c(0)$, where it only forces $f$ below a value of order $0.8$. Perturbative calculations (Baym, Blaizot, Holzmann, Laloë and Vautherin, 1999) indicate that a repulsive interaction raises the transition temperature by an amount that tends to zero with the interaction strength. In this landscape, Leggett (2001) derived an upper bound on $f(T)$ for a three-dimensional gas of spinless bosons with an interaction potential $V(\boldsymbol r)\ge0$ everywhere, of integrated strength $V_0=\int V(\boldsymbol r)\,d^3r$, in the thermodynamic limit. Bounding the interacting free energy $F$ from above by the Peierls–Bogoliubov inequality (a variational comparison of the free energies of two Hamiltonians) and from below by a trial-state argument that strips the condensate out of the interacting density matrix, he obtained the implicit bound

$$F_0(N(1-f),\Omega,T)-F_0(N,\Omega,T)\;\le\;NnV_0,$$

where $F_0$ is the free energy of the corresponding noninteracting gas, and then the explicit consequence

$$f\le(3\eta/\alpha_0)^{1/3}\quad\text{at }T=T_c(0),\qquad\eta\equiv nV_0/kT_c(0),$$

with $\alpha_0\simeq0.3$ an explicit numerical constant, and $f\lesssim\alpha_0^{-1}\eta\,\epsilon^{-2}$ for $\eta^{1/3}\ll\epsilon\ll1$, where $\epsilon=1-(T_c(0)/T)^{3/2}$. Under the standard pseudopotential replacement $V_0\to4\pi\hbar^2a_s/m$ — where $a_s$ is the s-wave scattering length, the strength parameter of low-energy two-body scattering — the bound at $T_c(0)$ becomes proportional to $(na_s^3)^{1/9}$, vanishing with the interaction strength: qualitatively stronger than the Hohenberg bound near $T_c(0)$.

The derivation rests, however, on an unproven statistical assumption: that the condensate occupation $N_0$ has Gaussian tails at the extensivity scale,

$$P(N_0)\le\zeta\exp\bigl(-(N_0-\langle\hat N_0\rangle)^2/(KN)\bigr),\qquad\zeta,K=O(1),$$

i.e. condensate-number fluctuations of order $\sqrt N$. This is exactly what fails for the ideal gas below $T_c$, where condensate fluctuations are anomalously large — of amplitude $\sim N^{2/3}$ in a three-dimensional box with an explicitly non-Gaussian limit, as established rigorously by Chatterjee and Diaconis (2014) — a manifestation of the inequivalence of the canonical ensemble (fixed $N$) and the grand-canonical ensemble (fixed chemical potential). Interactions are expected to suppress the anomalous fluctuations back to the Gaussian scale, but no rigorous result, and per a 2025 review of the field no exact result at all, is known for any interacting gas. The paper additionally assumes $V\ge0$ (used through the Fourier bound $V_{k-k'}\le V_0$), notes that real atomic potentials are not everywhere positive, and leaves open whether the harmonic-trap geometry — where the condensed state need not be the lowest single-particle state of the noninteracting system — can be handled. Leggett further notes that a lower bound on the interacting potential energy relative to its Hartree–Fock value (the simplest mean-field decoupling of the interaction energy) could tighten the bound considerably, and expresses doubt that arguments of this form can bound the transition temperature $T_c$ of the interacting gas.

Problem Statement

Establish a rigorous upper bound on the condensate fraction $f(T)$ of the three-dimensional dilute Bose gas with weak repulsive interactions at positive temperature in the thermodynamic limit, of the strength derived conditionally by Leggett (2001): with $\eta=nV_0/kT_c(0)$, where $n$ is the particle density, $V_0=\int V(\boldsymbol r)\,d^3r$ is the integrated strength of the interaction potential $V$, and $T_c(0)$ is the critical temperature of the corresponding noninteracting gas (or, under a rigorously justified pseudopotential substitution, the equivalent parametrization by the s-wave scattering length $a_s$, the strength parameter of low-energy two-body scattering), $f$ is bounded at $T=T_c(0)$ by an expression of order $\eta^{1/3}$ (equivalently $(na_s^3)^{1/9}$), and above $T_c(0)$ by an expression vanishing with the interaction strength. The problem comprises three linked components, exactly the gaps the source leaves open, with one common scientific objective — a rigorous condensate-fraction bound for the interacting gas:

(a) The load-bearing statistical gap: prove or appropriately weaken the Gaussian extensivity assumption on the condensate-number distribution, $P(N_0)\le\zeta\exp\bigl(-(N_0-\langle\hat N_0\rangle)^2/(KN)\bigr)$ with $\zeta,K=O(1)$ in the thermodynamic limit, for the interacting dilute gas — it is false for the ideal gas below $T_c$; any weakening must be accompanied by the corresponding modified rigorous bound on $f(T)$. A refutation of the assumption would not by itself settle this component, since the target is the rigorous bound on $f(T)$, which in that case would remain to be established by other means.

(b) The potential-class gap: remove the hypothesis $V(\boldsymbol r)\ge0$ everywhere — either by extending the argument to realistic sign-changing atomic potentials, or by rigorously justifying the pseudopotential replacement $V_0\to4\pi\hbar^2a_s/m$ at the level of the condensate-fraction bound (not merely at the level of the free energy).

(c) The geometry gap: extend the bound to dilute gases in a harmonic trap, where condensation need not occur in the lowest single-particle state of the noninteracting system (Leggett's own 2003 trap extension remains conditional on the same unproved assumption as (a)).

Components (a) and (b) together settle the homogeneous-gas problem, and (c) extends it to the experimentally relevant trapped geometry. Two further directions raised in the source are auxiliary: a lower bound on the interacting potential energy relative to the Hartree–Fock value, which would tighten the bound, and the question — which the source itself considers unlikely to succeed — whether arguments of this form can also yield rigorous bounds on the transition temperature $T_c$ of the interacting gas, beyond the non-sharp upper bound of Seiringer–Ueltschi (2009).

References

1. A. J. Leggett, 'An upper bound on the condensate fraction in a Bose gas with weak repulsion', New J. Phys. 3 (2001) 23, DOI: 10.1088/1367-2630/3/1/323, https://doi.org/10.1088/1367-2630/3/1/323
2. A. J. Leggett, 'Bose–Einstein condensation in a harmonic trap: effect of interactions on $T_c$', J. Stat. Phys. 110 (2003) 903–910, DOI: 10.1023/A:1022124222128, https://doi.org/10.1023/A:1022124222128
3. G. Baym, J.-P. Blaizot, M. Holzmann, F. Laloë, D. Vautherin, 'The transition temperature of the dilute interacting Bose gas', Phys. Rev. Lett. 83 (1999) 1703–1706, DOI: 10.1103/PhysRevLett.83.1703, https://doi.org/10.1103/PhysRevLett.83.1703
4. P. C. Hohenberg, 'Existence of long-range order in one and two dimensions', Phys. Rev. 158 (1967) 383–386, DOI: 10.1103/PhysRev.158.383, https://doi.org/10.1103/PhysRev.158.383
5. M. Holzmann, G. Baym, 'Condensate density and superfluid mass density of a dilute Bose–Einstein condensate near the condensation transition', Phys. Rev. Lett. 90 (2003) 040402, DOI: 10.1103/PhysRevLett.90.040402, https://doi.org/10.1103/PhysRevLett.90.040402
6. R. Seiringer, 'Free energy of a dilute Bose gas: lower bound', Commun. Math. Phys. 279 (2008) 595–636, DOI: 10.1007/s00220-008-0428-2, https://doi.org/10.1007/s00220-008-0428-2
7. J. Yin, 'Free energies of dilute Bose gases: upper bound', J. Stat. Phys. 141 (2010) 683–726, DOI: 10.1007/s10955-010-0066-x, https://doi.org/10.1007/s10955-010-0066-x
8. R. Seiringer, D. Ueltschi, 'Rigorous upper bound on the critical temperature of dilute Bose gases', Phys. Rev. B 80 (2009) 014502, DOI: 10.1103/PhysRevB.80.014502, arXiv:0904.0050, https://arxiv.org/abs/0904.0050
9. A. Deuchert, R. Seiringer, J. Yngvason, 'Bose–Einstein condensation in a dilute, trapped gas at positive temperature', Commun. Math. Phys. 368 (2019) 723–776, DOI: 10.1007/s00220-018-3239-0, arXiv:1803.05180, https://arxiv.org/abs/1803.05180
10. A. Deuchert, R. Seiringer, 'Semiclassical approximation and critical temperature shift for weakly interacting trapped bosons', J. Funct. Anal. 281 (2021) 109096, DOI: 10.1016/j.jfa.2021.109096, arXiv:2009.00992, https://arxiv.org/abs/2009.00992
11. S. Chatterjee, P. Diaconis, 'Fluctuations of the Bose–Einstein condensate', J. Phys. A: Math. Theor. 47 (2014) 085201, DOI: 10.1088/1751-8113-47-8-085201, https://doi.org/10.1088/1751-8113-47-8-085201
12. G. Basti, C. Boccato, S. Cenatiempo, A. Deuchert, 'A new upper bound on the specific free energy of dilute Bose gases', arXiv:2507.20877 (2025), https://arxiv.org/abs/2507.20877
13. M. B. Kruk, P. Kulik, M. F. Andersen, P. Deuar, M. Gajda, K. Pawłowski, E. Witkowska, J. J. Arlt, K. Rzążewski, 'On the fluctuations of the number of atoms in the condensate', Rep. Prog. Phys. 88 (2025) 106401, DOI: 10.1088/1361-6633/ae0e33, arXiv:2502.10880, https://arxiv.org/abs/2502.10880
14. A. Deuchert, X. Li, 'Large deviations for the Bose–Einstein condensate of the ideal gas in the canonical ensemble', arXiv:2608.26378 (2026), https://arxiv.org/abs/2608.26378
15. A. Deuchert, P. T. Nam, M. Napiórkowski, 'The Gibbs state of the mean-field Bose gas', arXiv:2501.19396 (2025), https://arxiv.org/abs/2501.19396