For a trapped ideal Bose gas with a fixed total particle number NN, Politzer (1996) showed that the condensate occupation N0N_0 retains its grand-canonical mean while the condensate-number fluctuations become microscopic, and he evaluated the canonical partition function numerically for N=102N=10^2 to 10610^6. The paper itself states that this is 'nowhere near to a "theory" of the large-NN asymptotics' of the canonical ideal Bose gas: the next-order, finite-NN corrections — which would explain the magnitude, shape, and sign of the canonical-versus-grand-canonical discrepancy in N0N_0 near the transition — remain to be derived analytically; a direct microcanonical evaluation and a systematic treatment of weakly interacting trapped gases, quantifying the crossover between interaction-induced and fixed-NN-induced suppression of fluctuations, are likewise left open. The next-to-leading-order analytic theory for three-dimensional traps, whose next-to-leading finite-NN corrections remain to be derived beyond the closed leading-order asymptotic expressions of Holthaus, Kalinowski, and Kirsten (1998), and the statistical behavior of weakly interacting gases, remain open; a 2025 Reports on Progress in Physics review describes the latter as 'less thoroughly resolved'.

Background

Bose-Einstein condensation (BEC) is the macroscopic occupation of a single one-particle state by a gas of bosons. In the dilute atomic gases first confined in magnetic traps in 1995, the confining potential is well approximated by a three-dimensional isotropic harmonic oscillator with level spacing ϵ\epsilon and level energies ϵm=mϵ\epsilon_m=m\epsilon of degeneracy (m+1)(m+2)/2(m+1)(m+2)/2. The three standard statistical ensembles describe different physical situations: the grand canonical ensemble fixes the temperature TT and a chemical potential μ\mu (a system in diffusive contact with a particle reservoir), the canonical ensemble fixes TT and the total particle number NN, and the microcanonical ensemble fixes both NN and the total energy EE (an isolated system). Textbook arguments guarantee equivalence of the three ensembles in the thermodynamic limit — here N→∞N\to\infty with N(ϵ/T)3N(\epsilon/T)^3 held fixed — provided the particles interact, because a finite compressibility makes macroscopic density fluctuations energetically costly. For a non-interacting (ideal) gas the argument fails in one dramatic respect: grand canonically, each one-particle level obeys Bose statistics with mean occupation Ni=1/(e(ϵi−μ)/T−1)N_i=1/(e^{(\epsilon_i-\mu)/T}-1) and variance Ni(Ni+1)N_i(N_i+1), so once the ground-state occupation N0N_0 is macroscopic below the critical temperature TcT_c, the grand-canonical fluctuation of the condensate number is of order N0N_0 itself — an anomalous, unphysically large fluctuation for a finite isolated cloud. The quantity ζ(3)≈1.202\zeta(3)\approx1.202 is a Riemann zeta value; the excited-particle number is Ne=ζ(3)(T/ϵ)3N_e=\zeta(3)(T/\epsilon)^3, giving Tc=N1/3ζ(3)−1/3ϵT_c=N^{1/3}\zeta(3)^{-1/3}\epsilon; the fugacity λ=eμ/T\lambda=e^{\mu/T} approaches 1 below TcT_c.

Politzer (1996) analyzed exactly this regime for a fixed, large NN. Because the excited-state occupations depend on the condensate only through corrections of order 1/N01/N_0, he showed that all mean occupations retain their grand-canonical values while the condensate number is fixed by N0=N−NeN_0=N-N_e; hence ΔN0=ΔNe∼(T/ϵ)3/2\Delta N_0=\Delta N_e\sim(T/\epsilon)^{3/2} is microscopic (of order N1/2N^{1/2}, versus order NN grand canonically), and he derived the estimate ΔN0N0=1N(T/Tc)3/21−(T/Tc)3π26ζ(3)\frac{\Delta N_0}{N_0}=\frac{1}{\sqrt{N}}\frac{(T/T_c)^{3/2}}{\sqrt{1-(T/T_c)^3}}\sqrt{\frac{\pi^2}{6\zeta(3)}} together with the induced inter-level correlation ⟨Δn0Δn1⟩/(N0N1)\langle\Delta n_0\Delta n_1\rangle/(N_0N_1) (with nin_i the actual occupation of level ii in a given configuration), equal to −N−2/3T/Tc1−(T/Tc)3ζ(3)−1/3-N^{-2/3}\frac{T/T_c}{1-(T/T_c)^3}\zeta(3)^{-1/3}. He evaluated the canonical partition function, represented as a Fourier integral Z(N,T)=12π∫−ππdz eiNz∏m(1−e−mϵ/T−iz)−(m+1)(m+2)/2Z(N,T)=\frac{1}{2\pi}\int_{-\pi}^{\pi}dz\,e^{iNz}\prod_m(1-e^{-m\epsilon/T-iz})^{-(m+1)(m+2)/2}, for N=102N=10^2 to 10610^6, confirming these scalings. Three limitations were stated explicitly. First, 'this is nowhere near to a “theory” of the large-NN asymptotics of the canonical ideal Bose gas': the next level of approximation, needed to account for the numerically observed finite-NN discrepancy between canonical and grand-canonical N0N_0 (his Fig. 2; decreasing roughly like N−1.15N^{-1.15} at intermediate T/TcT/T_c but only slowly just below TcT_c), 'would require a detailed analysis of the canonical or microcanonical partition function and may be very difficult to determine analytically'. Second, a direct numerical evaluation of the microcanonical partition function 'would be considerably more difficult' and was not performed. Third, Appendix B gave only a mean-field estimate of fluctuation damping by repulsive interactions (pairwise ground-state energy λn02\lambda n_0^2, where λ\lambda — a different λ\lambda from the fugacity above — is the two-particle interaction contribution to the ground-state energy, equal to a/2πa/\sqrt{2\pi} in natural oscillator units ℏ=m=ω0=1\hbar=m=\omega_0=1 for scattering length aa): fixing NN dominates over interactions only if that pairwise energy is below O(N−2/3ϵ)O(N^{-2/3}\epsilon), while interactions dominate when λ/ϵ≳(T/ϵ)−2\lambda/\epsilon\gtrsim(T/\epsilon)^{-2} — leaving open which effect controls real, weakly interacting, finite clouds.

Later work built substantially on this. Exact recurrence relations for the canonical partition function (e.g. ZN=1N∑l=1NglZN−lZ_N=\frac{1}{N}\sum_{l=1}^{N}g_lZ_{N-l} with gl=∑mdme−lϵm/Tg_l=\sum_md_me^{-l\epsilon_m/T}, exposited by Mullin and Fernández) removed the floating-point limitations of the Fourier-integral method. Holthaus, Kalinowski, and Kirsten used the Mellin-Barnes transformation (a contour-integral representation that expresses thermodynamic sums through the poles of an associated zeta function) to derive closed leading-order asymptotic expressions for the canonical N0N_0, its fluctuation, and the canonical-minus-microcanonical difference, and studied microcanonical statistics of one- and three-dimensional harmonic traps by the saddle-point method (approximating an integral through the exponential behavior near its dominant stationary point), exploiting the exact correspondence between the one-dimensional trapped gas and the partition theory of integers. The Kocharovsky-Scully school developed a canonical-ensemble quasiparticle formulation yielding all higher moments of condensate fluctuations and, later, an analytical theory of condensation statistics at mesoscopic (finite, intermediate) particle numbers covering the whole critical region; Klünder and Pelster constructed a systematic semiclassical expansion of the harmonically trapped ideal gas valid down to small NN; Chatterjee and Diaconis proved rigorous orders and limiting distributions (condensate-fraction fluctuations of order n−1/2n^{-1/2} with normal limit for the three-dimensional harmonic trap, n−1/3n^{-1/3} for the box). On the interacting side, number-conserving Bogoliubov treatments (a quasiparticle formulation that, unlike the standard Bogoliubov treatment, preserves the total particle number), hybrid master-equation approaches, and a stochastic 'Fock state sampling' method (Monte Carlo sampling of many-body occupation-number states) appeared; and number-stabilized experiments observed condensate atom-number fluctuations directly in 2019, revealing their microcanonical character in 2021. A 2025 Reports on Progress in Physics review summarizes that, for the ideal gas, explicit analytic results remain 'primarily one-dimensional' and that the statistical behavior of weakly interacting Bose gases is 'less thoroughly resolved'.

Problem Statement

Construct the quantitative theory of the statistics of the number of condensed atoms in a trapped Bose gas at fixed total particle number, beyond the leading grand-canonical-equivalent behavior established by Politzer (1996), at the level of control the source paper itself called for. Notation: NN is the total atom number, N0N_0 the condensate (ground-state) occupation, ΔN0\Delta N_0 the standard deviation of N0N_0, TT the temperature, and TcT_c the condensation critical temperature, which for the isotropic three-dimensional harmonic trap with level spacing ϵ\epsilon is Tc=N1/3ζ(3)−1/3ϵT_c=N^{1/3}\zeta(3)^{-1/3}\epsilon (ζ\zeta is the Riemann zeta function). The requested results, which serve one common objective — replacing heuristic and purely numerical fixed-N results by a systematic, experimentally testable theory — are:

(1) Ideal-gas analytic theory. An analytic asymptotic theory of the canonical partition function of the ideal Bose gas in a harmonic trap at large NN, including the next-to-leading finite-NN corrections, that accounts for the magnitude, shape, and sign of the difference between the canonical condensate occupation N0N_0 and the grand-canonical limit N(1−(T/Tc)3)N(1-(T/T_c)^3) across the transition region (the discrepancy exhibited in Fig. 2 of Politzer 1996, which decreases roughly like N−1.15N^{-1.15} at intermediate T/TcT/T_c but only slowly just below TcT_c), together with the corresponding corrections to the condensate fluctuation ΔN0\Delta N_0 and the inter-level correlations ⟨ΔniΔnj⟩\langle\Delta n_i\Delta n_j\rangle (with nin_i the occupation of the ii-th single-particle level).

(2) Microcanonical ensemble. The corresponding microcanonical statistics — with both NN and the total energy EE fixed — at the same analytic level as part (1), i.e. beyond the leading-order asymptotic results of Holthaus, Kalinowski, and Kirsten (1998), including the canonical-versus-microcanonical difference in fluctuations near TcT_c.

(3) Weakly interacting gas. A systematic treatment of the dilute weakly interacting trapped gas that quantifies, for experimentally relevant trap frequencies, densities, and scattering lengths, the crossover between interaction-induced damping of condensate fluctuations (estimated only at mean-field level in Appendix B of Politzer 1996, with fixed-NN dominance requiring the ground-state pairwise interaction energy to be below O(N−2/3ϵ)O(N^{-2/3}\epsilon)) and fixed-NN suppression, reconciling canonical, microcanonical, and interaction effects.

The completion boundary is a theory whose predictions for the mean and variance of N0N_0 (and their correlations) are closed-form or systematically ordered in the small parameters 1/N1/N, the scattering length, and the distance from TcT_c; agree with numerically exact fixed-NN ensemble computations; and are quantitatively consistent with measured condensate fluctuation data in weakly interacting trapped gases. The theory is to address all three parts.

References

  1. H. David Politzer, 'Condensate fluctuations of a trapped, ideal Bose gas', Physical Review A 54, 5048-5054 (1996). DOI: 10.1103/PhysRevA.54.5048; arXiv:quant-ph/9606009. URL: https://arxiv.org/abs/quant-ph/9606009
  2. Klaus Kirsten and David J. Toms, 'Bose-Einstein condensation of atomic gases in a general harmonic-oscillator confining potential trap', Physical Review A 54, 4188-4203 (1996). DOI: 10.1103/PhysRevA.54.4188. URL: https://doi.org/10.1103/PhysRevA.54.4188
  3. Siegfried Grossmann and Martin Holthaus, 'Fluctuations of the Particle Number in a Trapped Bose-Einstein Condensate', Physical Review Letters 79, 3557-3560 (1997). DOI: 10.1103/PhysRevLett.79.3557. URL: https://doi.org/10.1103/PhysRevLett.79.3557
  4. Martin Holthaus, Eva Kalinowski, and Klaus Kirsten, 'Condensate Fluctuations in Trapped Bose Gases: Canonical vs. Microcanonical Ensemble', Annals of Physics 270, 198-230 (1998). DOI: 10.1006/aphy.1998.5852. URL: https://doi.org/10.1006/aphy.1998.5852
  5. W. J. Mullin and J. P. Fernández, 'Bose-Einstein condensation, fluctuations, and recurrence relations in statistical mechanics', American Journal of Physics 71, 661-669 (2003). DOI: 10.1119/1.1544520. URL: https://doi.org/10.1119/1.1544520
  6. V. V. Kocharovsky, Vl. V. Kocharovsky, and Marlan O. Scully, 'Condensation of N bosons. III. Analytical results for all higher moments of condensate fluctuations in interacting and ideal dilute Bose gases via the canonical ensemble quasiparticle formulation', Physical Review A 61, 053606 (2000). DOI: 10.1103/PhysRevA.61.053606. URL: https://doi.org/10.1103/PhysRevA.61.053606
  7. Vitaly V. Kocharovsky and Vladimir V. Kocharovsky, 'Analytical theory of mesoscopic Bose-Einstein condensation in an ideal gas', Physical Review A 81, 033615 (2010). DOI: 10.1103/PhysRevA.81.033615. URL: https://doi.org/10.1103/PhysRevA.81.033615
  8. S. V. Tarasov, Vl. V. Kocharovsky, and V. V. Kocharovsky, 'Universal scaling in the statistics and thermodynamics of a Bose-Einstein condensation of an ideal gas in an arbitrary trap', Physical Review A 90, 033605 (2014). DOI: 10.1103/PhysRevA.90.033605. URL: https://doi.org/10.1103/PhysRevA.90.033605
  9. B. Klünder and A. Pelster, 'Systematic semiclassical expansion for harmonically trapped ideal Bose gases', European Physical Journal B 68, 457-465 (2009). DOI: 10.1140/epjb/e2009-00112-9; arXiv:0806.2960. URL: https://arxiv.org/abs/0806.2960
  10. Sourav Chatterjee and Persi Diaconis, 'Fluctuations of the Bose-Einstein condensate', Journal of Physics A: Mathematical and Theoretical 47, 085201 (2014). DOI: 10.1088/1751-8113/47/8/085201. URL: https://doi.org/10.1088/1751-8113/47/8/085201
  11. Olivier Giraud, Aurélien Grabsch, and Christophe Texier, 'Correlations of occupation numbers in the canonical ensemble and application to a Bose-Einstein condensate in a one-dimensional harmonic trap', Physical Review A 97, 053615 (2018). DOI: 10.1103/PhysRevA.97.053615. URL: https://doi.org/10.1103/PhysRevA.97.053615
  12. Hongwei Xiong, Shujuan Liu, Guoxiang Huang, Zhijun Xu, and Cunyuan Zhang, 'Fluctuations of the condensate in ideal and interacting Bose gases', Journal of Physics B 34, 4203-4215 (2001). DOI: 10.1088/0953-4075/34/21/310. URL: https://doi.org/10.1088/0953-4075/34/21/310
  13. Z. Idziaszek and K. Rzążewski, 'Two characteristic temperatures for a Bose-Einstein condensate of a finite number of particles', Physical Review A 68, 035604 (2003). DOI: 10.1103/PhysRevA.68.035604. URL: https://doi.org/10.1103/PhysRevA.68.035604
  14. Zbigniew Idziaszek, 'Microcanonical fluctuations of the condensate in weakly interacting Bose gases', Physical Review A 71, 053604 (2005). DOI: 10.1103/PhysRevA.71.053604. URL: https://doi.org/10.1103/PhysRevA.71.053604
  15. Anatoly A. Svidzinsky and Marlan O. Scully, 'Condensation of N Interacting Bosons: A Hybrid Approach to Condensate Fluctuations', Physical Review Letters 97, 190402 (2006). DOI: 10.1103/PhysRevLett.97.190402. URL: https://doi.org/10.1103/PhysRevLett.97.190402
  16. M. A. Kristensen, M. B. Christensen, M. Gajdacz, M. Iglicki, K. Pawlowski, C. Klempt, J. F. Sherson, K. Rzążewski, A. J. Hilliard, and J. J. Arlt, 'Observation of Atom Number Fluctuations in a Bose-Einstein Condensate', Physical Review Letters 122, 163601 (2019). DOI: 10.1103/PhysRevLett.122.163601. URL: https://doi.org/10.1103/PhysRevLett.122.163601
  17. M. B. Christensen, T. Vibel, A. J. Hilliard, M. B. Kruk, K. Pawlowski, D. Hryniuk, K. Rzążewski, M. A. Kristensen, and J. J. Arlt, 'Observation of Microcanonical Atom Number Fluctuations in a Bose-Einstein Condensate', Physical Review Letters 126, 153601 (2021). DOI: 10.1103/PhysRevLett.126.153601. URL: https://doi.org/10.1103/PhysRevLett.126.153601
  18. Maciej Bartlomiej Kruk, Dawid Hryniuk, Mick Kristensen, Toke Vibel, Krzysztof Pawlowski, Jan Arlt, and Kazimierz Rzążewski, 'Microcanonical and canonical fluctuations in atomic Bose-Einstein condensates – Fock state sampling approach', SciPost Physics 14, 036 (2023). DOI: 10.21468/SciPostPhys.14.3.036. URL: https://doi.org/10.21468/SciPostPhys.14.3.036
  19. Maciej B Kruk, Piotr Kulik, Malthe F Andersen, Piotr Deuar, Mariusz Gajda, Krzysztof Pawlowski, Emilia Witkowska, Jan J Arlt, and Kazimierz Rzążewski, 'On the fluctuations of the number of atoms in the condensate', Reports on Progress in Physics 88, 106401 (2025). DOI: 10.1088/1361-6633/ae0e33; arXiv:2502.10880. URL: https://arxiv.org/abs/2502.10880
  20. T. Vibel, M. B. Christensen, R. M. F. Andersen, L. N. Stokholm, K. Pawlowski, K. Rzążewski, M. A. Kristensen, and J. J. Arlt, 'Atom number fluctuations in Bose gases — statistical analysis of parameter estimation', Journal of Physics B 57, 195301 (2024). DOI: 10.1088/1361-6455/ad7458. URL: https://doi.org/10.1088/1361-6455/ad7458
  21. Giacomo Gradenigo, Dario Lucente, and Luca Salasnich, 'Ideal Bose-Einstein condensation in the canonical ensemble: exact asymptotic estimates from large deviations' (2026). arXiv:2608.24625. URL: https://arxiv.org/abs/2608.24625