arXiv:2610.20434
Universal enhancement of the pairing gap in a dilute Fermi superfluid
Silas R. Beane, Roland C. Farrell, Cameron P. Shotwell
Abstract
At zero temperature, a dilute Fermi gas with an attractive contact interaction forms a superfluid with gapped fermionic excitations. We show that the gap at weak coupling can be organized as a trans-series expansion in the interaction strength and the asymptotic Gor'kov-Melik-Barkhudarov gap. Low-energy virtual quasiparticles generate corrections that are linear in the gap with a leading coefficient nearly 25$\times$ larger than the other known coefficients. This substantially increases the gap at intermediate densities, highlighting that systematic corrections are crucial for quantitative predictions.
AI Summary
For a zero-temperature, dilute spin-$1/2$ Fermi gas with attractive contact interactions, this paper asks how to systematically expand the fermionic excitation gap beyond the Gor’kov–Melik-Barkhudarov (GMB) asymptotic result. It organizes $\log x$, with $x=\Delta/E_F$, as a trans-series in $\lambda=2k_Fa/\pi<0$ and the exponentially small GMB gap $x_0$. The calculation starts from the exact zero-momentum Goldstone-pole condition for the pairing vertex. A diagrammatic expansion through explicit third order in the interaction, combined with separation of hard and soft quasiparticle energies, identifies a previously uncomputed correction linear in the gap. Soft virtual quasiparticles contribute through both the irreducible pairing interaction and the on-shell self-energy. The resulting coefficient is $c_{2,1}=24.32356\ldots$, with contributions $17.63884\ldots$ and $6.68473\ldots$, respectively. Correcting an earlier calculation also gives $c_{1,0}=-0.15263\ldots$. The retained prediction is $x=x_0[1+c_{1,0}\lambda+c_{2,1}\lambda^2x_0]$; the linear-in-gap term dominates the known corrections for $-k_Fa>0.49$. The trans-series structure relies on stated assumptions about diagrammatic expansions and logarithmic enhancements; completeness of $c_{2,1}$ additionally assumes higher-order kernels generate no enhancing outer Cooper logarithms. Truncation bands use independent Gaussian priors for omitted coefficients, rather than rigorous error bounds. The prediction lies about $2\sigma$ above available QMC and experimental points near $-k_Fa\approx1$, where the retained corrections exceed the GMB gap and the expansion is considered unreliable. Data at $-k_Fa\in[0.6,0.9]$ are proposed as a test.
Central claim may not hold: The claimed trans-series structure is established only conditionally on three assumptions, including an all-orders bound on logarithmic powers that the paper does not prove. The reported leading linear-in-gap coefficient is complete only if higher pairing-irreducible kernels lack logarithmic enhancements that the paper assumes, rather than establishes, are absent.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 47 pages, 6 figures