betaFeedback

WorksMathematics

Even Asymptotic Expansions and Uniform Hard-Edge Approximations for Wigner 3j3j and 6j6j Symbols at Fixed Rank

Abstract

We study Wigner 3j3j symbols with two large and one small angular momentum, and the 6j6j symbols with one small entry (Edmonds' limit), both at fixed rank: the small labels stay fixed while the large ones grow. Their classical limits are Wigner rotation matrix elements. Combining a Regge symmetry with the Weyl shift j↦j+12j\mapsto j+\tfrac12 singles out midpoint variables: for 3j3j, the large scale N=j−+j++1N=j_-+j_++1 (j∓j_\mp the two large angular momenta) and x=2μ/Nx=2\mu/N (2μ2\mu the difference of the two large magnetic indices), and, for 6j6j, large lengths A,B,CA,B,C shifted by offsets that depend on the small labels. Among constant shifts of the large labels and magnetic coordinate, and up to rank-independent gauge factors, these are the only centerings that remove the generic first-order term of the exact recurrences. In these variables the normalized symbols are values of analytic functions of the inverse square of the large scale, for integer and half-integer small labels alike (including spin–orbit coefficients): the expansions contain only even powers and converge. The first correction is a closed formula in one Wigner dd function and its derivative, improving the error from second to fourth order in the inverse large scale; all higher corrections are constructed as local differential operators. The exact finite problems are Hahn and Racah spectral systems with undeformed Wigner spectra. Near the ends of the physical lattice (the hard edges), an explicit product of Pochhammer factors carries the difference between the discrete and continuum weights; extracting it gives expansions that are uniform on the whole lattice, including the 6j6j corner where two large lengths nearly coincide (A≈BA\approx B). We also show that centered Hahn and Racah polynomials are even deformations of Jacobi polynomials.

Introduction

Wigner 3j3j symbols with two large and one small angular momentum are the coupling coefficients of squeezed configurations: they appear in Gaunt integrals and the spin-weighted couplings of full-sky cosmic microwave background (CMB) bispectrum and lensing estimators [1, 2], in the Hönl–London line strengths of rotational spectroscopy [3, 4], and, with half-integer small labels, in spin–orbit coupling. The 6j6j symbols with one small entry are the recoupling coefficients of a small-rank tensor operator acting on one of two coupled large angular momenta [5]. In both cases the classical leading approximation is a Wigner rotation matrix element [5, 6, 7]. This paper shows why the averaged (midpoint) variables are the right ones, gives the corrections to all orders, and makes the approximation uniform up to the edges of the physical lattice, all at fixed rank: the small labels are fixed while the large ones grow.

For the 3j3j symbol the leading form goes back to Edmonds, who stated without proof that it tends to a Wigner dd function whose angle is set by the magnetic index γ\gamma of one large spin cc, cos⁡θ=γ/c(c+1)\cos\theta=\gamma/\sqrt{c(c+1)} [5]. Brussaard and Tolhoek gave a heuristic derivation [6], Ponzano and Regge’s triangle geometry replaces c(c+1)\sqrt{c(c+1)} by c+12c+\frac{1}{2} [33], and Flude proved the limit with this one-sided scale [7 Thm. 1]. McFarlane replaced the one-sided angle by an average over the two large spins, cos⁡θ=(m+m′)/(l+l′+1)\cos\theta=(m+m')/(l+l'+1) for ⟨LM l′m′∣lm⟩\langle LM\,l'm'|lm\rangle, in which the difference l−l′l-l' becomes the second magnetic index of the dd function [27]. He chose this average by matching the exact rank recurrence to that of the dd functions: the diagonal coefficients then agree exactly, and the off-diagonal ones differ only through k2/(l+l′+1)2k^2/(l+l'+1)^2. Morehead derived the same form from the Weyl correspondence, which evaluates matrix elements on the average quantized torus, proved that its relative error is O(l−2)O(l^{-2}), and noted that one-sided choices such as cos⁡θ=m/l\cos\theta=m/l leave a first-order error [29 Eq. (3.31) and Appendix]. Carron and Lewis rederived the symmetric form from the Regge symmetry, with normalization 2ℓ+1\sqrt{2\ell+1} [1 App. B, Eq. (B19)]. For a 3j3j coefficient with a small angular momentum LL of magnetic index MM, two large sides j∓=ℓ∓σ/2j_{\mp}=\ell\mp\sigma/2 and magnetic indices m∓=±μ−M/2m_{\mp}=\pm\mu-M/2, it reads

(−1)2L+j++m+2ℓ+1(Lj−j+Mm−m+)⟶dMσL(θ),cos⁡θ=2μ2ℓ+1.(1)(-1)^{2L+j_{+}+m_{+}}\sqrt{2\ell+1} \begin{pmatrix} L & j_{-} & j_{+}\\ M & m_{-} & m_{+} \end{pmatrix} \longrightarrow d^{L}_{M\sigma}(\theta),\qquad\cos\theta=\frac{2\mu}{2\ell+1}. \tag*{(1)}

Here 2ℓ+1=j−+j++12\ell+1=j_{-}+j_{+}+1 is the mean of 2j−+12j_{-}+1 and 2j++12j_{+}+1, twice the midpoint of the Weyl lengths j∓+12j_{\mp}+\frac{1}{2}. The normalization [(2j−+1)(2j++1)]1/4[(2j_{-}+1)(2j_{+}+1)]^{1/4} of McFarlane and Morehead differs from 2ℓ+1\sqrt{2\ell+1} only by a factor 1+O(ℓ−2)1+O(\ell^{-2}), and the sign selects one analytic branch across both large-spin parity classes (§ 4). We prove that the physical branch is analytic in (2ℓ+1)−2(2\ell+1)^{-2}, and that the midpoint variables

N=2ℓ+1=j−+j++1,x=2μN(2)N=2\ell+1=j_{-}+j_{+}+1,\qquad x=\frac{2\mu}{N} \tag*{(2)}

are the only constant shifts of NN and 2μ2\mu for which the generic first-order term of the exact magnetic difference equation (30) vanishes, up to rank-independent gauge factors (Proposition 4.3). The one-sided scale c+12c+\frac{1}{2} of Ponzano–Regge and Flude is such a shift (2c+1=N∓σ(2c+1=N\mp\sigma and 2γ=±2μ−M2\gamma=\pm2\mu-M for c=j∓c=j_{\mp}, the sign of μ\mu being a reflection x↦−xx\mapsto-x), so unless σ=M=0\sigma=M=0 it carries a first-order error. Proposition 4.3 thus makes McFarlane’s recurrence matching and Morehead’s remark on one-sided expansion points precise, and Theorem 5.6 strengthens Morehead’s O(N−2)O(N^{-2}) error to a convergent expansion in N−2N^{-2} with explicit coefficients.

The directly usable consequence is the next term. Writing ΦLM,σ\Phi^{M,\sigma}_{L} for the normalized coefficient (17) and putting

f(x)=dMσL(arccos⁡x),J=L(L+1),(3)f(x)=d^{L}_{M\sigma}(\arccos x),\qquad J=L(L+1), \tag*{(3)}

one has, uniformly on compact subsets of −1<x<1-1<x<1,

ΦLM,σ=f+A2(x)f′(x)+B2(x)f(x)12N2(x2−1)+O(N−4),(4)\Phi^{M,\sigma}_{L}=f+\frac{A_{2}(x)f'(x)+B_{2}(x)f(x)}{12N^{2}(x^{2}-1)}+O(N^{-4}), \tag*{(4)}

where

A2(x)=2J(x3−x)−4x(M2+σ2)+2Mσ(x2+3)+4x,B2(x)=J(x2−3)+2Mσx.(5)\begin{aligned} A_{2}(x)&=2J(x^{3}-x)-4x(M^{2}+\sigma^{2})+2M\sigma(x^{2}+3)+4x,\\ B_{2}(x)&=J(x^{2}-3)+2M\sigma x. \tag*{(5)} \end{aligned}

The Regge-symmetric dd approximation therefore has an intrinsic O(N−2)O(N^{-2}) error in the bulk (away from the ends x=±1x=\pm1 of the grid), and one dd matrix and one derivative improve it to O(N−4)O(N^{-4}) (Theorem 6.1).

For the one-small 6j6j symbol the classical limit of Edmonds and Flude is

\begin{Bmatrix} c & b & a\\ f & a+n & b+m \end{Bmatrix} \sim\text{phase}\times(\text{large normalization})^{-1}d^{f}_{nm}(\phi) \tag*{(6)}\end{Bmatrix}

with ϕ\phi a law-of-cosines angle of the large labels, for fixed f,m,nf,m,n and large a,b,ca,b,c [5, 7] (the symbol is that of § 7 with its columns permuted). McFarlane gave an averaged version, again by matching rank recurrences [28 Eqs. (16), (A10)]. In terms of the Weyl-shifted lengths

A=a+n+12,B=b+m+12,C=c+12.(7)A=a+\frac{n+1}{2},\qquad B=b+\frac{m+1}{2},\qquad C=c+\frac{1}{2}. \tag*{(7)}

which are the midpoints of the Weyl lengths of the pairs (a,a+n)(a, a+n) and (b,b+m)(b, b+m) and the Weyl length of cc, his angle and normalization agree through first order with x=cos⁡θ=(A2+B2−C2)/(2AB)x=\cos\theta=(A^2+B^2-C^2)/(2AB) and 1/(2AB)1/(2\sqrt{AB}); his cosine is x+(m2+n2)/(8AB)x+(m^2+n^2)/(8AB). Within the same class of ff-independent gauges, (7) is the unique affine centering of the exact recurrence (Proposition 7.2), and the centering must depend on the offsets: replacing every large spin by j+12j+\frac{1}{2} (the case u=v=w=12u=v=w=\frac{1}{2} of Proposition 7.2) leaves a generic first-order error unless m=n=0m=n=0, and Flude’s variables leave one even then when f>0f>0, through his normalization (Section 9.3). The first correction is again one dd function and one derivative, with coefficients homogeneous of degree −2-2 in (A,B,C)(A,B,C) (Theorem 9.1).

Results. The small labels remain fixed: LL, MM, σ\sigma as N→∞N\to\infty for 3j3j, and ff, mm, nn as A,B→∞A,B\to\infty for 6j6j. In both families they may be integers or half-integers; for 3j3j the half-integer case is that of large spins in opposite integer/half-integer classes, such as spin–orbit coefficients. The bulk 6j6j results require CC to grow within a strict-triangle cone; the uniform results allow every physical CC, including bounded CC near the corner A=BA=B, provided A/BA/B stays bounded away from 00 and ∞\infty (λ≤A/B≤λ−1\lambda\le A/B\le\lambda^{-1} for a fixed λ∈(0,1)\lambda\in(0,1)).

FamilyRegionResult
3j3j bulkxx in a compact subset of (−1,1)(-1,1)Convergent even expansion in N−2N^{-2}; explicit terms through N−4N^{-4} (Theorems 5.6, 6.1, 6.3).
3j3j uniformwhole physical gridΦ=EN[d+O(N−2)W]\Phi=E_N[d+O(N^{-2})W] with an explicit edge factor ENE_N, and an all-order even expansion with polynomial coefficients after ENWE_NW is extracted (Theorem 10.1).
6j6j bulkcompact strict-triangle coneHomogeneous all-even expansion; explicit degree −2-2, recursive degree −4-4 (Theorems 8.7, 9.1, 9.2 and (245)).
6j6j uniformwhole cc-lattice, λ≤A/B≤λ−1\lambda\le A/B\le\lambda^{-1}Φf=EA,B[d+O((A+B)−2)W]\Phi_f=E_{A,B}[d+O((A+B)^{-2})W], including the corner A≈BA\approx B; all orders (Theorem 10.1).

Table 1.

Here Φ\Phi and Φf\Phi_f are the normalized symbols, ENE_N and EA,BE_{A,B} explicit edge factors, and WW is the endpoint envelope of the Wigner function, so the uniform estimates are additive and remain valid at zeros of the leading mode. At the 3j3j and separated 6j6j hard edges (the ends of the lattice, where the link coefficients of the exact recurrences vanish) they give relative Laguerre laws, and the lower 6j6j corner carries a second explicit factor (Corollaries 10.4 and 10.5, Remark 10.7). Behind these formulas are exact structures that we also prove: the finite problems are Hahn and Racah spectral systems with the undeformed spectra L(L+1)L(L+1) and f(f+1)f(f+1), unitarily equivalent to truncated continuum Wigner problems (Theorem 5.1 and Corollary 5.3; Theorem 8.2 and Corollary 8.3), and every correction is a local differential operator constructed recursively (Theorems 6.5 and 9.2).

The even expansions rest on exact square-root-free closed forms of the physical kernels (Lemmas 5.4 and 8.5) in which the large scale enters only through its inverse square, so no odd power can occur. Midpoint centering alone does not force this: it makes the centered recurrences invariant under reversal of the large scale, and it is the physical initial data that select the even sector (Remark 3.3).

References

  1. [1]J. Carron and A. Lewis, Spherical spectrum expansion and quadratic estimators, JCAP 07 (2024) 067; doi:10.1088/1475-7516/2024/07/067; arXiv:2404.16797.DOI
  2. [2]T. Okamoto and W. Hu, CMB lensing reconstruction on the full sky, Phys. Rev. D 67 (2003) 083002; doi:10.1103/PhysRevD.67.083002; arXiv:astro-ph/0301031.DOI
  3. [3]H. Hönl and F. London, Über die Intensitäten der Bandenlinien, Z. Phys. 33 (1925) 803–809; doi:10.1007/BF01328367.
  4. [4]A. Hansson and J. K. G. Watson, A comment on Hönl–London factors, J. Mol. Spectrosc. 233 (2005) 169–173; doi:10.1016/j.jms.2005.06.009.DOI
  5. [5]A. R. Edmonds, Angular Momentum in Quantum Mechanics, Princeton University Press, Princeton (1957).
  6. [6]P. J. Brussaard and H. A. Tolhoek, Classical limits of Clebsch–Gordan coefficients, Racah coefficients and Dmn functions, Physica 23 (1957) 955–971; doi:10.1016/S0031-8914(57)95547-7.DOI
  7. [7]J. P. M. Flude, The Edmonds asymptotic formulas for the 3j and 6j symbols, J. Math. Phys. 39 (1998) 3906–3915; doi:10.1063/1.532474; arXiv:physics/9710037.DOI
  8. [8]A. Lewis, Semiclassical Asymptotics of Wigner 3j and 6j Symbols at Growing Rank: Phase Horizons and Turning Connections, companion paper, in preparation (2026).
  9. [9]K. Schulten and R. G. Gordon, Exact recursive evaluation of 3j- and 6j-coefficients for quantum-mechanical coupling of angular momenta, J. Math. Phys. 16 (1975) 1961–1970; doi:10.1063/1.522426.DOI
  10. [10]K. Schulten and R. G. Gordon, Semiclassical approximations to 3j- and 6j-coefficients for quantum-mechanical coupling of angular momenta, J. Math. Phys. 16 (1975) 1971–1988; doi:10.1063/1.522427.DOI
  11. [11]M. W. Reinsch and J. J. Morehead, Asymptotics of Clebsch–Gordan coefficients, J. Math. Phys. 40 (1999) 4782–4806; doi:10.1063/1.533000; arXiv:math-ph/9906007.DOI
  12. [12]S. Karlin and J. McGregor, The Hahn polynomials, formulas and an application, Scripta Math. 26 (1961) 33–46.
  13. [13]T. H. Koornwinder, Clebsch–Gordan coefficients for SU(2) and Hahn polynomials, Nieuw Arch. Wisk. (3) 29 (1981) 140–155.
  14. [14]A. C. P. Bitencourt, M. Ragni, R. G. Littlejohn, R. Anderson and V. Aquilanti, The screen representation of vector coupling coefficients or Wigner 3j symbols: exact computation and illustration of the asymptotic behavior, Lecture Notes in Computer Science 8579 (2014) 468–481; doi:10.1007/978-3-319-09144-0_32; arXiv:1409.8205.DOI
  15. [15]A. F. Nikiforov, S. K. Suslov and V. B. Uvarov, Classical Orthogonal Polynomials of a Discrete Variable, Springer, Berlin (1991); doi:10.1007/978-3-642-74748-9.DOI
  16. [16]R. Koekoek, P. A. Lesky and R. F. Swarttouw, Hypergeometric Orthogonal Polynomials and Their q-Analogues, Springer, Berlin (2010); doi:10.1007/978-3-642-05014-5.DOI
  17. [17]W. Donnelly, L. Freidel, S. F. Moosavian and A. J. Speranza, Matrix quantization of gravitational edge modes, JHEP 05 (2023) 163, Appendix A.5; doi:10.1007/JHEP05(2023)163; arXiv:2212.09120.DOI
  18. [18]M. Dupuis and E. R. Livine, Pushing further the asymptotics of the 6j-symbol, Phys. Rev. D 80 (2009) 024035; doi:10.1103/PhysRevD.80.024035; arXiv:0905.4188.DOI
  19. [19]W. Kamiński and S. Steinhaus, Coherent states, 6j symbols and properties of the next to leading order asymptotic expansions, J. Math. Phys. 54 (2013) 121703; doi:10.1063/1.4849515; arXiv:1307.5432.DOI
  20. [20]R. G. Littlejohn and L. Yu, Uniform semiclassical approximation for the Wigner 6j symbol in terms of rotation matrices, J. Phys. Chem. A 113 (2009) 14904–14922; doi:10.1021/jp905056y; arXiv:0905.4240.DOI
  21. [21]L. Yu and R. G. Littlejohn, Semiclassical analysis of the Wigner 9j symbol with small and large angular momenta, Phys. Rev. A 83 (2011) 052114; doi:10.1103/PhysRevA.83.052114; arXiv:1104.1499.DOI
  22. [22]V. Bonzom and P. Fleury, Asymptotics of Wigner 3nj-symbols with small and large angular momenta: an elementary method, J. Phys. A: Math. Theor. 45 (2012) 075202; doi:10.1088/1751-8113/45/7/075202; arXiv:1108.1569.DOI
  23. [23]D. A. Varshalovich, A. N. Moskalev and V. K. Khersonskii, Quantum Theory of Angular Momentum, World Scientific, Singapore (1988); doi:10.1142/0270.DOI
  24. [24]NIST Digital Library of Mathematical Functions, Chapter 34, 3j, 6j, 9j Symbols, Secs. 34.3 and 34.5; https://dlmf.nist.gov/34.
  25. [25]T. Regge, Symmetry properties of Clebsch–Gordan’s coefficients, Nuovo Cimento 10 (1958) 544–545; doi:10.1007/BF02859841.DOI
  26. [26]D. Neville, A technique for solving recurrence relations approximately and its application to the 3-J and 6-J symbols, J. Math. Phys. 12 (1971) 2438–2453.DOI
  27. [27]S. C. McFarlane, Angular momentum and Heisenberg’s correspondence principle, J. Phys. B 25 (1992) 4045–4057.
  28. [28]S. C. McFarlane, Angular momentum and Heisenberg’s correspondence principle. II. More complex matrix elements, J. Phys. B 26 (1993) 1871–1884.DOI
  29. [29]J. J. Morehead, Semiclassical integrable matrix elements, Phys. Rev. A 53 (1996) 1285–1294; doi:10.1103/PhysRevA.53.1285.DOI
  30. [30]T. Regge, Symmetry properties of Racah’s coefficients, Nuovo Cimento 11 (1959) 116–117; doi:10.1007/BF02724914.DOI
  31. [31]NIST Digital Library of Mathematical Functions, Secs. 18.25–18.26, Wilson class: definitions and representations; https://dlmf.nist.gov/18.25.
  32. [32]R. Askey and J. Wilson, A set of orthogonal polynomials that generalize the Racah coefficients or 6-j symbols, SIAM J. Math. Anal. 10 (1979) 1008–1016; doi:10.1137/0510092.DOI
  33. [33]G. Ponzano and T. Regge, Semiclassical limit of Racah coefficients, in Spectroscopic and Group Theoretical Methods in Physics, ed. F. Bloch, North-Holland, Amsterdam (1968) 1–58.
  34. [34]J. H. Luscombe and M. Luban, Simplified recursive algorithm for Wigner 3j and 6j symbols, Phys. Rev. E 57 (1998) 7274–7277; doi:10.1103/PhysRevE.57.7274.
  35. [35]H. T. Johansson and C. Forssén, Fast and accurate evaluation of Wigner 3j, 6j, and 9j symbols using prime factorization and multiword integer arithmetic, SIAM J. Sci. Comput. 38 (2016) A376–A384; doi:10.1137/15M1021908; arXiv:1504.08329.DOI
  36. [36]M.-M. Derriennic, Sur l’approximation de fonctions intégrables sur [0, 1] par des polynômes de Bernstein modifiés, J. Approx. Theory 31 (1981) 325–343; doi:10.1016/0021-9045(81)90101-5.
  37. [37]H. Berens and Y. Xu, On Bernstein–Durrmeyer polynomials with Jacobi weights, in Approximation Theory and Functional Analysis, Academic Press, Boston (1991) 25–46.
  38. [38]Z. Ciesielski, Explicit formula relating the Jacobi, Hahn and Bernstein polynomials, SIAM J. Math. Anal. 18 (1987) 1573–1575; doi:10.1137/0518113.DOI
  39. [39]A. Makur and L. Zheng, Polynomial spectral decomposition of conditional expectation operators, in Proc. 54th Annual Allerton Conference on Communication, Control, and Computing, IEEE (2016) 633–640; doi:10.1109/ALLERTON.2016.7852291.DOI

Paper details

vlnc:W_w7q59afn
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 6

downloads · 0

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

  • Mathematics

Anthropic Claude Opus 5.5, Anthropic Claude Sonnet 5.5, OpenAI Codex 6.1 sol, Google Gemini 4 Hydrogen/Argon/Carbon

View statement
Looking for related papers…
Contents