Statistical laws of compound-nucleus resonance spacings and widths and their invariance under the long-range channel interaction

Wigner's 1951 paper asked whether the probability distributions of nuclear resonance spacings and widths, measured in units of their local averages, are invariant under changes of the long-range interaction of the scattered particle (the channel boundary functions), and what the explicit laws of these distributions are. In R-matrix theory the internal compound-nucleus dynamics enters the scattering through a meromorphic R-function, while the long-range channel physics acts only through a real fractional-linear transformation; Wigner proved the resulting invariance only for a special class of R-functions and left the general statement as an unproven surmise, without determining the distribution laws. Random-matrix theory later supplied the laws — the Wigner/Gaussian Orthogonal Ensemble spacing law and the Porter-Thomas width law — and embedded them in a complete statistical theory of compound-nucleus reactions in which channel dependence enters only through transmission coefficients (measures of how open each decay channel is); neutron resonance data broadly confirm this picture. Two precise questions nevertheless remain open. First, no derivation of these laws from the underlying nuclear dynamics exists: the universality of the ensemble statistics, and Wigner's invariance principle for general classes of statistical R-functions, are unproven, and the width sector is known not to be universal in general. Second, the width law itself is contested: several careful analyses reject the Porter-Thomas distribution for reduced neutron and gamma-decay widths, and the competing explanations — channel-coupling terms that violate orthogonal invariance, realistic energy dependence of the neutron channel, or statistical artifacts of normalization — have not been reconciled. The problem is to determine the actual spacing and width distribution laws, to establish the exact conditions under which they are invariant under changes of the channel coupling, and to bring theory and measured width statistics into definitive agreement.

Background

When a slow neutron or proton is captured by a heavy nucleus, the resulting equilibrated system — the compound nucleus — lives long enough that the cross section shows a dense forest of narrow resonances, each characterized by a position $H_\mu$ and a width $w_\mu$ (the energy range over which the peak contributes). Wigner's R-matrix theory divides space at a channel radius $a$: inside, the projectile feels the short-range nuclear interaction; outside, only the long-range Coulomb and centrifugal forces act. In each partial wave (a fixed orbital angular momentum and spin coupling) the interior dynamics enters the scattering only through the R-function, the reciprocal of the logarithmic derivative of the interior wave function at the boundary,

$$R(E)=\sum_\mu \frac{\gamma_\mu^2}{E_\mu-E},$$

a meromorphic function of energy whose poles $E_\mu$ are the internal compound-nucleus levels and whose positive residues $\gamma_\mu^2$ are called reduced widths. The exterior physics — everything distinguishing a neutron from a proton or a different partial wave — enters only through the values of the exterior channel solutions and their derivatives at $r=a$ (written $u,u',F,G,F',G'$, with $F$ and $G$ the regular and irregular solutions); these convert $R$ into the observed logarithmic derivative $Q$ by a real fractional-linear (Möbius) transformation, and through the penetration factor — the barrier-transmission factor controlling tunnelling through the Coulomb and centrifugal barrier — they set the scale of the observed widths. Two deterministic facts anchor the statistical analysis: exactly one pole $H_\mu$ of $Q$ lies between successive poles $E_\mu$ of $R$, so the mean spacing $D$ of observed resonances equals that of the internal levels; and the mean observed width is the internal coupling strength per unit energy (the strength function) multiplied by a channel-dependent penetration factor.

When an experimental energy window contains many resonances, spacings and widths admit only a statistical description, through the probability distributions of the normalized spacing $(H_{\mu+1}-H_\mu)/D$ and the normalized width $w_\mu/\bar{w}$; correspondingly the actual $R$-function in the window is replaced by a statistical R-function, an ensemble of functions of the same pole form whose pole positions and residues are random but statistically stationary across the window. Wigner's 1951 question was whether these normalized laws depend on the long-range interaction of the projectile: the mean width must (through the penetration factor), but he surmised that the normalized distributions do not, while for spacings even the mean is channel-independent. His argument introduced the invariant derivative

$$f'(E)=\frac{dR/dE}{1+R^2},$$

which is unchanged by the real orthogonal Möbius transformations relating different boundary conditions, so that the statistics of the poles and residues read off from $f'$ should be the same before and after a change of channel. The argument was explicitly heuristic — Wigner compared it to the ergodic hypothesis and carried out a rigorous proof only for a special class of R-functions — and the explicit distribution laws were left undetermined; the analysis also assumed a single elastic channel with energy-independent boundary functions, and a distant-level constant $R_0$ (the net shift from levels outside the window) estimated from admittedly crude strength-function assumptions.

The question was answered within a different framework. Random matrix theory postulates that the effective Hamiltonian of the compound nucleus is a typical member of the Gaussian Orthogonal Ensemble (GOE) — real symmetric matrices with independent Gaussian-distributed entries, the ensemble appropriate to time-reversal-invariant systems. The spacings of GOE eigenvalues then follow the Wigner surmise $P(s)=(\pi/2)s\,e^{-\pi s^2/4}$ for $s$ in units of the mean spacing, whose exact GOE counterpart was computed by Mehta and Gaudin in terms of determinants of integral operators (Fredholm determinants); and the reduced width amplitudes, being projections of GOE eigenvectors, are Gaussian, so the reduced widths follow the Porter-Thomas distribution — a chi-square law with one degree of freedom, $P(x)=(2\pi x)^{-1/2}e^{-x/2}$ for $x=\gamma^2/\overline{\gamma^2}$. Haq, Pandey and Bohigas found that the nuclear data ensemble (the compilation of resolved neutron resonances across many nuclei) agrees with GOE statistics, and Bohigas, Giannoni and Schmit conjectured the general link between classically chaotic dynamics and random-matrix statistics. In the stochastic S-matrix theory of compound-nucleus reactions, the GOE is embedded inside R-matrix scattering theory and all fluctuation properties of the cross section are computed exactly; channel physics enters only through the average scattering matrix, or equivalently the transmission coefficients $T_c$ (which measure how open each decay channel $c$ is), and the construction extends to overlapping resonances, where widths become comparable to the spacing — a regime Wigner's analysis excluded. This is the precise modern descendant of his separation between internal statistics and long-range channel physics.

What survives of the 1951 question is then sharply posed. First, universality: the GOE and Porter-Thomas laws are postulates about the internal dynamics. Level-correlation statistics are universal for matrix ensembles with smoothly varying element distributions (Kamien, Politzer and Wise), but width and eigenvector statistics are in that generality system-dependent, and rigorous universality theorems cover random-matrix ensembles rather than deterministic nuclear Hamiltonians; no derivation of the laws from the underlying many-body dynamics exists, and Wigner's invariance principle for general statistical R-functions has never been proved as a theorem. Second, the width law itself is contested: several analyses of reduced neutron widths (in the platinum isotopes and in the nuclear data ensemble) reject the Porter-Thomas distribution with stated high significance, and so do partial gamma-decay widths in $^{150}$Nd measured by nuclear resonance fluorescence (exciting nuclear levels with monochromatic gamma rays and observing their decay); the proposed explanations invoke channel-coupling terms that break the orthogonal invariance underlying the law — such as the Thomas-Ehrman shift, a threshold-dependent energy shift of weakly bound levels caused by coupling to the particle continuum — or non-statistical radiative decay, or the realistic energy dependence of the neutron channel, or statistical artifacts of the normalization; these accounts currently stand unreconciled.

Problem Statement

Determine the statistical laws of compound-nucleus resonance spectra and the exact domain of their invariance under the long-range channel interaction, as posed by Wigner (1951) and sharpened by the intervening literature. Concretely:

(1) Determine the actual probability distributions of the normalized spacings $(H_{\mu+1}-H_\mu)/D$ (with $D$ the local mean spacing of the observed resonance positions $H_\mu$) and of the normalized widths $w_\mu/\bar{w}$ of compound-nucleus resonances (resonance widths in units of their local average, equivalently — when the penetration factor is effectively constant across the window — the reduced widths $\gamma_\mu^2$ in units of theirs) — their functional forms or exact implicit characterizations — for the standard regime of isolated resonances (level spacings large compared to widths) and for the regimes the 1951 analysis excluded: open reaction channels and overlapping resonances, appreciable energy variation of the channel boundary functions across the resonance window, and the neighborhood of maxima or thresholds of the neutron strength function.

(2) Establish Wigner's invariance principle as a theorem with precise scope, or delimit exactly where it fails: specify the class of statistical R-functions (or of underlying dynamical models) for which the normalized spacing and width distributions are invariant under the real Möbius transformations induced by changes of the channel boundary functions, and quantify how the distributions change when those conditions fail.

(3) Reconcile theory with the measured width statistics: either derive from stated microscopic or ensemble assumptions a width distribution — including channel-coupling and non-statistical corrections — that accounts for the reported deviations from the Porter-Thomas law in reduced neutron widths (platinum isotopes, nuclear data ensemble) and in partial gamma-decay widths, or demonstrate rigorously that the reported deviations are artifacts of the statistical normalization and analysis.

These three components serve one objective: a universal, channel-independent statistical description of compound-nucleus resonance spectra, validated against data — component (1) fixes the laws, component (2) fixes their invariance domain, and component (3) confronts both with experiment. The known standard-regime laws (the Wigner/GOE spacing law and the Porter-Thomas width law, derived under the ensemble hypothesis) are the baseline, not the answer: what remains open is the invariance domain, the width law under realistic channel coupling, and their reconciliation with measurement. These three components are jointly posed and together constitute the problem.

References

1. Eugene P. Wigner, 'On the statistical distribution of the widths and spacings of nuclear resonance levels', Mathematical Proceedings of the Cambridge Philosophical Society 47(4), 790-798 (1951), DOI 10.1017/S0305004100027237, https://doi.org/10.1017/S0305004100027237
2. E. P. Wigner, 'On a Class of Analytic Functions from the Quantum Theory of Collisions', Annals of Mathematics 53, 36 (1951), DOI 10.2307/1969342, https://doi.org/10.2307/1969342
3. A. M. Lane and R. G. Thomas, 'R-Matrix Theory of Nuclear Reactions', Reviews of Modern Physics 30, 257 (1958), DOI 10.1103/RevModPhys.30.257, https://doi.org/10.1103/RevModPhys.30.257
4. C. E. Porter and R. G. Thomas, 'Fluctuations of Nuclear Reaction Widths', Physical Review 104, 483 (1956), DOI 10.1103/PhysRev.104.483, https://doi.org/10.1103/PhysRev.104.483
5. E. P. Wigner, 'On the Distribution of the Roots of Certain Symmetric Matrices', Annals of Mathematics 67, 325 (1958), DOI 10.2307/1970008, https://doi.org/10.2307/1970008
6. M. L. Mehta, 'On the statistical properties of the level-spacings in nuclear spectra', Nuclear Physics 18, 395 (1960), DOI 10.1016/0029-5582(60)90413-2, https://doi.org/10.1016/0029-5582(60)90413-2
7. M. Gaudin, 'Sur la loi limite de l'espacement des valeurs propres d'une matrice aléatoire', Nuclear Physics 25, 447-458 (1961), DOI 10.1016/0029-5582(61)90176-6, https://doi.org/10.1016/0029-5582(61)90176-6
8. F. J. Dyson, 'Statistical Theory of the Energy Levels of Complex Systems. I', Journal of Mathematical Physics 3, 140 (1962), DOI 10.1063/1.1703773, https://doi.org/10.1063/1.1703773
9. R. U. Haq, A. Pandey and O. Bohigas, 'Fluctuation Properties of Nuclear Energy Levels: Do Theory and Experiment Agree?', Physical Review Letters 48, 1086 (1982), DOI 10.1103/PhysRevLett.48.1086, https://doi.org/10.1103/PhysRevLett.48.1086
10. O. Bohigas, M. J. Giannoni and C. Schmit, 'Characterization of Chaotic Quantum Spectra and Universality of Level Fluctuation Laws', Physical Review Letters 52, 1 (1984), DOI 10.1103/PhysRevLett.52.1, https://doi.org/10.1103/PhysRevLett.52.1
11. J. J. M. Verbaarschot, H. A. Weidenmüller and M. R. Zirnbauer, 'Grassmann integration in stochastic quantum physics: The case of compound-nucleus scattering', Physics Reports 129, 367-438 (1985), DOI 10.1016/0370-1573(85)90070-5, https://doi.org/10.1016/0370-1573(85)90070-5
12. Randall D. Kamien, H. David Politzer and Mark B. Wise, 'Universality of Random-Matrix Predictions for the Statistics of Energy Levels', Physical Review Letters 60, 1995 (1988), DOI 10.1103/PhysRevLett.60.1995, https://doi.org/10.1103/PhysRevLett.60.1995
13. G. E. Mitchell, A. Richter and H. A. Weidenmüller, 'Random matrices and chaos in nuclear physics: Nuclear reactions', Reviews of Modern Physics 82, 2845-2901 (2010), DOI 10.1103/RevModPhys.82.2845, https://doi.org/10.1103/RevModPhys.82.2845
14. P. E. Koehler, F. Bečvář, M. Krtička, J. A. Harvey and K. H. Guber, 'Anomalous Fluctuations of s-Wave Reduced Neutron Widths of $^{192,194}$Pt Resonances', Physical Review Letters 105, 072502 (2010), DOI 10.1103/PhysRevLett.105.072502, https://doi.org/10.1103/PhysRevLett.105.072502
15. H. A. Weidenmüller, 'Distribution of Partial Neutron Widths for Nuclei Close to a Maximum of the Neutron Strength Function', Physical Review Letters 105, 232501 (2010), DOI 10.1103/PhysRevLett.105.232501, https://doi.org/10.1103/PhysRevLett.105.232501
16. P. E. Koehler, 'Reduced neutron widths in the nuclear data ensemble: Experiment and theory do not agree', Physical Review C 84, 034312 (2011), DOI 10.1103/PhysRevC.84.034312, https://doi.org/10.1103/PhysRevC.84.034312
17. G. L. Celardo, N. Auerbach, F. M. Izrailev and V. G. Zelevinsky, 'Distribution of Resonance Widths and Dynamics of Continuum Coupling', Physical Review Letters 106, 042501 (2011), DOI 10.1103/PhysRevLett.106.042501, https://doi.org/10.1103/PhysRevLett.106.042501
18. H. A. Weidenmüller, 'Open problems in applying random-matrix theory to nuclear reactions', Journal of Physics G: Nuclear and Particle Physics 41, 094010 (2014), DOI 10.1088/0954-3899/41/9/094010, https://doi.org/10.1088/0954-3899/41/9/094010
19. Alexander Volya, Hans A. Weidenmüller and Vladimir Zelevinsky, 'Neutron Resonance Widths and the Porter-Thomas Distribution', Physical Review Letters 115, 052501 (2015), DOI 10.1103/PhysRevLett.115.052501, https://doi.org/10.1103/PhysRevLett.115.052501
20. P. Fanto, Y. Alhassid and H. A. Weidenmüller, 'The Statistical Model of Nuclear Reactions: Open Problems', arXiv:1810.13385 (2018), https://arxiv.org/abs/1810.13385
21. P. Fanto, G. F. Bertsch and Y. Alhassid, 'Neutron width statistics in a realistic resonance-reaction model', Physical Review C 98, 014604 (2018), DOI 10.1103/PhysRevC.98.014604, https://doi.org/10.1103/PhysRevC.98.014604
22. H.-L. Harney and H. A. Weidenmüller, 'Tests of the Porter-Thomas Distribution for Reduced Partial Neutron Widths', arXiv:2106.06311 (2021), https://arxiv.org/abs/2106.06311
23. O. Papst, J. Isaak, V. Werner, D. Savran, N. Pietralla, G. Battaglia, T. Beck, M. Beuschlein, S. W. Finch, U. Friman-Gayer, K. E. Ide, R. V. F. Janssens, M. D. Jones, J. Kleemann, B. Löher, M. Scheck, M. Spieker, W. Tornow, R. Zidarova and A. Zilges, 'Deviations from the Porter-Thomas Distribution due to Nonstatistical $\gamma$ Decay below the $^{150}$Nd Neutron Separation Threshold', Physical Review Letters 135, 052501 (2025), DOI 10.1103/n6zw-zhz2, https://doi.org/10.1103/n6zw-zhz2
24. K. Fujio, T. Kawano, A. E. Lovell and D. Neudecker, 'Statistical properties of neutron-induced reaction cross sections using a random-matrix approach', Physical Review C 112, 064608 (2025), DOI 10.1103/mvz3-46vy, https://doi.org/10.1103/mvz3-46vy (preprint arXiv:2510.00360, https://arxiv.org/abs/2510.00360)
25. L. Erdős and H.-T. Yau, 'Universality of local spectral statistics of random matrices', Bulletin of the American Mathematical Society 49, 377-414 (2012), DOI 10.1090/s0273-0979-2012-01372-1, https://doi.org/10.1090/s0273-0979-2012-01372-1