---
title: Regge Poles in Scattering Theory
url: https://www.emergentmind.com/topics/regge-poles
type: topic
---

# Regge Poles in Scattering Theory

Regge poles are poles of a scattering amplitude or $S$-matrix after analytic continuation of the angular-momentum variable from integer values to the complex angular-momentum (CAM) plane. At fixed real frequency or energy, they occur at complex values of a shifted angular momentum, commonly $\lambda=\ell+\tfrac12$, and define Regge trajectories $\lambda_n(\omega)$ or $\alpha(t)$. They provide a spectral description complementary to ordinary resonances and quasinormal modes: resonances are poles in complex frequency at fixed integer angular momentum, whereas Regge poles are poles in complex angular momentum at fixed real frequency. Their residues determine excitation strengths, and their locations control phase accumulation, attenuation, diffraction, high-energy asymptotics, and the organization of resonant spectra.

## 1. Historical and analytic foundations

The theory of Regge poles originated in the S-matrix program of the late 1950s and early 1960s, which sought to constrain scattering amplitudes through unitarity, analyticity, and crossing symmetry rather than through a perturbative expansion around a microscopic interaction. Tullio Regge’s decisive step was to regard the integer angular momentum $\ell$ as the restriction of a function defined for complex angular momentum. The convenient shifted variable is

$$
\lambda=\ell+\frac12.
$$

For a partial-wave expansion,

$$
f(k,\cos\theta)
 =
\frac{1}{2ik}
\sum_{\ell=0}^{\infty}
(2\ell+1)
\left[S_\ell(k)-1\right]P_\ell(\cos\theta),
$$

the Sommerfeld–Watson transformation replaces the discrete sum by a contour integral in the complex $\lambda$-plane. The poles of the analytically continued partial-wave $S$-matrix are then exposed when the contour is deformed. A Regge pole is locally represented by

$$
S(\lambda,k)\sim
\frac{\beta_n(k)}{\lambda-\lambda_n(k)},
$$

where $\lambda_n(k)$ is the pole position and $\beta_n(k)$ is its residue. The trajectories $\lambda_n(k)$ move as the energy or momentum changes.

In potential scattering, Regge poles can equivalently be defined through Jost functions. For a radial Schrödinger operator with complexified angular momentum $v$, the regular solution can be expanded in outgoing and incoming Jost solutions. If $\beta(v)$ denotes the relevant Jost function, then

$$
\text{Regge poles are the zeros of }\beta(v),
$$

or, equivalently, poles of the meromorphic Regge interpolation of the generalized phase shift [1502.02276]. Under the analytic and short-range assumptions studied there, the physically relevant poles lie in the first quadrant of the complex angular-momentum plane.

The distinction between Regge poles and other singularities is fundamental. Bound-state poles are poles in the complex momentum plane at fixed angular momentum; ordinary resonances are generally poles in the complex energy plane; branch points and threshold singularities generate cuts in energy or momentum; Regge poles are poles in complex angular momentum while energy or momentum is held fixed. In relativistic scattering, a trajectory $\ell=\alpha(t)$ can connect angular momentum to momentum transfer. A pole then produces high-energy behavior of the form

$$
f(s,t)\sim C(t)s^{\alpha(t)}
$$

at fixed $t$ and large $s$ [1807.02456].

## 2. Definition in wave scattering and relation to resonances

For a fixed real frequency $\omega>0$, the partial-wave scattering matrix is analytically continued from integer $\ell$ to complex $\lambda$. The Regge poles in the first quadrant are denoted

$$
\lambda_n(\omega),\qquad n=1,2,\ldots.
$$

In black-hole and compact-object scattering, the radial solution is usually selected to be purely ingoing at the horizon or regular at the center, while at infinity it is decomposed into incoming and outgoing components. For a Schwarzschild black hole,

$$
\phi_{\omega\ell}^{\rm in}
\sim
e^{-i\omega r_*},
\qquad r_*\to-\infty,
$$

and

$$
\phi_{\omega\ell}^{\rm in}
\sim
A_\ell^{(-)}(\omega)e^{-i\omega r_*}
+
A_\ell^{(+)}(\omega)e^{+i\omega r_*},
\qquad r_*\to+\infty.
$$

The analytically continued $S$-matrix is

$$
S_{\lambda-1/2}(\omega)
=
e^{i(\lambda+1/2)\pi}
\frac{A_{\lambda-1/2}^{(+)}(\omega)}
{A_{\lambda-1/2}^{(-)}(\omega)}.
$$

A Regge pole satisfies

$$
A_{\lambda_n(\omega)-1/2}^{(-)}(\omega)=0.
$$

The corresponding Regge mode has no incoming component at infinity: it is purely outgoing there and satisfies the appropriate ingoing or regularity condition on the inner boundary. Its residue is determined by the derivative of the incoming coefficient,

$$
r_n(\omega)=
e^{i\pi[\lambda_n(\omega)+1/2]}
\left[
\frac{A_{\lambda-1/2}^{(+)}(\omega)}
{\dfrac{d}{d\lambda}A_{\lambda-1/2}^{(-)}(\omega)}
\right]_{\lambda=\lambda_n(\omega)}.
$$

The same radial equation generates quasinormal modes (QNMs), but the spectral variables are interchanged. QNMs have integer $\ell$ and complex frequencies

$$
\omega_{\ell n}
=
\omega_{\ell n}^{(o)}
-\frac{i}{2}\Gamma_{\ell n},
$$

whereas Regge poles have real $\omega$ and complex $\lambda$. Thus the two spectra are dual analytic manifestations of the same resonant scattering problem [0906.2601].

The connection can be expressed by a Bohr–Sommerfeld-type condition. For Schwarzschild spacetime,

$$
\operatorname{Re}\lambda_n
\left(\omega_{\ell n}^{(0)}\right)
=
\ell+\frac12.
$$

More generally, in $d$ dimensions the condition is

$$
\operatorname{Re}\lambda_n
\left(\omega_{\ell n}^{(0)}\right)
=
\ell+\frac{d-3}{2}.
$$

The imaginary part of the Regge trajectory determines the resonance width. In the regime where the frequency dependence of $\operatorname{Im}\lambda_n$ is weak,

$$
\frac{\Gamma_{\ell n}}{2}
\simeq
\left.
\frac{\operatorname{Im}\lambda_n(\omega)}
{d\operatorname{Re}\lambda_n/d\omega}
\right|_{\omega=\omega_{\ell n}^{(0)}}.
$$

Consequently, one Regge trajectory generates a family of QNMs indexed by $\ell$, while the pole index $n$ labels distinct surface-wave or resonance families.

## 3. Photon-sphere surface waves and WKB trajectories

For a static, spherically symmetric black hole,

$$
ds^2=-f(r)\,dt^2+\frac{dr^2}{f(r)}+r^2d\sigma_{d-2}^2,
$$

a photon sphere at $r=r_c$ satisfies

$$
f'(r_c)-\frac{2f(r_c)}{r_c}=0,
$$

with instability condition

$$
f''(r_c)-\frac{2f(r_c)}{r_c^2}<0.
$$

The associated instability parameter is

$$
\eta_c=
\frac12
\sqrt{4f_c-2r_c^2f_c^{(2)}},
\qquad f_c=f(r_c),
$$

and is related to the Lyapunov exponent of the null orbit. The leading Regge trajectory is

$$
\lambda_n(\omega)
\sim
\frac{r_c}{\sqrt{f_c}}\omega
+i\eta_c\left(n-\frac12\right).
$$

Its real part is controlled by the orbital period of the unstable null orbit, while its imaginary part is controlled by the instability and leakage of waves trapped near that orbit [1002.0121].

For Schwarzschild spacetime, $r_c=3M$, $\eta_c=1$, and

$$
\lambda_n(\omega)
\sim
3\sqrt3\,M\omega
+i\left(n-\frac12\right).
$$

A third-order WKB treatment gives

$$
\lambda_n(\omega)\approx
\left[
\frac{r_c^2}{f_c}\omega^2
+a_n
+2\eta_c^2\left(n-\frac12\right)^2\epsilon_n(\omega)
\right]^{1/2}
+i\eta_c\left(n-\frac12\right)
\left[1+\epsilon_n(\omega)\right],
$$

where

$$
\epsilon_n(\omega)=
\frac{b_n}
{(r_c^2/f_c)\omega^2+a_n+\eta_c^2(n-\frac12)^2}.
$$

The coefficients $a_n$ and $b_n$ depend on the dimension, the metric function and its derivatives at the photon sphere, and the overtone index. At high frequency,

$$
\operatorname{Re}\lambda_n
\sim
\frac{r_c}{\sqrt{f_c}}\omega,
\qquad
\operatorname{Im}\lambda_n
\sim
\eta_c\left(n-\frac12\right),
$$

while the subleading terms generate nonlinear dispersion and frequency-dependent attenuation.

For Schwarzschild perturbations of spin $s=0,1,2$, the high-frequency expansion is

$$
\lambda_n(\omega)
=
3\sqrt3\,M\omega
+\frac{\sqrt3\,a_n}{18M\omega}
+i\left(n-\frac12\right)
\left[
1+\frac{b_n}{27(M\omega)^2}
\right]
+\mathcal O(\omega^{-3}).
$$

The leading trajectory is approximately spin-independent, but the coefficients $a_n$ and $b_n$ distinguish scalar, electromagnetic, and gravitational perturbations [0906.2601].

The pole contribution to a scattered field contains factors such as

$$
\exp\!\left[i\lambda_n(\omega)\theta-i\omega t\right].
$$

With arc length $\mathcal L=r_c\theta$, this becomes

$$
\exp[-k_n''(\omega)\mathcal L]\,
\exp\!\left[i(k_n'(\omega)\mathcal L-\omega t)\right],
$$

where

$$
k_n'(\omega)=
\frac{\operatorname{Re}\lambda_n(\omega)}{r_c},
\qquad
k_n''(\omega)=
\frac{\operatorname{Im}\lambda_n(\omega)}{r_c}.
$$

The quantities $k_n'$ and $k_n''$ are respectively the azimuthal propagation constant and attenuation constant. The terminology “surface wave” is a wave-theoretic scattering analogy: these are not material waves on a physical surface, but waves temporarily localized near an unstable photon orbit and leaking toward the horizon and infinity.

## 4. CAM representations, residues, and scattering phenomena

The Sommerfeld–Watson transformation expresses a partial-wave amplitude as a background integral plus Regge-pole residues. For scalar Schwarzschild scattering,

$$
f(\omega,\theta)
=
\frac{1}{2\omega}
\int_{\mathcal C}d\lambda\,
\frac{\lambda}{\cos(\pi\lambda)}
\left[S_{\lambda-1/2}(\omega)-1\right]
P_{\lambda-1/2}(-\cos\theta).
$$

After contour deformation,

$$
f(\omega,\theta)
=
f^{\rm B}(\omega,\theta)
+
f^{\rm RP}(\omega,\theta),
$$

where the Regge-pole contribution is

$$
f^{\rm RP}(\omega,\theta)
=
-\frac{i\pi}{\omega}
\sum_{n=1}^{\infty}
\frac{\lambda_n(\omega)r_n(\omega)}
{\cos[\pi\lambda_n(\omega)]}
P_{\lambda_n(\omega)-1/2}(-\cos\theta).
$$

The full background-plus-pole representation is equivalent to the original partial-wave expansion. A finite pole sum without the background is an approximation whose accuracy depends on frequency, angle, and the number of poles retained [1901.03965].

At intermediate and high frequencies, Regge poles efficiently resum information distributed over many partial waves. A small number of low-lying poles describes backward glory scattering, while additional poles reproduce orbiting oscillations over small and intermediate scattering angles. In the short-wavelength regime, the background integral is often numerically small, although it remains important at low frequencies and in the extreme forward direction.

For electromagnetic waves, spin dependence enters through an angular differential operator and a factor $(\lambda^2-\tfrac14)^{-1}$. For gravitational waves, the even- and odd-parity $S$-matrices have common Regge-pole positions but different residues. The helicity-preserving and helicity-reversing amplitudes involve respectively

$$
r_n^{(e)}+r_n^{(o)}
\quad\text{and}\quad
r_n^{(e)}-r_n^{(o)}.
$$

The high-frequency gravitational glory is described by a Bessel function of order four, reflecting the spin-2 character of the field [1906.01441].

Regge poles also enter gravitational radiation from a source. For a particle falling radially into a Schwarzschild black hole, a CAM representation of the waveform $\Psi_4$ contains a Fourier transform of Regge-pole contributions weighted by excitation factors. A single low-lying Regge pole can reproduce much of the QNM ringdown, while additional poles improve the description when many multipoles and overtones are excited. Unlike an isolated QNM expansion, the Regge-pole Fourier integral does not require an externally imposed ringdown starting time [1807.09056].

The same structure appears in absorption. The exact absorption cross section for spin $s$ is

$$
\sigma_s(\omega)
=
\frac{\pi}{\omega^2}
\sum_{\ell=s}^{\infty}
(2\ell+1)\Gamma_{\ell,s}(\omega),
$$

where $\Gamma_{\ell,s}$ is the greybody factor. CAM continuation decomposes it into smooth real- and imaginary-axis backgrounds and a discrete Regge-pole series. At high frequency, the real-axis background approaches the geometric capture cross section

$$
\sigma_{\rm geo}=27\pi M^2,
$$

while the first Regge pole supplies the leading oscillatory correction. For spin $s=0,1,2$, subleading phase corrections depend on $s$ and generate fine structure in the absorption spectrum [2504.19324].

## 5. Generalizations beyond isolated Schwarzschild black holes

The geometric framework extends to a broad class of static, spherically symmetric black holes, including Schwarzschild–Tangherlini, Reissner–Nordström, canonical acoustic, and Schwarzschild–de Sitter geometries. The leading trajectory depends on the photon-sphere radius, the redshift factor $f_c$, and the instability parameter $\eta_c$.

For a Schwarzschild–Tangherlini metric,

$$
f(r)=1-\left(\frac{r_0}{r}\right)^{d-3},
$$

one has

$$
r_c=r_0\left(\frac{d-1}{2}\right)^{1/(d-3)},
\qquad
\eta_c=\sqrt{d-3}.
$$

For a canonical acoustic hole with

$$
f(r)=1-\left(\frac{r_0}{r}\right)^4,
$$

the photon-sphere radius and instability parameter are

$$
r_c=3^{1/4}r_0,
\qquad
\eta_c=2.
$$

The associated Regge poles and QNMs have the same geometric organization, with the orbital frequency determined by $r_c/\sqrt{f_c}$ and the damping determined by $\eta_c$ [1002.0121]. Direct numerical calculations for the canonical acoustic hole confirm that inverse-$L$ expansions become accurate for moderate and large angular momentum, while their accuracy deteriorates for low multipoles and for $l$ comparable to or smaller than the overtone number [1407.3904].

Compact objects produce additional branches of poles. For neutron-star-like objects, the spectrum can contain surface-wave poles and broad resonances. For ultracompact objects with $R<3M$, a trapping cavity between the surface and the light-ring barrier produces narrow resonances with small imaginary parts. In models with a discontinuity in the effective potential, the broad-resonance imaginary part is controlled by the discontinuity, while the real part is controlled by the cavity size and surface position [1912.11348].

A Schwarzschild black hole surrounded by a thin spherical shell can exhibit inner surface waves, outer surface waves, creeping modes associated with the shell, and broad resonances. The inner and outer surface-wave branches are associated with distinct light rings, while shell-guided modes arise when the shell lies outside the outer light ring. The shell position controls the real parts of broad resonances, and the shell stress-energy and potential discontinuity control their imaginary parts [2211.17147].

Recent analyses of black-hole potentials with finite discontinuities distinguish spectral instability from observable instability. Higher-order Regge poles can bifurcate and deform strongly when a discontinuity is introduced, while low-lying poles and integrated absorption cross sections remain comparatively stable. In the high-frequency regime, the geometric-optics background and first Regge pole dominate, suppressing the effect of strongly deformed high-order poles [2504.13265].

## 6. Inverse problems, localization, and mathematical structure

Regge poles are also central to inverse scattering and inverse spectral problems. For radial Schrödinger operators at fixed energy, the complexified angular momentum enters the centrifugal term,

$$
-u''(r)
+
\left(
\frac{v^2-\frac14}{r^2}+q(r)
\right)u(r)
=
u(r).
$$

The Regge poles are zeros of the analytically continued Jost function. Their localization depends strongly on the decay and analytic properties of the potential. For nonzero super-exponentially decreasing potentials, there are infinitely many Regge poles whose real parts are unbounded in the first quadrant. For compactly supported potentials, the poles approach the positive imaginary axis with asymptotics

$$
\operatorname{Re}v_p
=
\frac{\pi^2}{2}
\frac{p}{(\log p)^2}
\bigl(1+o(1)\bigr),
\qquad
\operatorname{Im}v_p
=
\frac{\pi p}{\log p}
\bigl(1+o(1)\bigr).
$$

By contrast, for potentials analytically extendable into the right half-plane with polynomial decay, the Regge poles are bounded to the right in the first quadrant [1502.02276].

These distinctions determine which analytic continuation and uniqueness arguments are available. Localization of the poles allows Hardy-class, Phragmén–Lindelöf, Cartwright, Paley–Wiener, and Borg–Marchenko-type methods to be applied to phase shifts and Jost functions. Under the stated assumptions, super-exponentially close phase shifts imply equality of the potentials. For compactly supported perturbations added to analytically decaying backgrounds, sharper high-angular-momentum phase-shift asymptotics yield local uniqueness outside a prescribed radius.

A related inverse problem arises on warped balls. Separation of variables converts the fixed-energy Dirichlet problem into a half-line Schrödinger equation with spectral parameter $-z^2$. The Regge poles are poles of the meromorphic Weyl–Titchmarsh function

$$
M(-z^2)=\frac{\psi'(0,z)}{\psi(0,z)},
$$

or equivalently zeros of the Jost function $\psi(0,z)$. Under compact-support and regularity assumptions, the poles split into two families: real poles near the negative integers, generated primarily by the universal exponentially decaying background, and complex strings whose asymptotics depend on the support endpoint and the derivative discontinuity of the perturbation. The full Regge-pole set determines the associated Schrödinger potential and, with boundary data, the warping function [2203.13850].

## 7. Relativistic Regge theory, Regge cuts, and open issues

In relativistic scattering, a Regge pole at

$$
j=\alpha(t)
$$

produces an amplitude behaving as

$$
A(s,t)\sim s^{\alpha(t)}.
$$

The trajectory can interpolate resonance spins through

$$
\alpha(s_p)=J_p,
$$

where $s_p$ is a complex resonance pole. Finite widths require the complex trajectory to reproduce both the real and imaginary parts of the resonance condition. Fits to $N^*$ and $\Delta^*$ poles therefore use complex trajectory models rather than only Chew–Frautschi plots of $J$ against $\operatorname{Re}s_p$. The real-part trajectories are approximately linear, whereas the imaginary parts exhibit an approximate square-root behavior associated with decay phase space [1809.01954].

Regge poles do not exhaust the singularity structure. Regge cuts arise from multi-Reggeon exchange and correspond to branch-point singularities in the complex angular-momentum plane. In perturbative QCD, the negative-signature Reggeized gluon is a Regge pole, while the BFKL Pomeron is a two-Reggeon cut in the positive-signature sector. At NNLL accuracy, three-Reggeon exchange produces an odd-signature Regge cut and violates simple pole factorization. A practical pole–cut prescription identifies the pole with single-Reggeon exchange plus planar multi-Reggeon exchange, while the nonplanar remainder is assigned to the cut [2112.11098].

The separation is important because the Reggeized-gluon pole factorizes into impact factors and an exponentiated trajectory, whereas the cut contains nontrivial color mixing and additional infrared singularities. The infrared poles of the Regge trajectory are governed by the cusp anomalous dimension, while the cut probes non-dipole correlations in the soft anomalous dimension. In $2\to3$ scattering, the Lipatov vertex occupies the factorized structure between two Reggeized-gluon exchanges; multi-Reggeon contributions must be removed before extracting its pole contribution [2412.20577].

The distinction between poles and cuts is not universally resolved in all relativistic settings. In QCD, the precise separation in the adjoint channel remains dependent on perturbative organization, color representation, and infrared factorization. In meromorphic dual amplitudes, crossing symmetry and positivity impose a different constraint: under the assumptions studied in [2405.21057], a finite number of Regge trajectories cannot produce a reggeizing amplitude. The argument implies that an infinite collection of trajectories is required, as in the Veneziano amplitude, because an infinite sum of individually regular contributions can collectively generate a complex-angular-momentum pole.

Regge-pole theory therefore encompasses several related but distinct structures: isolated CAM poles in potential and black-hole scattering; trajectories connecting resonance poles and high-energy behavior; photon-sphere surface waves and their residues; inverse-scattering data encoded in Jost and Weyl–Titchmarsh functions; and Regge cuts generated by multi-Reggeon dynamics. Across these settings, the common principle is analytic continuation in angular momentum: the singularities of the continued scattering data reorganize partial waves into trajectories, residues, backgrounds, and cuts whose physical manifestations include resonances, quasinormal ringing, orbiting, glory and rainbow scattering, absorption oscillations, high-energy asymptotics, and inverse spectral information.

Source: https://www.emergentmind.com/topics/regge-poles