WorksPhysics and Astronomy
Spurious center-of-mass mode as an -matrix pole within the Jost-random-phase approximation
Abstract
Within the Jost-random-phase approximation (Jost-RPA), the spurious center-of-mass mode is a zero of the Jost determinant and hence a pole of the matrix. A zero at zero excitation energy , whether from self-consistency or from a calibration of the residual interaction, is a pole of exactly second order when it has a single independent solution, every unoccupied bound orbital lies above the holes, and each hole is paired with every allowed partial wave; the order follows from the position of the zero, not from self-consistency. The state of this double pole, given by the Autonne-Takagi factorization of its coefficient, is the zero-energy solution normalized by its inertia. In a self-consistent model of one energy density of the neutron and proton densities, every physical -matrix element continued to carries the term, contrary to an earlier conclusion. Subtracting the mode's principal part from the response changes the strength at real only through the smoothing and, in such a model, coincides with the projection of the ideal spurious state. In uncalibrated Woods-Saxon models of O and Pb the mode is a pair of poles on the imaginary axis.
Introduction
In the random-phase approximation (RPA) on a self-consistent mean field, the translation of the nucleus is a solution at zero excitation energy , the spurious center-of-mass (c.m.) mode [1]. It is not an intrinsic excitation, yet it enters the response to external fields. In practice, discrete RPA and its extensions give a finite real or an imaginary eigenvalue [2, 3, 4, 5, 6]. Continuum calculations place it on the real axis as a bound state of the Green function [7]; when it is imaginary, only that fact has been reported [7, 8, 9, 10]. Remedies rescale the residual interaction to restore zero energy [11, 12, 13, 14, 15], remove the c.m. coordinate from the external field [9, 10, 16, 17], project the ideal spurious state out of the response [18], or remove its coupling from the RPA matrices [19].
Is the mode a pole of the matrix? Ginocchio, Schucan, and Weidenmüller derived the RPA matrix of nucleon–nucleus scattering and answered no [20]. Finite-Fermi-system theory treats c.m. oscillations in an external field as poles of the vertex function [21], and continuum RPA with a separable force fits the coupling so that a dispersion function vanishes at [22]. In discrete RPA a zero mode with nonzero inertia forms a Jordan pair [23, 24, 25], which gives a factor in the RPA determinant and an term in the response [26, 27], approximately also in a semimicroscopic continuum model [28]. These concern the response, whereas the 1970 question concerns the physical -matrix elements. As an -matrix pole the mode has been shown only with the separable-force fit, never as the double pole of a self-consistent model or as the imaginary pair, whose energy has not been reported.
This Letter answers yes within the Jost-RPA [29, 30, 31], which keeps the scattering boundary conditions of every channel and finds the -matrix poles among the zeros of the Jost determinant: the spurious mode is such a zero. At , whether from self-consistency or from a calibration of the residual interaction, it is, with one state, a pole of exactly second order, and in a self-consistent model every physical -matrix element continued to has this pole, contrary to Ref. [20]. In the Woods–Saxon model used here, which approximates a self-consistent theory, the mode is a pair on the imaginary axis. Subtracting the principal part of the response function at these poles, its Mittag-Leffler terms, changes the strength at real only through the smoothing; in a self-consistent model it equals the projection of Ref. [18], and for an imaginary pair it needs no calibration. Pole expansions of responses and matrices must contain these poles, whose terms carry 90–95% of the energy-weighted sum rule of the field here.
Jost-RPA
At multipolarity , each of the particle-hole configurations [29, 30, 31] pairs a hole of isospin , angular momenta , and energy with a particle partial wave and gives a forward channel and a backward channel , with and the wave numbers , of the nucleon energies , . The Green function, regular at and outgoing at large , obeys
with the nucleon mass , the unit matrix , the Pauli matrices , on the forward and backward blocks, , and the mean field of isospin in the partial wave ; the radial hole functions solve ${-(\hbar^2/2m)[d^2/dr^2-l_{h_\alpha}(l_{h_\alpha}+1)/r^2]+\ . The residual interaction and the response function are
with the real contact strengths , the spherical harmonics , the vector whose component is times for and 0 otherwise, and .
The regular solution and the Jost solutions solve with, for all channels ,
where and are the spherical Bessel and Neumann functions and . The mean field and the residual interaction vanish beyond a radius, where each column of is the Hankel function of its own channel alone; this fixes the solutions and the Jost matrices defined by
With the step function , the diagonal matrix of the , in , and , the flux-adjusted matrix and the Green function are
Throughout, lies on the first Riemann sheet, where in every channel, with equality only on the cuts of the real axis; off the cuts, decays and grows. On the upper side of the cut at the open-channel block of is the physical matrix. At a zero of off the cuts, a common nonzero left null vector of and would, by Eq. (8), give a vanishing combination of the columns of , which are independent by Eq. (6); so the zero is a pole of and, since are analytic there and only the first term of Eq. (10) contains , a pole of , as a function of and , of the same order. Conversely, a solution regular at that decays is with by Eq. (8); a pole has one state if all such solutions at it are multiples of one. The Wronskian of with itself vanishes, so Eq. (8) and the Wronskians of the Jost solutions give , and at a simple pole with one state the Autonne–Takagi factorization defines the state,
at a pole of higher order the coefficient matrix of the most negative power of replaces the residue matrix. Since , , and , the matrix solves the equation at with Eqs. (6), (7), and Eq. (8) gives
so is even in : zeros come in pairs , as found for the RPA matrix in Ref. [20], and a zero at has even order.
The spurious mode as a zero of the Jost determinant. For , a zero at with one state is double, even without self-consistency, under four premises: natural parity, so that ; every occupied orbital is a hole; (S1) every unoccupied bound orbital lies strictly above the holes of its isospin; and (S2) the configurations of each hole contain every allowed , occupied ones included. For a channel function , let be the solution of that is regular at and decays. The rigid shift of the occupied orbitals has the channel function
with the component of configuration and . By (S2) its transition density is , with the ground-state density of isospin .
A pair of : Under (S1), (S2), the zeros of off the cuts lie on the real or the imaginary axis, and a negative restoring energy of the rigid shift gives an imaginary pair. Let be a zero off the cuts and any nonzero solution of that is regular at and decays, i.e., with . Let be its forward and backward parts and . The difference of the two rows gives , and their sum then gives
Multiplying by and integrating gives
Both integrals are real, as and are real symmetric and there are no surface terms. The following pair argument shows that the last integral is positive unless $E_g is real, so is real. An occupied orbital of isospin and partial wave is an eigenfunction of with eigenvalue . Unless , which is real, (15) and give the component
As is symmetric, this component contributes times its squared modulus. By (S2) the configuration of the hole and the partial wave exists. The component of along , with the opposite eigenvalue of , has the same modulus: both contain the same radial integral, and the squared reduced matrix elements of are symmetric. These contributions thus cancel in pairs. By (S1) the part orthogonal to the occupied orbitals contributes positively unless it vanishes. If it vanished, and would lie along occupied orbitals. The contributions of to the transition density would cancel in the same pairs, would not act, and the projected equations would give .
On functions whose occupied-orbital components cancel in pairs likewise, only the parts orthogonal to the occupied orbitals, where is positive by (S1), enter Eq. (16). On them Eq. (15) is a Hermitian problem for with the scalar product and a continuum from . Such a function is , as . Hence, as in discrete RPA [1], the lowest eigenvalue obeys
so a negative restoring energy, the numerator, gives a discrete , an imaginary pair [32].
The Woods–Saxon model used here [29] is such a case. Its mean field is
with the upper sign for neutrons and the lower for protons, the neutron and proton numbers and , the mass number , , and the Coulomb potential of a uniformly charged sphere of protons and radius . The depth , spin-orbit strength , radius parameter and diffuseness are those of Ref. [29] for and, for , MeV, MeV fm, fm, fm [33]. The contact strengths , Eqs. (59) and (60) of Ref. [29], are the second derivatives in , at the ground-state densities, of the term and the
term (density exponent 1) of a Skyrme energy density, with MeV fm, MeV fm, and the spin-exchange parameters , ; MeV fm; the radial step is 0.2 fm () and 0.25 fm (). At 20 fm the Jost solutions take the free Hankel values of Eq. (7) in every channel, protons included, and are integrated inward with all potentials, Coulomb included. For both nuclei the model satisfies (S1) and (S2). The pair lies at MeV () [Figure 1(a)] and 8.07i MeV (), with below the bounds and MeV of Eq. (18).

Figure 1. for , , on the first Riemann sheet at MeV, with contours every 0.25; is the Jost matrix of Eq. (8) with . (a) Uncalibrated Woods–Saxon model: the pair of zeros at MeV (circles). (b) Calibrated model ( multiplied by 0.7215): the zero of order two at (circle).
A zero at : In a self-consistent model the rigid shift gives a decaying zero-energy solution, so . Self-consistency means that the mean field and the contact strengths are the first and second derivatives of one energy density of , and vanish beyond the radius. The hole equation gives , which cancels by the transition density of and , so . Beyond the radius, is a multiple of , as belongs to . The Hankel-function recurrences give, for ,
so , regular at , decays. By Eq. (8) this solution is with and
In a model that is not self-consistent, a calibration multiplies the residual interaction by a positive constant chosen so that ; for the Woods–Saxon model it is 0.7215 () and 0.6116 () [Figure 1(b)]. As then doubles from 0.05 and 0.25 MeV, grows by 4.000 () and 3.971 (); with the nonzero term of below, this shows one state. In the self-consistent and the calibrated models the zero is, like every zero off the cuts, a pole of . Let it have one state, with a real solution , as is real. Then no has a zero-energy bound solution. By (S1) such a solution would be an occupied orbital of the partial wave with . With the configuration of and in the pair argument, as . The vector whose only nonzero components are in channel and in channel has zero transition density and, by Eq. (4), solves the equation at . It and its image under are two independent solutions, contrary to one state. Hence the forward and backward parts of , whose difference maps to zero at , are equal: with ; in the self-consistent model.
Exactly second order: With one state, the pole at and the zero of are of order two in both the self-consistent and the calibrated models. Since is not identically zero, the pole of at has a finite order . Then with , since its columns are zero-energy solutions regular at that decay and by Eq. (10) and . As maps to , multiplying Eq. (1) from the left by and by the transpose of that vector, and integrating over , gives
The inertia is positive by the pair argument at , as no has a zero-energy bound solution. Times , the left side tends to , while the right side diverges for and vanishes for . Hence and
In the self-consistent model the hole equation gives the components , so , the c.m. inertia. With one state the null space of is one-dimensional, so some minor of is nonzero. With no common left null vector of , the zero of , an Evans-type determinant [34], has the pole’s order, two, as in Ref. [35].
The state of the double pole: The state is , in the self-consistent model the translation solution, normalized by the inertia. The coefficient matrix of replaces the residue matrix in Eqs. (11), (12). It has rank one by Eqs. (10), (23), and its Autonne–Takagi factorization gives
Physical elements: In the self-consistent model every physical -matrix element, continued to , carries this term. The element between forward channels is at real above both thresholds. Under (S1), (S2) no zero of lies in the quadrant , , so its continuation through that quadrant to near 0 is :
Column of is beyond the radius, so, with , is fixed by the tail of and is nonzero in every forward channel by Eq. (20) and (S2).
Answer to Ginocchio et al.: Reference [20] (p. 72) answered that the matrix has no pole at , because the particle wave functions in the radial integral of its eq:90 have none. They have none, as no has a zero-energy bound solution, but in the self-consistent model that integral, which extends to infinity, diverges as . There it reads , with the scattering solution of in channel . The integral one obtains has as for small . Two integrations by parts, with and from the hole equation, give
Its right side is regular at , as vanishes beyond the radius. At , as , it is a Wronskian of with beyond the radius, nonzero by Eq. (20) and (S2).
Removal of the Mittag-Leffler terms. As is even in , its principal part at the poles of the mode follows from their states. Since and , the matrix solves Eq. (1) at , is regular at , and decays. Such a solution is unique at off the cuts with , so , and makes even. For a pair of simple poles, each with one state, the residue of at is thus . Subtracting the two pole terms, the principal part of at gives
For the double pole at the even has no term, and Eq. (23) gives the principal part, its term:
which is in the self-consistent model.
For a real radial field , the strength at real is , with the smoothing width ; is the same with . As is real, uniqueness also gives off the cuts, so the subtracted terms are real at real for the double pole at and for a pair on the imaginary axis. Hence, as , at every real in both cases.
At the double pole leaves in the tail
which falls off as for . In Figure 2 ( MeV) it is 9% of at 7 MeV for E1 and 7–22% at 6.5–10 MeV for .

Figure 2. (solid) and with Eq. (28) (dashed) at MeV for , , in the calibrated Woods–Saxon model ( multiplied by 0.6116). Left: E1, , . Right: isoscalar compression, . Top: the vertical dotted line marks the lowest threshold, 7.045 MeV. Bottom: their energy-weighted sums ; the dotted lines are their limits, the right sides of Eqs. (30) and (31), 405.6 and 248.5 MeV fm (left), and MeV fm (right). Up to 80 MeV the sums reach 99.6% of them, above 99.9% with the fraction that the terms add beyond.
Under (S2), if has no zero at , obeys for every the energy-weighted sum rule [1], here without self-consistency. With the double pole at , Eqs. (28) and (29) then give that of :
Equation (30) follows from the term at large of . Multiplying Eq. (1) from the left by , integrating over by parts without surface terms, and repeating this for the remainder give no term, since , and the coefficient of . The residual interaction drops out of it, as velocity-independent contact forces do [36]: the forward and backward parts of cancel in the terms of Eq. (2). With the hole equation and (S2), under which the spin-orbit term of Eq. (19) cancels, the coefficient is , the ground-state expectation value of the double commutator of the field with the kinetic energy [36]. The evenness of and make odd in , and, with no zero at , closing the contour in the upper half plane gives Eq. (31). In the calibrated models no zero of other than is found on the imaginary axis up to MeV in steps of MeV () and MeV (). If none lies beyond, none lies at , as the pair argument makes real.
For the field , whose a self-consistent model gives entirely to the double pole, half the coefficient of the subtracted terms at large is almost all of : () and () for the Woods–Saxon pair terms of Eq. (27), and and for the calibrated double pole of Eq. (28). The rest stays in .
Equality with the projection
For in the self-consistent model with one state at , the projection formula, Eq. (15) of Ref. [18], equals the subtraction of Eq. (28). In the radial form of Eq. (5) it is
Multiplying Eq. (22) from the right by , with the components of , gives the radial form of the identity of Ref. [37],
so . In discrete self-consistent RPA this projection leaves a remainder that gives no strength for local fields [38]; here it subtracts exactly the principal part, with continuum boundary conditions.
Conclusion
Within the Jost-RPA the spurious c.m. mode is a zero of the Jost determinant, an -matrix pole. In the uncalibrated Woods–Saxon models it is a pair on the imaginary axis. At , whether by self-consistency or by calibration, it is, under (S1) and (S2) and with one state, a pole of exactly second order, with its state from the Autonne–Takagi factorization. In a self-consistent model of one energy density of every physical -matrix element continued to carries the term. This corrects only the analytic conclusion of Ref. [20]: its physical statement that translation is not an intrinsic excitation stands. Subtracting the principal part of the response function at the poles leaves the strength at real unchanged as ; in such a self-consistent model it equals the projection of Ref. [18]. If has no zero at , the strength obeys, as in discrete RPA, the energy-weighted sum rule without self-consistency (Figure 2).
References
- [1]D. J. Thouless, Nucl. Phys. 22, 78 (1961).
- [2]D. Gambacurta, M. Grasso, and F. Catara, Phys. Rev. C 81, 054312 (2010).
- [3]P. Papakonstantinou, Phys. Rev. C 90, 024305 (2014).
- [4]K. Yoshida and T. Nakatsukasa, Phys. Rev. C 88, 034309 (2013).
- [5]D. Bianco, F. Knapp, N. Lo Iudice, P. Veselý, F. Andreozzi, G. De Gregorio, and A. Porrino, J. Phys. G 41, 025109 (2014).
- [6]F. Knapp, P. Papakonstantinou, P. Veselý, G. De Gregorio, J. Herko, and N. Lo Iudice, Phys. Rev. C 107, 014305 (2023).
- [7]B. K. Agrawal, S. Shlomo, and A. I. Sanzhur, Phys. Rev. C 67, 034314 (2003).
- [8]E. Khan, N. Sandulescu, M. Grasso, and N. Van Giai, Phys. Rev. C 66, 024309 (2002).
- [9]P. Papakonstantinou, T. S. Kosmas, J. Wambach, and A. Faessler, Phys. Rev. C 73, 035502 (2006).
- [10]I. Hamamoto and H. Sagawa, Phys. Rev. C 66, 044315 (2002).
- [11]S. Shlomo and G. F. Bertsch, Nucl. Phys. A 243, 507 (1975).
- [12]M. Matsuo, Nucl. Phys. A 696, 371 (2001).
- [13]K. Hagino and H. Sagawa, Nucl. Phys. A 695, 82 (2001).
- [14]S. Goriely, E. Khan, and M. Samyn, Nucl. Phys. A 739, 331 (2004).
- [15]K. Mizuyama, M. Matsuo, and Y. Serizawa, Phys. Rev. C 79, 024313 (2009).
- [16]N. Van Giai and H. Sagawa, Nucl. Phys. A 371, 1 (1981).
- [17]M. N. Harakeh and A. E. L. Dieperink, Phys. Rev. C 23, 2329 (1981).
- [18]K. Mizuyama and G. Colò, Phys. Rev. C 85, 024307 (2012).
- [19]J. Meyer-ter-Vehn, Z. Phys. A 289, 319 (1979).
- [20]J. N. Ginocchio, T. H. Schucan, and H. A. Weidenmüller, Phys. Rev. C 1, 55 (1970).
- [21]V. A. Khodel and E. E. Saperstein, Phys. Rep. 92, 183 (1982).
- [22]T. Vertse, R. J. Liotta, and E. Maglione, Nucl. Phys. A 584, 13 (1995).
- [23]D. J. Thouless and J. G. Valatin, Nucl. Phys. 31, 211 (1962).
- [24]H. Nakada, Prog. Theor. Exp. Phys. 2016, 063D02 (2016).
- [25]H. Nakada, Prog. Theor. Exp. Phys. 2017, 023D03 (2017).
- [26]Y. R. Shimizu, P. Donati, and R. A. Broglia, Phys. Rev. Lett. 85, 2260 (2000).
- [27]N. Niinohara, Phys. Rev. C 92, 034321 (2015).
- [28]M. L. Gorelik, S. Shlomo, B. A. Tulupov, and M. H. Urin, Phys. Rev. C 103, 034302 (2021).
- [29]K. Mizuyama, N. Nhu Le, T. Dieu Thuy, N. Hoang Tung, D. Quang Tam, and T. V. Nhan Hao, Phys. Rev. C 107, 024303 (2023).
- [30]K. Mizuyama, T. Dieu Thuy, and T. V. Nhan Hao, Phys. Rev. C 109, 054304 (2024).
- [31]K. Mizuyama and T. Dieu Thuy, Phys. Rev. C 111, 054323 (2025).
- [32]D. J. Thouless, Nucl. Phys. 21, 225 (1960).
- [33]A. Bohr and B. R. Mottelson, Nuclear Structure (Benjamin, New York, 1969), Vol. I.
- [34]F. Gesztesy, Y. Latushkin, and K. A. Makarov, Arch. Ration. Mech. Anal. 186, 361 (2007).
- [35]R. L. Pego and H. A. Warchall, J. Nonlinear Sci. 12, 347 (2002).
- [36]E. Lipparini and S. Stringari, Phys. Rep. 175, 103 (1989).
- [37]G. Vignale, Phys. Lett. A 209, 206 (1995).
- [38]V. Tselyaev, Phys. Rev. C 106, 064327 (2022).