Angular Momentum Projection in Nuclear Theory

• Angular Momentum Projection is a technique that restores rotational symmetry by extracting states with definite total angular momentum from deformed many-body wave functions.
• It utilizes numerical integration over Euler angles as well as algebraic and linear-system methods to compute projected norm and Hamiltonian kernels for accurate state reconstruction.
• The method underpins diverse nuclear structure applications, from generator-coordinate mixing to fission and rotational band analyses, highlighting both computational delicacies and physical insights.

Angular momentum projection is the symmetry-restoration procedure that extracts states with definite total angular momentum from intrinsically deformed or otherwise symmetry-breaking many-body wave functions. In nuclear structure theory, intrinsic Slater determinants, Hartree–Fock or Hartree–Fock–Bogoliubov vacua, antisymmetrized molecular dynamics states, and cluster-model states are flexible descriptions of deformation, pairing, and clustering, but they are generally not eigenstates of $\hat J^2$. Projection restores the laboratory-frame quantum numbers required by rotational invariance and is therefore a basic ingredient of projected mean-field theory, projected generator-coordinate methods, multiconfiguration mixing, and several algebraic alternatives to integral projection (Taniguchi, 2016, Pan et al., 2013, Johnson et al., 2017).

1. Standard operator framework

The standard projector onto total angular momentum $J$, laboratory projection $M$, and intrinsic projection $K$ is

$$\hat{P}_{MK}^J = \frac{2J + 1}{8\pi^2}\int D_{MK}^{J\ast} (\alpha, \beta, \gamma)\, \hat{R}(\alpha, \beta, \gamma)\, d\alpha\, d\beta\, d\gamma ,$$

with

$$\hat{R}(\alpha, \beta, \gamma) = e^{-i\alpha \hat{J}_z} e^{-i\beta \hat{J}_y} e^{-i\gamma \hat{J}_z}, \qquad D^J_{MK}(\alpha,\beta,\gamma)=e^{-iM\alpha}d^J_{MK}(\beta)e^{-iK\gamma}.$$

Acting on an intrinsic state $|\Phi\rangle$, $\hat P^J_{MK}|\Phi\rangle$ isolates the component carrying the specified $J,M,K$ quantum numbers. A useful factorization is

$$\hat{P}_{MK}^J = (2J + 1)\, \hat{P}_M^{(z)}\, \hat{P}_{MK}^{(y)J}\, \hat{P}_K^{(z)}, \qquad \hat{P}_K^{(z)} = \frac{1}{2\pi} \int_0^{2\pi} d\phi\, e^{i(K - \hat{J}_z) \phi},$$

which makes the $J$0-projection content explicit (Taniguchi, 2016).

For a single Hartree–Fock Slater determinant $J$1, one builds projected norm and Hamiltonian kernels,

$J$2

and then solves the generalized eigenvalue problem

$J$3

When the intrinsic state is represented by an occupied-orbital matrix $J$4, the rotated determinant is $J$5, the overlap is

$J$6

and the transition density enters the Hamiltonian kernel through the usual one-body and two-body contractions. This kernel-based formulation is the basic computational pattern for projected Hartree–Fock and related symmetry-restored methods (Staker et al., 2013).

2. Intrinsic $J$7, $J$8-distribution, and alignment

In an intrinsic frame, $J$9 denotes the projection of total angular momentum on the body-fixed $M$0-axis. For a general intrinsic state $M$1, the weight of a given $M$2 is

$M$3

The set $M$4 is the $M$5-distribution. A narrow $M$6-distribution is desirable in projected multiconfiguration calculations because each non-negligible $M$7 generates an additional basis state $M$8, while different $M$9-components of the same intrinsic state are generally non-orthogonal after projection. Broad $K$0-distributions therefore enlarge the basis, worsen conditioning of the norm matrix, and can produce severe near-linear dependence (Taniguchi, 2016).

The orientation of the intrinsic state is decisive. The same physical intrinsic wave function may have a very narrow $K$1-distribution in one orientation and a very broad one in another. To address this alignment problem, a general prescription is to minimize the variance of $K$2,

$K$3

or, more generally, to diagonalize the angular-momentum covariance matrix

$K$4

Because $K$5 transforms as $K$6, choosing the body-fixed $K$7-axis along the eigenvector with the smallest eigenvalue minimizes the $K$8 variance and concentrates the $K$9-distribution. This procedure was benchmarked for an $$\hat{P}_{MK}^J = \frac{2J + 1}{8\pi^2}\int D_{MK}^{J\ast} (\alpha, \beta, \gamma)\, \hat{R}(\alpha, \beta, \gamma)\, d\alpha\, d\beta\, d\gamma ,$$0–$$\hat{P}_{MK}^J = \frac{2J + 1}{8\pi^2}\int D_{MK}^{J\ast} (\alpha, \beta, \gamma)\, \hat{R}(\alpha, \beta, \gamma)\, d\alpha\, d\beta\, d\gamma ,$$1Mg cluster structure in $$\hat{P}_{MK}^J = \frac{2J + 1}{8\pi^2}\int D_{MK}^{J\ast} (\alpha, \beta, \gamma)\, \hat{R}(\alpha, \beta, \gamma)\, d\alpha\, d\beta\, d\gamma ,$$2Si, a triaxially superdeformed state in $$\hat{P}_{MK}^J = \frac{2J + 1}{8\pi^2}\int D_{MK}^{J\ast} (\alpha, \beta, \gamma)\, \hat{R}(\alpha, \beta, \gamma)\, d\alpha\, d\beta\, d\gamma ,$$3Ar, and Hartree–Fock states in light nuclei. It performs similarly to moment-of-inertia alignment in strongly quadrupole-deformed even-even nuclei, but it is markedly superior for cluster structures, nearly spherical odd and odd–odd nuclei, and cases where angular momentum is carried predominantly by valence nucleons rather than by collective quadrupole deformation. In $$\hat{P}_{MK}^J = \frac{2J + 1}{8\pi^2}\int D_{MK}^{J\ast} (\alpha, \beta, \gamma)\, \hat{R}(\alpha, \beta, \gamma)\, d\alpha\, d\beta\, d\gamma ,$$4O, moment-of-inertia alignment generated a very broad $$\hat{P}_{MK}^J = \frac{2J + 1}{8\pi^2}\int D_{MK}^{J\ast} (\alpha, \beta, \gamma)\, \hat{R}(\alpha, \beta, \gamma)\, d\alpha\, d\beta\, d\gamma ,$$5-distribution whose $$\hat{P}_{MK}^J = \frac{2J + 1}{8\pi^2}\int D_{MK}^{J\ast} (\alpha, \beta, \gamma)\, \hat{R}(\alpha, \beta, \gamma)\, d\alpha\, d\beta\, d\gamma ,$$6 and $$\hat{P}_{MK}^J = \frac{2J + 1}{8\pi^2}\int D_{MK}^{J\ast} (\alpha, \beta, \gamma)\, \hat{R}(\alpha, \beta, \gamma)\, d\alpha\, d\beta\, d\gamma ,$$7 components after $$\hat{P}_{MK}^J = \frac{2J + 1}{8\pi^2}\int D_{MK}^{J\ast} (\alpha, \beta, \gamma)\, \hat{R}(\alpha, \beta, \gamma)\, d\alpha\, d\beta\, d\gamma ,$$8 projection had overlap $$\hat{P}_{MK}^J = \frac{2J + 1}{8\pi^2}\int D_{MK}^{J\ast} (\alpha, \beta, \gamma)\, \hat{R}(\alpha, \beta, \gamma)\, d\alpha\, d\beta\, d\gamma ,$$9, causing numerical pathologies, whereas variance minimization yielded a pure $$\hat{R}(\alpha, \beta, \gamma) = e^{-i\alpha \hat{J}_z} e^{-i\beta \hat{J}_y} e^{-i\gamma \hat{J}_z}, \qquad D^J_{MK}(\alpha,\beta,\gamma)=e^{-iM\alpha}d^J_{MK}(\beta)e^{-iK\gamma}.$$0 state (Taniguchi, 2016).

3. Configuration mixing and symmetry-restored many-body methods

Angular momentum projection is most powerful when combined with configuration mixing. A general projected multiconfiguration state can be written as

$$\hat{R}(\alpha, \beta, \gamma) = e^{-i\alpha \hat{J}_z} e^{-i\beta \hat{J}_y} e^{-i\gamma \hat{J}_z}, \qquad D^J_{MK}(\alpha,\beta,\gamma)=e^{-iM\alpha}d^J_{MK}(\beta)e^{-iK\gamma}.$$1

so that each intrinsic configuration $$\hat{R}(\alpha, \beta, \gamma) = e^{-i\alpha \hat{J}_z} e^{-i\beta \hat{J}_y} e^{-i\gamma \hat{J}_z}, \qquad D^J_{MK}(\alpha,\beta,\gamma)=e^{-iM\alpha}d^J_{MK}(\beta)e^{-iK\gamma}.$$2 contributes through its projected $$\hat{R}(\alpha, \beta, \gamma) = e^{-i\alpha \hat{J}_z} e^{-i\beta \hat{J}_y} e^{-i\gamma \hat{J}_z}, \qquad D^J_{MK}(\alpha,\beta,\gamma)=e^{-iM\alpha}d^J_{MK}(\beta)e^{-iK\gamma}.$$3-components. This structure underlies HF–GCM, AMD, FMD, and Monte Carlo shell-model–type calculations (Taniguchi, 2016).

In triaxial projected GCM with the Gogny force, one combines particle-number and angular-momentum projection with shape mixing in the $$\hat{R}(\alpha, \beta, \gamma) = e^{-i\alpha \hat{J}_z} e^{-i\beta \hat{J}_y} e^{-i\gamma \hat{J}_z}, \qquad D^J_{MK}(\alpha,\beta,\gamma)=e^{-iM\alpha}d^J_{MK}(\beta)e^{-iK\gamma}.$$4 plane,

$$\hat{R}(\alpha, \beta, \gamma) = e^{-i\alpha \hat{J}_z} e^{-i\beta \hat{J}_y} e^{-i\gamma \hat{J}_z}, \qquad D^J_{MK}(\alpha,\beta,\gamma)=e^{-iM\alpha}d^J_{MK}(\beta)e^{-iK\gamma}.$$5

and solves the Hill–Wheeler–Griffin equation built from projected norm and Hamiltonian kernels. The first full triaxial implementation in the $$\hat{R}(\alpha, \beta, \gamma) = e^{-i\alpha \hat{J}_z} e^{-i\beta \hat{J}_y} e^{-i\gamma \hat{J}_z}, \qquad D^J_{MK}(\alpha,\beta,\gamma)=e^{-iM\alpha}d^J_{MK}(\beta)e^{-iK\gamma}.$$6 plane with the Gogny force reproduced the low-lying spectrum and transition probabilities of $$\hat{R}(\alpha, \beta, \gamma) = e^{-i\alpha \hat{J}_z} e^{-i\beta \hat{J}_y} e^{-i\gamma \hat{J}_z}, \qquad D^J_{MK}(\alpha,\beta,\gamma)=e^{-iM\alpha}d^J_{MK}(\beta)e^{-iK\gamma}.$$7Mg reasonably well and identified a ground band and a $$\hat{R}(\alpha, \beta, \gamma) = e^{-i\alpha \hat{J}_z} e^{-i\beta \hat{J}_y} e^{-i\gamma \hat{J}_z}, \qquad D^J_{MK}(\alpha,\beta,\gamma)=e^{-iM\alpha}d^J_{MK}(\beta)e^{-iK\gamma}.$$8 $$\hat{R}(\alpha, \beta, \gamma) = e^{-i\alpha \hat{J}_z} e^{-i\beta \hat{J}_y} e^{-i\gamma \hat{J}_z}, \qquad D^J_{MK}(\alpha,\beta,\gamma)=e^{-iM\alpha}d^J_{MK}(\beta)e^{-iK\gamma}.$$9 band. The same work introduced the RVAMPIR approximation, which gives a good description of the ground and gamma bands in the absence of strong mixing (Rodríguez et al., 2010).

Projected generator-coordinate methods also expose a methodological distinction between restoration before and after solving the collective secular equation. In ab initio PGCM studies of monopole resonances, performing angular-momentum restoration only after solving the secular equation contaminates the monopole response with an unphysical coupling to the rotational motion, whereas an a priori restoration is necessary to handle rotations and vibrations consistently in the same framework. This result aligns projected GCM with the broader view that symmetry restoration must enter the variational problem itself when collective motions are strongly coupled (Porro et al., 2024).

For rotational bands at high spin, the same logic leads to multicranked configuration mixing. Instead of projecting a single intrinsic state, one superposes several cranked mean-field states $|\Phi\rangle$0 after projection, implementing the Peierls–Thouless idea that the conjugate collective momentum must enter the generator set. In superdeformed yrast bands of $|\Phi\rangle$1Dy and $|\Phi\rangle$2Hg, this multicranked angular-momentum-projection scheme gives a good description of the rotational spectrum as well as $|\Phi\rangle$3 and $|\Phi\rangle$4, and particle-number projection on top of angular-momentum projection was performed for the first time with the multicranked configuration-mixing (Ushitani et al., 2019). In weakly bound systems, DRHBc+AMP extends projection to deformed relativistic Hartree–Bogoliubov theory in the continuum; with a Dirac Woods–Saxon basis and Bogoliubov pairing, it reproduces the ground-state rotational bands of $|\Phi\rangle$5Mg reasonably well (Sun et al., 2021).

4. Alternative formulations beyond Euler-angle quadrature

Although the integral projector is standard, angular momentum projection is not tied to numerical integration over Euler angles. One alternative is algebraic. For $|\Phi\rangle$6 particles with spins $|\Phi\rangle$7, one introduces collective lowering operators

$|\Phi\rangle$8

and highest-weight states

$|\Phi\rangle$9

Requiring $\hat P^J_{MK}|\Phi\rangle$0 leads to Bethe ansatz equations

$\hat P^J_{MK}|\Phi\rangle$1

These equations are equivalent to the condition that the $\hat P^J_{MK}|\Phi\rangle$2 are zeros of degree-$\hat P^J_{MK}|\Phi\rangle$3 Heine–Stieltjes polynomials satisfying a second-order Fuchsian equation. The formalism yields a constructive projection method for systems of arbitrary spins, connects angular momentum projection to the strong-pairing limit of Richardson–Gaudin models, and simplifies substantially for identical bosons, where the relevant zeros can be expressed in terms of Jacobi polynomials. For identical bosons or fermions the projected states are generally non-orthogonal in the multiplicity label and require orthonormalization (Pan et al., 2013).

A second alternative replaces the three-dimensional Euler-angle integral by the solution of a linear system. The rotated norm and Hamiltonian kernels can be written as finite sums,

$\hat P^J_{MK}|\Phi\rangle$4

$\hat P^J_{MK}|\Phi\rangle$5

so projection becomes an inversion problem for the unknown $\hat P^J_{MK}|\Phi\rangle$6 and $\hat P^J_{MK}|\Phi\rangle$7. In the implementation tested on $\hat P^J_{MK}|\Phi\rangle$8Cr and $\hat P^J_{MK}|\Phi\rangle$9Fe in the $J,M,K$0 shell, a quadrature with 40 points per Euler angle, $J,M,K$1 points, agreed with the linear-algebra projection to within about $J,M,K$2 keV for all $J,M,K$3 except $J,M,K$4, where the agreement was within about $J,M,K$5 keV. A “need-to-know” strategy reduced the number of Euler-angle evaluations for $J,M,K$6Cr from about $J,M,K$7 to about $J,M,K$8, while preserving keV-level agreement. A later convergence study showed that inversion methods, including Fomenko projection and trapezoidal quadrature in the $J,M,K$9-rotation angles, dramatically improve efficiency in many cases, with the optimal choice depending on how many $$\hat{P}_{MK}^J = (2J + 1)\, \hat{P}_M^{(z)}\, \hat{P}_{MK}^{(y)J}\, \hat{P}_K^{(z)}, \qquad \hat{P}_K^{(z)} = \frac{1}{2\pi} \int_0^{2\pi} d\phi\, e^{i(K - \hat{J}_z) \phi},$$0 values must be projected (Johnson et al., 2017, Johnson et al., 2018).

5. Spectroscopy, performance, and known failure modes

Projected Hartree–Fock based on a single Slater determinant is a controlled but limited approximation. A direct benchmark against configuration-interaction shell-model calculations in identical model spaces showed that low-lying excitation spectra for rotational nuclides are well reproduced, whereas spectra for vibrational nuclides and, more generally, the complex spectra for odd-$$\hat{P}_{MK}^J = (2J + 1)\, \hat{P}_M^{(z)}\, \hat{P}_{MK}^{(y)J}\, \hat{P}_K^{(z)}, \qquad \hat{P}_K^{(z)} = \frac{1}{2\pi} \int_0^{2\pi} d\phi\, e^{i(K - \hat{J}_z) \phi},$$1 and odd-odd nuclides are less well reproduced in detail. In the $$\hat{P}_{MK}^J = (2J + 1)\, \hat{P}_M^{(z)}\, \hat{P}_{MK}^{(y)J}\, \hat{P}_K^{(z)}, \qquad \hat{P}_K^{(z)} = \frac{1}{2\pi} \int_0^{2\pi} d\phi\, e^{i(K - \hat{J}_z) \phi},$$2-shell nucleus $$\hat{P}_{MK}^J = (2J + 1)\, \hat{P}_M^{(z)}\, \hat{P}_{MK}^{(y)J}\, \hat{P}_K^{(z)}, \qquad \hat{P}_K^{(z)} = \frac{1}{2\pi} \int_0^{2\pi} d\phi\, e^{i(K - \hat{J}_z) \phi},$$3Mg, the projected-HF ground state lies $$\hat{P}_{MK}^J = (2J + 1)\, \hat{P}_M^{(z)}\, \hat{P}_{MK}^{(y)J}\, \hat{P}_K^{(z)}, \qquad \hat{P}_K^{(z)} = \frac{1}{2\pi} \int_0^{2\pi} d\phi\, e^{i(K - \hat{J}_z) \phi},$$4 MeV above CI, projection recovers $$\hat{P}_{MK}^J = (2J + 1)\, \hat{P}_M^{(z)}\, \hat{P}_{MK}^{(y)J}\, \hat{P}_K^{(z)}, \qquad \hat{P}_K^{(z)} = \frac{1}{2\pi} \int_0^{2\pi} d\phi\, e^{i(K - \hat{J}_z) \phi},$$5 of the correlation energy, and the shifted excitation spectrum has an rms deviation of $$\hat{P}_{MK}^J = (2J + 1)\, \hat{P}_M^{(z)}\, \hat{P}_{MK}^{(y)J}\, \hat{P}_K^{(z)}, \qquad \hat{P}_K^{(z)} = \frac{1}{2\pi} \int_0^{2\pi} d\phi\, e^{i(K - \hat{J}_z) \phi},$$6 MeV. In $$\hat{P}_{MK}^J = (2J + 1)\, \hat{P}_M^{(z)}\, \hat{P}_{MK}^{(y)J}\, \hat{P}_K^{(z)}, \qquad \hat{P}_K^{(z)} = \frac{1}{2\pi} \int_0^{2\pi} d\phi\, e^{i(K - \hat{J}_z) \phi},$$7Al, by contrast, the projected-HF ground state is $$\hat{P}_{MK}^J = (2J + 1)\, \hat{P}_M^{(z)}\, \hat{P}_{MK}^{(y)J}\, \hat{P}_K^{(z)}, \qquad \hat{P}_K^{(z)} = \frac{1}{2\pi} \int_0^{2\pi} d\phi\, e^{i(K - \hat{J}_z) \phi},$$8 MeV above CI, projection recovers only about $$\hat{P}_{MK}^J = (2J + 1)\, \hat{P}_M^{(z)}\, \hat{P}_{MK}^{(y)J}\, \hat{P}_K^{(z)}, \qquad \hat{P}_K^{(z)} = \frac{1}{2\pi} \int_0^{2\pi} d\phi\, e^{i(K - \hat{J}_z) \phi},$$9 of the correlation energy, and the shifted excitation spectrum has an rms deviation of $J$00 MeV. The same survey found ground-state $J$01 correctness frequencies of $J$02 for even-even $J$03-shell nuclei, $J$04 for even-even $J$05-shell nuclei, but much lower success rates for odd-$J$06 and odd-odd systems (Staker et al., 2013).

A common simplifying approximation in deformed QRPA calculations is the needle approximation, which treats different rotations of a strongly deformed intrinsic state as effectively orthogonal. Exact angular-momentum projection in axially deformed proton–neutron QRPA modifies both ground-state energies and transition strengths: it reduces calculated beta-decay half-lives from those that use the needle approximation by up to $J$07, and even more when the effects of projection on the ground-state energy are included (Chen et al., 18 Oct 2025). This does not imply that exact projection automatically improves agreement with experiment in an unrecalibrated EDF; rather, it shows that symmetry restoration is a quantitatively large effect that must be incorporated consistently.

The same caution applies to post hoc symmetry restoration in collective-response calculations. In deformed PGCM calculations of monopole resonances, a posteriori angular-momentum restoration can generate large artificial peaks because the unprojected excited states retain overlap with the projected ground state. This suggests that the mathematical order of projection and configuration mixing is not a technicality: when rotations couple strongly to the mode of interest, restoration after the fact can misidentify rotational admixtures as physical response (Porro et al., 2024).

6. Collective modes, discrete symmetries, and broader scope

Angular momentum projection is also the natural laboratory for rotational modes that depend on triaxiality or discrete point-group symmetry. In microscopic projection from cranked triaxially deformed mean-field states, wobbling bands appear as multiple projected rotational sequences. The resulting spectra reflect the dynamics of the angular-momentum vector in the intrinsic frame and distinguish transverse from longitudinal wobbling, paralleling the phenomenological triaxial-rotor and particle–rotor pictures (Shimada et al., 2018). In the same framework, chiral doublet bands arise naturally in $J$08Cs and $J$09Rh without any explicit rotor core; the projected bands display the characteristic $J$10 and $J$11 properties expected from the triaxial particle–rotor model (Shimada et al., 2018).

When parity is broken together with rotational symmetry, angular momentum projection can be combined with parity projection. A microscopic study of tetrahedrally symmetric nuclei showed that for pure tetrahedral deformation the projected excitation patterns satisfy the characteristic features predicted by group-representation theory for the tetrahedral symmetry group, including the sequence $J$12, and that the energy-versus-spin behavior evolves from approximately linear to the rigid-rotor $J$13 pattern as the tetrahedral deformation increases (Tagami et al., 2013).

In microscopic fission theory, projection is used differently but no less fundamentally. TDDFT fragment states near scission are deformed and oriented, hence they are not eigenstates of $J$14. Applying the standard projector

$J$15

yields fragment spin distributions

$J$16

A comparison between projected fragment-spin distributions and a spin cut-off formula derived from the uncertainty relation between orientation angle and angular momentum found that a large portion of the spin distribution obtained from projection methods can be explained by quantum uncertainty associated with fragment orientation, mainly due to quadrupole deformation and, to a lesser extent, octupole deformation (Scamps et al., 1 Dec 2025).

Taken together, these developments define angular momentum projection as more than a post-processing step. It is the operator-theoretic mechanism by which intrinsic, symmetry-breaking descriptions are converted into laboratory-frame spectroscopy, and its practical success depends on how projection is interleaved with alignment, pairing, configuration mixing, and the numerical representation of the many-body state. The recurring lesson across projected HF, Gogny and relativistic EDF frameworks, PGCM, algebraic projection, and fission applications is that restoring rotational symmetry is both conceptually essential and computationally delicate.