Fermionic Molecular Dynamics

• Fermionic Molecular Dynamics (FMD) is a microscopic method that constructs nuclear wave functions from antisymmetrized Gaussian packets, enabling ab initio calculations.
• FMD restores broken symmetries through projection operators and variational optimization, effectively capturing both shell-model and clustered nuclear states.
• FMD models scattering, bound states, and bulk fermion matter using analytical matrix methods and periodic boundary conditions for accurate observables.

Fermionic Molecular Dynamics (FMD) is a fully microscopic many-body method that constructs nuclear and bulk fermion wave functions from antisymmetrized products of localized Gaussian wave packets, incorporating restoration of broken symmetries and variational optimization of single-particle parameters. This framework systematically describes both shell-model–like states and spatially correlated cluster structures, and enables ab initio calculations of nuclear structure, resonances, reaction cross sections, and bulk fermionic matter properties (Neff et al., 2010, Vantournhout et al., 2011, Feldmeier et al., 2016, Neff, 2012, Feldmeier et al., 2013, Vantournhout et al., 2010, Neff et al., 2010, Gulminelli et al., 2011).

1. Foundation: Wave-Packet Ansatz and Antisymmetrization

The essential constituent of FMD is the single-particle wave packet,

$$\varphi_k(\mathbf{r}) = \exp\left\{-\frac{(\mathbf{r}-\mathbf{b}_k)^2}{2a_k}\right\} \otimes \chi_k \otimes \xi_k,$$

where $\mathbf{b}_k\in\mathbb{C}^3$ is the centroid (encoding mean position and momentum), $a_k\in\mathbb{C}$ is the complex width, $\chi_k$ is a two-component spinor, and $\xi_k$ is an isospin label (proton/neutron). The intrinsic many-body state is built as a Slater determinant: $$|Q\rangle = \mathcal{A}\bigl\{\varphi_1\otimes\varphi_2\otimes\cdots\otimes\varphi_A\bigr\},$$ with full antisymmetrization ensuring all fermionic correlations. The parameters $\{\mathbf{b}_k, a_k, \chi_k\}$ serve as variational degrees of freedom (Neff et al., 2010, Feldmeier et al., 2016).

This basis is maximally flexible: it recovers the harmonic oscillator shell model for $b_k=0$ and equal widths, and naturally incorporates Brink-type cluster models as special limits. No explicit cluster degrees of freedom are introduced; spatial correlations and clustering emerge due to energy minimization (Feldmeier et al., 2016, Feldmeier et al., 2013).

2. Symmetry Restoration and Variational Principles

Intrinsic Slater determinants break physical symmetries (parity, angular momentum, translational invariance). FMD restores these symmetries using projection operators:

• Parity: $P^\pi = \frac{1}{2} (1 + \pi \Pi)$,
• Angular momentum: $$P^J_{MK} = \frac{2J+1}{8\pi^2}\int d\Omega\, D^{J*}_{MK}(\Omega) R(\Omega)$$,
• Center of mass: $$P^{\vec{P}=0} = \int \frac{d^3R}{(2\pi)^3} e^{-i\vec{P}\cdot\vec{R}} T(\vec{R})$$.

Symmetry restoration is performed before or during the variation. In the Variation After Projection (VAP) scheme, all single-particle parameters are varied to minimize the projected energy,

$$E[Q] = \frac{\langle Q;J^\pi M K|H|Q;J^\pi M K\rangle}{\langle Q;J^\pi M K|Q;J^\pi M K\rangle},$$

often under additional constraints (notably, on the radius to control cluster separations and generate polarized states) (Neff et al., 2010, Feldmeier et al., 2016). These techniques allow for optimized configuration mixing in a non-orthogonal basis, producing low-lying excitations, cluster structures, and polarized configurations.

3. Hamiltonian, Effective Interactions, and Bulk Formalism

The FMD Hamiltonian takes the form

$$H = \sum_{i<j} \left[ T_{ij} + V^{NN}_{ij} + V^C_{ij} \right] + T_{cm},$$

where $T_{ij}$ is relative kinetic energy, $V^{NN}_{ij}$ is a realistic nucleon-nucleon interaction (commonly the UCOM-transformed Argonne v18), $V^C_{ij}$ is the Coulomb interaction, and $T_{cm}$ subtracts center-of-mass motion (Neff et al., 2010, Neff et al., 2010, Feldmeier et al., 2013).

In bulk (infinite) fermion systems, FMD introduces periodic boundary conditions via a Bravais lattice. The infinite wave function is constructed as

$$|Q_\infty\rangle = \hat{A} \bigotimes_{R\in B} T(R)\{|q_1\rangle\otimes\cdots\otimes|q_A\rangle\},$$

with the block-Toeplitz structure of the overlap and Hamiltonian matrices handled by Brillouin-zone integrals. Observables per unit cell reduce to $A\times A$ matrix inversions and $k$-space integration,

$$\langle B_1\rangle = \frac{1}{V_{BZ}}\int_{BZ} d^3k\, \sum_{p,q} B_{pq}(k) O_{qp}(k),$$

where $B_{pq}(k)$ are Fourier-transformed matrix elements and $O_{qp}(k)$ is the inverse overlap. This structure allows for exact implementation of the long-range antisymmetrization required for bulk systems, as in neutron-star crust modeling (Vantournhout et al., 2011, Vantournhout et al., 2010, Gulminelli et al., 2011).

Boundary conditions may employ either twist-averaged (Bloch) phases or a replica/Wannier construction. The latter is formally exact for variational ansätze and robustly handles inter-cell correlation, representing current best practice for periodic FMD (Gulminelli et al., 2011).

4. Model Spaces: Clustering, Polarization, and Configuration Mixing

FMD model spaces generally consist of two sectors:

• Frozen cluster configurations: Direct-product Slater determinants representing separated clusters (e.g., $^3$He + $^4$He), built at a grid of intercluster separations.
• Polarized configurations: Intrinsic basis states obtained by VAP with constraints (e.g., on the cluster–cluster distance or deformation), resulting in cluster structures with re-adjusted internal wave functions due to nuclear polarization (Neff et al., 2010, Feldmeier et al., 2016, Feldmeier et al., 2013).

The total solution is constructed by diagonalizing the Hamiltonian in the full non-orthogonal space of symmetry-projected energetically and structurally distinct configurations,

$$|\Psi^J_n\rangle = \sum_\alpha c^n_\alpha |Q_\alpha;J\rangle,$$

yielding converged energies, radii, and transition strengths. This approach accommodates both shell-model–like and highly-clustered limits, capturing the dynamical emergence of complex structural features (Feldmeier et al., 2016, Neff, 2012).

5. Treatment of Scattering, Bound States, and Reaction Observables

Scattering and reaction observables are computed via a microscopic R-matrix method:

• The model space is divided at a channel radius $a$; for $r<a$ the full FMD Hamiltonian is used, while for $r>a$ only the Coulomb interaction remains.
• The Bloch operator,

$$L(a) = \frac{\hbar^2}{2\mu} \delta(r-a)\left(\frac{d}{dr} - \frac{B}{a}\right),$$

implements boundary conditions matching internal and external wave functions (Neff et al., 2010).

For bound states and resonances, diagonalization yields wave functions with correct asymptotics. Scattering solutions are matched to known Coulomb or Whittaker functions at the boundary; phase shifts and K-matrix elements are extracted from asymptotics (Neff et al., 2010, Neff et al., 2010, Neff, 2012).

Radiative-capture cross sections, such as for $^3$He$(\alpha,\gamma)^7$Be, are computed by evaluating electromagnetic matrix elements between many-body scattering and bound states: $$\sigma_{E1}(E) = \frac{16\pi}{9}\frac{k_\gamma^3}{k_i} |\langle \Psi_f || \mathcal{O}(E1) || \Psi_i \rangle|^2,$$ with the astrophysical S-factor given by $S(E) = \sigma(E) E e^{2\pi\eta}$ (Neff et al., 2010, Neff et al., 2010). The calculation includes all relevant S- and D-wave contributions; comparison with experiment demonstrates excellent agreement in both absolute normalization and energy dependence.

6. Applications, Model Validation, and Physical Insights

FMD has been applied to a broad spectrum of nuclear and bulk phenomena:

• Nuclear clustering and exotic structures: Charge radii in Neon isotopes, two-proton halo in $^{17}$Ne, and three-alpha clustering in $^{12}$C—where the Hoyle state emerges as a spatially extended dilute triangle—are explained through explicit configuration mixing. No explicit clusters are imposed; collective structures arise through energetically-favored localizations of wave packets (Feldmeier et al., 2016, Feldmeier et al., 2013).
• Ab initio capture reaction calculations: In $^3$He$(\alpha,\gamma)^7$Be and $^3$H$(\alpha,\gamma)^7$Li, FMD with UCOM-transformed AV18 interaction gives $S(0)\approx0.56$ keV b for the former in agreement with experimental S-factor data, and $S(0)\approx0.10$ keV b for the latter, approximately 15% above experimental values. The bound-state energies, phase shifts, and charge radii are reproduced to within a few percent (Neff et al., 2010, Neff et al., 2010, Neff, 2012).
• Bulk fermion matter and neutron-star crusts: FMD with periodic boundary conditions rigorously implements long-range Pauli correlations, reproduces Fermi-gas limits, and enables study of inhomogeneous matter ("nuclear pasta") with spatial localization (Vantournhout et al., 2011, Vantournhout et al., 2010, Gulminelli et al., 2011).

Benchmark calculations using both constrained and variationally optimized bases demonstrate that physically-relevant observables (energies, radii, phase shifts, S-factors) are robustly predicted within the FMD framework for light and medium-mass nuclei, and for periodic bulk systems, with accuracy competitive with the most advanced ab initio methods.

7. Computational Strategies and Best Practice

• All Hamiltonian and overlap kernels between Slater determinants of Gaussians are evaluated analytically or by straightforward matrix algebra. For periodic systems, Brillouin-zone integrals are performed using Monkhorst–Pack $k$-meshes; matrices requiring inversion are of size $A\times A$ per $k$-point.
• Variational optimization (VAP) is required for physically accurate cluster and deformed states; additional constraints (e.g., radius or quadrupole moment) are employed to span reaction or collective coordinate spaces.
• Boundary conditions for bulk systems should use the replica/Wannier method rather than pure TABC/Bloch phases if the variational ansatz may not capture true plane-wave eigenstates, to avoid artefacts and guarantee thermodynamic-limit observables (Gulminelli et al., 2011).
• Convergence in nuclear observables is obtained with moderate basis sizes (on the order of several tens of Slater determinants per $J^\pi$ channel), and FMD computations—though numerically intensive—are feasible on modern clusters for light and medium-mass systems.

Table: Key Features of FMD

Feature	Implementation	Reference
Single-particle basis	Gaussian wave packets (complex widths, centroids, spin/isospin)	(Neff et al., 2010, Feldmeier et al., 2016)
Many-body state	Antisymmetric Slater determinant	(Neff et al., 2010, Vantournhout et al., 2011)
Symmetry restoration	Angular momentum, parity, c.m. projection	(Neff et al., 2010, Feldmeier et al., 2016)
Hamiltonian	UCOM-transformed $V_{NN}$ + Coulomb	(Neff et al., 2010, Neff et al., 2010)
Bulk extensions	Periodic boundary (Bravais lattice), replica method	(Vantournhout et al., 2011, Gulminelli et al., 2011)
Reaction observables	Microscopic R-matrix, Bloch operator	(Neff et al., 2010, Neff et al., 2010)

Fermionic Molecular Dynamics provides a unified, fully microscopic approach to nuclear structure, clustering, resonances, and ab initio reaction cross sections, and extends rigorously to quantum-consistent descriptions of inhomogeneous bulk fermion systems. Its predictive power derives from the unique combination of a physically complete Gaussian wave-packet basis, exact symmetry restoration, configuration mixing, and the ability to implement both finite and infinite boundary conditions (Neff et al., 2010, Vantournhout et al., 2010, Feldmeier et al., 2016, Vantournhout et al., 2011, Gulminelli et al., 2011, Neff et al., 2010, Feldmeier et al., 2013, Neff, 2012).