In-Medium Similarity Renormalization Group

• IMSRG is an ab initio method that uses continuous unitary transformations to decouple a reference state and yield effective Hamiltonians for many-body systems.
• The Gamow IMSRG extends this approach by incorporating a complex-energy Berggren basis to consistently treat weakly bound, resonant, and continuum states in exotic nuclei.
• Robust generators (Wegner and White), normal-ordering approximations, and the Magnus expansion enable accurate decoupling and computational efficiency in modeling nuclear structures.

The In-Medium Similarity Renormalization Group (IMSRG) method is an ab initio, nonperturbative approach for solving quantum many-body problems through continuous unitary transformations of the Hamiltonian. The IMSRG generates a flow that systematically decouples a chosen reference state from the remainder of the Hilbert space, yielding tractable effective Hamiltonians in finite or infinite systems. The Gamow in-medium similarity renormalization group (Gamow IMSRG) extends this framework by employing the complex-energy Berggren basis, enabling unified and consistent treatment of weakly bound, resonant, and continuum states crucial for describing exotic nuclei near the drip lines. This article systematically reviews the theoretical formulation, algorithmic strategies, and selected applications of the (Gamow) IMSRG method, incorporating results and methods from (Hu et al., 2019, Zhen et al., 5 Jan 2025, Morris et al., 2015), and other foundational studies.

1. Theoretical Foundation of the (Gamow) IMSRG

The IMSRG is based on the similarity renormalization group concept, employing a continuous sequence of unitary transformations on a general many-body Hamiltonian,

$H(0) = T_{\rm int} + V_{NN} + V_{3N},$

where $T_{\rm int}$ is the intrinsic kinetic energy, $V_{NN}$ a two-nucleon interaction, and $V_{3N}$ a three-nucleon interaction. Modern implementations utilize chiral effective field theory (EFT) forces at various orders.

The central idea is to generate a unitarily-equivalent, decoupled Hamiltonian $H(s)$ via

$H(s) = U(s) H(0) U^{-1}(s),$

where $s$ is a continuous flow parameter. Differentiation yields the flow equation

$\frac{dH(s)}{ds} = [\eta(s), H(s)],$

with $\eta(s) = \frac{dU}{ds} U^{-1}(s)$ an anti-Hermitian generator. In the Berggren basis adopted by the Gamow IMSRG, the conjugation property becomes $U^{-1}(s) = U^T(s)$, and anti-Hermiticity is replaced by complex-symmetry: $\eta(s) = -\eta^T(s)$, $H(s) = H^T(s)$.

For weakly bound and unbound nuclear systems, the complex-energy Berggren basis $\{\lvert u_n \rangle\}$ is employed, comprising bound, resonant (Gamow), and non-resonant continuum states. This basis makes use of the completeness relation: $$\sum_n |u_n\rangle\langle u_n| + \int_{L^+}\!\!dk\,|u(k)\rangle\langle u(k)| = 1,$$ with summation over bound and resonance states, and the integral over the complex contour $L^+$ for the continuum (Hu et al., 2019).

2. Choice of IMSRG Generator and Flow Equation Structure

The transformation generator $\eta(s)$ controls which off-diagonal elements are suppressed. Two standard choices are:

• Wegner generator:

$\eta(s) = [H_d(s),\, H_{od}(s)],$

which systematically drives the off-diagonal block $H_{od}(s)$ to zero; here $H_d$ is the "diagonal" part (defined by a partition in the reference basis), and $H_{od}$ the remaining "off-diagonal" part.

• White generator (energy-denominator form):

$$\eta_{ij}(s) = \frac{H^{od}_{ij}(s)\left[H_{jj}(s)-H_{ii}(s)\right]} {\varepsilon_i - \varepsilon_j},$$

where $\varepsilon_i$ are approximate single-particle energies. The White generator provides numerically robust and uniform exponential suppression of off-diagonal elements and is often less stiff for numerical integration than the Wegner form (Hu et al., 2019, Zhen et al., 5 Jan 2025).

The generator and flow structure are preserved in the complex-symmetric (Berggren) basis of the Gamow IMSRG, though transposition replaces Hermitian conjugation in all commutator relations.

3. Normal-Ordering and Truncation

All operators are normal-ordered relative to a chosen reference state, typically a Hartree–Fock or Gamow–Hartree–Fock Slater determinant $|\Phi\rangle$. The normal-ordered expansion: $$H(s) = E(s) + \sum_{pq} f_{pq}(s) \{ a_p^\dagger a_q \} + \frac{1}{4} \sum_{pqrs} \Gamma_{pqrs}(s) \{ a_p^\dagger a_q^\dagger a_s a_r \} + \cdots,$$ defines the zero-body ($E$), one-body ($f$), and two-body ($\Gamma$) terms.

The IMSRG(2) truncation retains only terms up to the two-body sector, discarding residual normal-ordered three-body and higher operators. This truncation has been shown to capture the dominant many-body correlations, particularly for light and medium-mass nuclei and in nuclear matter at moderate densities (Morris et al., 2015, Zhen et al., 5 Jan 2025).

In the Gamow IMSRG, normal-ordering and Wick’s theorem are implemented for complex operators in the Berggren basis. The truncation is similarly enforced at the two-body level, with experience indicating that this approximation remains accurate for a wide range of systems (Hu et al., 2019).

4. Algorithms and Computational Aspects

The practical implementation of IMSRG methods proceeds via:

• Model-space selection: For open quantum systems, a limited set of key partial-wave channels (e.g., $d_{3/2}$, $s_{1/2}$) are expanded in the full complex-energy Berggren basis, while remaining channels employ a standard discrete (real-energy) expansion.
• Continuum discretization: The non-resonant sector of each partial wave is discretized by $N$ Gauss–Legendre points ($N \sim 30{-}40$ per channel) along the contour $L^+$ (Hu et al., 2019).
• Magnus expansion: Instead of directly integrating the flow equation for $H(s)$, one usually solves for an anti-Hermitian Magnus operator $\Omega(s)$:

$$U(s) = e^{\Omega(s)}, \quad \Omega(s) = -\Omega^T(s),$$

and constructs the evolved Hamiltonian as

$H(s) = e^{\Omega(s)} H(0) e^{-\Omega(s)}.$

The Magnus approach enables the use of simple first-order integrators without loss of unitarity or fidelity, reduces memory usage, and permits the straightforward evolution of arbitrary operators via the truncated Baker–Campbell–Hausdorff series (Morris et al., 2015).

• Block-diagonalization and decoupling: The flow proceeds such that, in the limit $s \rightarrow \infty$, the off-diagonal block $H_{od}(s)$ (coupling the reference to its excitations, including into the continuum) is suppressed, and $H(\infty)$ is block-diagonal in the Fock-space partition. This enables direct extraction of the ground-state energy or effective Hamiltonians for further configuration interaction/equation-of-motion treatments.
• Complex arithmetic: All operator algebra, including commutators, exponentiation, and matrix operations, is performed using complex-symmetric linear algebra, as dictated by the Berggren basis.

5. Applications in Weakly Bound and Unbound Nuclei

The Gamow IMSRG framework has been employed in the structural calculation of neutron-rich carbon and oxygen isotopes. Notable results include:

• $^{24}$O (neutron dripline nucleus): Calculations using chiral NNLO$_{\rm opt}$ (two-nucleon) and NNLO$_{\rm sat}$ (three-nucleon included) interactions reproduce the experimentally observed ground-state energy and first $2^+$ excitation ($\sim4.8$ MeV), and accurately predict the position and width of the $d_{3/2}$ resonance ($\sim7.4$ MeV, $\Gamma \sim 0.1$ MeV). Higher-lying $2^+, 3^+, 4^+$ resonances at $8$–$9$ MeV with widths $\sim0.2$ MeV are also found, in agreement with observed continuum peaks (Hu et al., 2019).
• $^{22}$C (Borromean two-neutron halo): Inclusion of the $s$-wave continuum in the Gamow HF basis yields a matter radius $r_{\rm m} \approx 3.1$–$3.2$ fm, in agreement with the measured $3.4\pm0.1$ fm. Real-energy truncations underestimate $r_{\rm m}$ by $\sim10\%$. The ground-state density displays a pronounced exponential tail characteristic of the halo structure. Low-lying $2^+, 1^+, 3^+, 4^+$ resonances at $E_x \approx 3.5{-}4.0$ MeV with widths $\Gamma \approx 0.15{-}0.25$ MeV are predicted.

These results demonstrate that the Gamow IMSRG can treat bound, resonant, and non-resonant continuum states on an equal footing, yielding quantitatively accurate spectra, resonance widths, and spatial distributions for nuclei at the limits of stability (Hu et al., 2019).

6. Broader Significance and Current Extensions

The IMSRG and its Gamow extension represent a unifying framework for ab initio-many-body theory, applicable to both finite nuclei and infinite matter (Zhen et al., 5 Jan 2025). The key advantages and limitations are:

• Advantages:
  • Unitary transformation ensures preservation of eigenvalues and consistent operator evolution.
  • Robust Magnus-based integration reduces computational overhead.
  • The ab initio treatment of continuum and resonance structure in open quantum systems is made possible through the Berggren basis.
  • Comparable computational scaling to CCSD, with the Hamiltonian brought to a block-diagonal Hermitian (or complex-symmetric) form.
• Limitations:
  • IMSRG(2) neglects induced three-body (and higher) operators beyond the normal-ordered two-body approximation; the impact increases with nuclear density and hardness of the interaction.
  • Full inclusion of three-body evolution (IMSRG(3)) is computationally demanding, especially in open or continuum-coupled systems.
• Ongoing and future directions:
  • Systematic deployment of IMSRG(3) (or approximate truncations) for improved accuracy, especially in high-density nuclear matter or for harder interactions.
  • Extension to finite-temperature systems and astrophysical applications.
  • Integration with machine learning emulators (e.g., IMSRG-Net) for accelerated operator flow and on-the-fly calculations.
  • Applications to ultracold atoms and further strongly correlated Fermion systems.

The Gamow IMSRG leverages advances in many-body theory, complex-energy quantum mechanics, and computational linear algebra, providing a quantitatively precise, systematically improvable approach to the structure and reactions of exotic nuclear systems (Hu et al., 2019, Zhen et al., 5 Jan 2025, Morris et al., 2015).