A Geometric Theory of Fermion-to-Qubit Encodings

Technical Summary: A Geometric Theory of Fermion-to-Qubit Encodings

Problem Statement
Exact fermion-to-qubit transformations (e.g., Jordan–Wigner, Bravyi–Kitaev) are conventionally treated as algorithmic tools designed to map interacting fermionic Hamiltonians onto qubit operators for quantum simulation. Their utility is typically assessed by computational metrics such as operator locality, circuit depth, and gate count. However, a fundamental question remains unanswered: do these encodings possess an intrinsic geometric organization that reveals physical information beyond mere spectral equivalence? While every exact representation preserves the many-body spectrum, it is unclear whether the structure of the encoded Hamiltonian itself encodes physically meaningful data about interaction-driven reorganization, independent of the many-body wavefunction or the solution to the underlying quantum problem.

Methodology
The authors propose a geometric framework that treats the encoded qubit Hamiltonian as the primary object of study, rather than the quantum state. This framework utilizes two complementary representations constructed from the Pauli decomposition of the encoded Hamiltonian, $H_q = \sum J_\alpha P_\alpha$:

1. Spectral Geometry (Bravyi–Kitaev Representation):

   • The encoded Hamiltonian is mapped to a weighted hypergraph $\mathcal{H} = (V, E, w)$, where vertices represent qubits and hyperedges represent Pauli strings with weights equal to the magnitude of their coupling coefficients ($|J_\alpha|$).
   • The hypergraph is converted to a weighted graph via clique expansion to construct a combinatorial Laplacian matrix ($L = D - A$).
   • Geometric observables are derived from the Laplacian spectrum, specifically the Fiedler eigenvalue ($\lambda_2$), which serves as a measure of algebraic connectivity.
   • The Hubbard Hamiltonian is decomposed into kinetic ($H_{hop}$) and interaction ($H_{int}$) sectors, yielding two distinct hypergraphs. A dimensionless geometric observable, $\rho(U) = \lambda_2(L_{hop}) / \lambda_2(L_{int})$, is introduced to quantify the competition between kinetic delocalization and Coulomb repulsion.

2. Coupling-Space Geometry (Xia–Bian–Kais Representation):

   • The encoded Hamiltonian is transformed into an exactly equivalent diagonal Ising Hamiltonian, defining a probability measure $\mu(J)$ over effective Ising couplings.
   • The evolution of this coupling distribution as physical parameters change is quantified using optimal transport theory, specifically the Wasserstein distance ($W$).
   • This approach characterizes the "reweighting" of the Hamiltonian in coupling space, measuring the cost to transform one coupling distribution into another, independent of network topology changes.

Key Contributions and Results

• Exact Geometric Competition Law: The authors derive an exact analytical relationship for the geometric observable $\rho(U)$. Since the kinetic hypergraph weights are constant (determined by hopping $t$) and interaction weights scale linearly with interaction strength $U$, the ratio follows $\rho(U) \propto 1/U$. This allows for the determination of a characteristic interaction scale $U^*$ purely from the geometry of the encoding, without diagonalizing the many-body Hamiltonian.
• Two Geometric Universality Classes: Analysis of the characteristic scale $U^*$ reveals that Bravyi–Kitaev encodings separate into two distinct families: tapered and untapered. These classes exhibit different finite-size scaling behaviors. Tapered encodings show systematic scaling that extrapolates to a value close to the Mott crossover regime in 2D Hubbard models, while untapered encodings form a separate branch. This distinction is intrinsic to the encoding geometry, not the lattice size.
• Exact Spectral Partition: The study uncovers a robust, universal internal organization within the Bravyi–Kitaev interaction hypergraph. The Laplacian spectrum of the interaction sector partitions exactly into two families of modes:
  • Tree-dominated modes: Exactly $N_{tree} = n_q/4$ modes, where $n_q$ is the number of qubits.
  • Site-dominated modes: Exactly $N_{site} = 3n_q/4$ modes.
    This 1:3 ratio holds exactly across all system sizes studied (from $4\times4$ to $16\times16$), indicating that the binary-tree architecture of the BK transformation imposes a fixed spectral structure independent of the physical system size.
• Coupling-Space Reorganization: Using the XBK representation, the authors demonstrate that increasing interaction strength does not alter the network topology of the encoded Hamiltonian but rather induces a continuous redistribution of coupling weights. The Wasserstein distance between coupling distributions exhibits a pronounced maximum in the intermediate interaction regime, coinciding with the region of most rapid change in electronic correlations (e.g., double occupancy).
• Universality Across Models: The geometric framework is applied to the 1D and 2D Hubbard models, the spinless $t-V$ model, the single-impurity Anderson model (SIAM), and the Kitaev chain. In all cases, the spectral and transport geometric observables consistently identify the parameter regimes associated with physical crossovers, phase transitions, or Kondo screening, demonstrating that the framework captures general features of encoded quantum Hamiltonians rather than artifacts of a specific model.

Significance and Claims
The paper establishes that fermion-to-qubit encodings are not merely computational mappings but define intrinsic geometric representations of quantum many-body Hamiltonians. The authors claim that:

1. Geometric Information is Intrinsic: The structure of the encoded operator contains physically meaningful information about interaction-driven reorganization that can be extracted without solving the many-body problem or referencing the wavefunction.
2. Dual Geometric Languages: Hypergraph connectivity (BK) and coupling-space transport (XBK) provide complementary, consistent descriptions of the same physical phenomena.
3. New Framework for Understanding: This work introduces "hypergraph geometry" as a new framework for connecting quantum encodings, spectral geometry, and optimal transport. It suggests that these mappings serve a dual role: enabling quantum computation while simultaneously acting as geometric representations that reveal the structural organization of correlated quantum matter.
4. Efficiency: The geometric observables can be computed efficiently using sparse graph operations, growing smoothly with system size and avoiding the exponential cost of direct many-body diagonalization.

The authors conclude that this perspective shifts the focus from the quantum state to the encoded operator, revealing that the "geometry of the encoding" itself is a rich source of physical insight.