Interactions are missing from band-based quantum geometry
Quantum geometry describes how a quantum state changes as momentum moves across the Brillouin zone. Its central object, the quantum geometric tensor, splits into a real part called the quantum metric and an imaginary part called the Berry curvature. The quantum metric measures the distance between neighbouring ground states in Hilbert space; the Berry curvature controls topological and Hall-type responses. Both are normally computed from a band structure, and a band structure by construction does not contain electron-electron interactions.
The paper's stated question is how to put interaction effects into this geometric language. Its answer is to use gauge/gravity duality, where a strongly coupled boundary theory is described by fermions propagating in a higher-dimensional curved bulk. The authors propose to build the effective Hamiltonian, and therefore the quantum geometric tensor, from the boundary retarded Green's function rather than from a single-particle band. This is the methodological move that lets a strongly interacting system be compared with a free Dirac theory on the same geometric footing.
A pseudoscalar coupling opens the holographic gap
The insulating state is realised by adding a pseudoscalar term to the bulk fermion action, written as iηϕ(z)Γ₅ with ϕ(z) = M₅z. Here η is a coupling constant and M₅ sets the size of the gap; the authors describe ϕ(z) as an external source representing a lattice effect that creates the gap. Solving the bulk Dirac equation gives the retarded Green's function at the boundary, and its trace of the imaginary part defines the fermionic spectral function.
In the zero-temperature limit the authors use pure anti-de Sitter geometry and set the fermion mass m_f = 0, which allows an analytic Green's function. The resulting spectral function gives a gapped dispersion with excitation energy ω₁₀ = 2√(k² + M₅²). That gap is the quantity that later separates insulating from metallic behaviour in the conductivity, so the dielectric result depends on it directly.
The Green's function defines a two-band effective Hamiltonian
From the retarded Green's function the authors construct a topological Hamiltonian, H_T(k) = −G_R⁻¹(k, 0), evaluated at zero frequency. Because the holographic setup uses one flavour of fermion, the boundary Green's function is a 2×2 matrix, so H_T can be written in the Pauli basis as h₀I₂ + h·σ. Diagonalising it gives the ground and excited states from which the quantum metric and Berry curvature are computed.
This is the step that replaces a band structure: the eigenvectors of the effective Hamiltonian play the role of Bloch states, even though the underlying theory is strongly coupled. The authors note that the construction can be understood in the limit m → 1/2, where the bulk fermion is known to behave as a Fermi liquid. The approach also avoids solving bulk gauge field equations, which the paper presents as the practical advantage over earlier one-loop holographic transport calculations that suffered from convergence problems in self-energy and vertex corrections.
The quantum metric matches a Dirac material exactly
For the gapped holographic state the authors obtain explicit expressions. The quantum metric has components built from k_y² + M₅², −k_xk_y and k_x² + M₅², each divided by 4(k² + M₅²)², and its trace is (k² + 2M₅²)/[4(k² + M₅²)²]. The Berry curvature is nonzero only off-diagonal, with magnitude M₅/[2(k² + M₅²)^{3/2}]. Integrating the Berry curvature over momentum gives a Chern number equal to sign(M₅)/(4π).
These expressions are identical to those of a two-dimensional Dirac Hamiltonian with mass m once M₅ is identified with m. The authors explain why: the quantum geometric tensor depends only on the normalised vector ĥ = h/|h|, not on the overall magnitude h. Strong-coupling interaction enters the static Green's function as an overall scalar dressing that does not rotate the eigenvectors, so the metric and curvature are unchanged. The paper calls this property semi-topological — the geometry is not fully topological, but it is insensitive to that particular kind of interaction dressing. The authors explicitly flag as open whether this persists beyond two bands.
Imaginary conductivity converts geometry into a dielectric constant
Transport is computed with a Kubo formula written in terms of the quantum metric and Berry curvature. At zero temperature the intraband contribution vanishes because the Fermi-surface derivative becomes a delta function at zero density of states, leaving interband transitions as the only source of reactive and optical response. For the holographic insulator the longitudinal conductivity evaluates to σ_xx(ω) = (i/4π)[(1/4 + M₅²/ω²)log((2|M₅| + ω)/(2|M₅| − ω)) − |M₅|/ω].
Below the gap, Re[σ_xx] = 0 for ω ≤ 2M₅, which is the statement that the system is insulating. The remaining imaginary part in the low-frequency limit is σ_xx = iω/(12π|M₅|). The dielectric constant is related to conductivity by κ = 1 + σ_xx/(iωϵ₀), so substituting the low-frequency result gives κ = 1 + 1/(12πϵ₀|M₅|). The authors identify this as a main result and note it coincides with the two-dimensional Dirac material answer under m = M₅.
The same low-frequency expansion also produces a Hall-like term, σ_μν = iωδ_μν/(12π|M₅|) − sign(M₅)ϵ_μν/(4π). The first term is the reactive piece that fixes the dielectric constant; the second is the topological piece tied to the Chern number. Keeping them separate matters, because only the imaginary longitudinal part enters κ.
Gap closing drives a frequency and temperature crossover
At zero temperature the conductivity turns on when the driving frequency exceeds the gap size, so the response changes character at ω = 2M₅. The authors describe this as a frequency-driven phase transition signalling a quantum phase transition, and they compare it with the field-theory Dirac result, which shows the same insulator-to-metal behaviour once the photon energy exceeds the gap and can excite electrons across it.
At finite temperature the background becomes an AdS–Schwarzschild black hole, and temperature enters through the spectral function and the excitation gap. As temperature rises the spectral gap progressively closes and the AC conductivity evolves from sharp gapped behaviour to a smoother low-frequency response, which the authors read as a temperature-driven insulator–metal crossover. Using a retarded prescription ω → ω + iϵ to regularise a finite σ_DC, they find dσ_DC/dT > 0 below T_c = 0.5M₅ and dσ_DC/dT < 0 above it. The paper states plainly that this DC quantity is a regularised transport diagnostic of the crossover, not a complete microscopic calculation, because thermally activated intraband transport is not included.
The match is a statement about geometry, not about all dynamics
The agreement between the holographic and Dirac results is exact for the gauge-invariant quantum geometric tensor and for the quantities built from it, including the dielectric constant. The authors' explanation is that the interaction dressing is an overall scalar on the static Green's function, so it cannot reorient the eigenvectors. This is why the topology of the interacting system matches the non-interacting one, and the paper presents its analytic derivation as the reason behind a stability previously seen in full numerical work.
The scope of that claim is narrower than a general equivalence between holographic and free theories. The zero-temperature analytic Green's function is obtained in pure anti-de Sitter space with m_f = 0, an approximation the authors say lets them trace each step but whose low-energy dynamical details they would not trust. The finite-temperature transport calculation is restricted to interband transitions, with the intraband Drude weight neglected as an explicit approximation. The dielectric constant quoted is the zero-temperature, low-frequency result, and the semi-topological behaviour of the quantum metric is established here for a two-band system only.
The paper's own priority is therefore the identification M₅ ↔ m as the parameter that carries the gap, and the resulting formula κ = 1 + 1/(12πϵ₀|M₅|) as the concrete prediction. A reader comparing this with a Dirac-material estimate should match on that gap parameter and on the zero-temperature, low-frequency regime; the finite-temperature dσ_DC/dT sign change at T_c = 0.5M₅ is a separate crossover signature, not an input to the dielectric constant.
Sources used in this answer
Dielectric constant of holographic insulator from quantum geometry
The gapped holographic fermion system with pseudoscalar coupling has the same quantum metric and Berry curvature as a 2D Dirac Hamiltonian under M₅ = m, giving κ = 1 + 1/(12πϵ₀|M₅|) at zero temperature from the imaginary part of conductivity.
