Higher Berry Invariant

• Higher Berry invariant is a higher-degree topological quantity for families of gapped quantum systems, defined via closed (d+2)-forms and quantized integrals.
• It employs mathematical frameworks like gerbe theory and tensor-network realizations to capture phenomena such as Chern-number pumping and anomalous boundary behavior.
• The invariant bridges physical observables with cohomological classifications, offering insights through scattering data, boundary CFT, and differential form formulations.

The higher Berry invariant is a higher-degree topological quantity attached not to a single gapped Hamiltonian but to a family of gapped quantum systems over a parameter space. In the Kapustin–Spodyneiko framework, a $d$-dimensional parametrized gapped system carries a closed $(d+2)$-form $\Omega^{(d+2)}$; its cohomology class is the higher Berry or KS invariant, while integration over a closed $(d+2)$-cycle gives a quantized higher Berry number. In one spatial dimension this degree-three structure governs phenomena such as Chern-number pumping, admits gerbe-theoretic and tensor-network realizations, and in free-fermion settings can be detected from boundary scattering data (Kapustin et al., 2020, Wen et al., 2021, Lo et al., 24 Feb 2026).

1. Terminology and conceptual scope

A parametrized gapped system is a continuous family of finite-range gapped lattice Hamiltonians

$H:X\to GH,$

with fixed spatial dimension, local Hilbert-space type, symmetry class, and parameter space $X$. When $X=\mathrm{pt}$, one recovers an ordinary gapped phase; when $X\neq \mathrm{pt}$, the family itself can carry global topology even if each individual Hamiltonian is topologically trivial as a phase over a point (Wen et al., 2021).

The literature distinguishes several related objects. In one standard usage, higher Berry curvature is a local closed differential form on parameter space, higher Berry invariant or KS invariant is the corresponding cohomology class, and higher Berry number or KS number is the quantized integral of that form over an appropriate closed cycle (Wen et al., 2021). A closely related formulation defines the higher Berry invariant as the de Rham cohomology class of a closed $(D+2)$-form $\Omega^{(D+2)}$ for a $(d+2)$0-dimensional gapped family (Kapustin et al., 2020). Other papers, especially in free-fermion boundary or tensor-network settings, sometimes use “higher Berry invariant” for the quantized closed-cycle value itself; this terminological variation is intrinsic to the current literature (Lo et al., 24 Feb 2026).

The conceptual motivation is that ordinary phase classification detects connected components of the space of gapped Hamiltonians, whereas parametrized families probe higher homotopy and higher cohomology. In one-dimensional families, a nonzero higher Berry invariant signals a gerbe structure of the ground-state family, generalizing the line-bundle structure underlying ordinary Berry phase. In the free-fermion boundary-scattering formulation, a nontrivial invariant also obstructs a globally continuous MPS tensor over parameter space and diagnoses a quantized Chern-number pump rather than an ordinary charge pump (Lo et al., 24 Feb 2026, Ohyama et al., 2024).

2. Mathematical formulations

For finite-dimensional quantum mechanics, the ordinary Berry curvature may be written in resolvent form as

$(d+2)$1

and satisfies $(d+2)$2. In extended systems, the total trace diverges, so the construction is replaced by a local/descent formalism built from local Hamiltonian terms $(d+2)$3 and chains $(d+2)$4 obeying

$(d+2)$5

For a $(d+2)$6-dimensional lattice family, the higher Berry form is

$(d+2)$7

where the $(d+2)$8 are step-like functions selecting coarse spatial directions. This form is closed, and changing the $(d+2)$9 only changes it by an exact form, so the cohomology class $\Omega^{(d+2)}$0 is well defined (Kapustin et al., 2020).

In one dimension, the KS higher Berry curvature is a closed $\Omega^{(d+2)}$1-form

$\Omega^{(d+2)}$2

with $\Omega^{(d+2)}$3 interpolating from $\Omega^{(d+2)}$4 at the far left to $\Omega^{(d+2)}$5 at the far right. For a step function $\Omega^{(d+2)}$6, this becomes

$\Omega^{(d+2)}$7

Its cohomology class is phase-invariant, and for invertible systems Kapustin and Spodyneiko argued that spherical periods are quantized: $\Omega^{(d+2)}$8 If $\Omega^{(d+2)}$9 is an oriented closed $(d+2)$0-manifold, the integrated KS number is

$(d+2)$1

More generally, for a $(d+2)$2-dimensional system over a $(d+2)$3-manifold, the higher Berry number is $(d+2)$4 (Wen et al., 2021).

For translationally invariant $(d+2)$5 free fermions with three parameters $(d+2)$6, the same cohomology class is represented by the Brillouin-zone integral of the degree-four Chern character of the occupied-band Berry curvature: $(d+2)$7 This identification ties the higher Berry invariant to characteristic classes familiar from higher-dimensional band topology (Kapustin et al., 2020).

3. Physical meaning: flow, pumping, and bulk–boundary correspondence

The main physical interpretation developed in the KS literature is that higher Berry curvature measures a flow of lower-degree Berry data. In one spatial dimension, the higher Berry curvature is a flow of ordinary Berry curvature to or from the boundary: $(d+2)$8 where $(d+2)$9 is a boundary $H:X\to GH,$0-form. In $H:X\to GH,$1 dimensions, $H:X\to GH,$2 is a flow of $H:X\to GH,$3-dimensional higher Berry curvature. A nonzero bulk KS number therefore forces anomalous boundary behavior—isolated Weyl-like singularities or more general boundary gap closings—and the boundary flux of the lower-dimensional invariant equals the bulk KS number (Wen et al., 2021).

This flow interpretation leads to a pumping picture. In the solvable $H:X\to GH,$4 model over $H:X\to GH,$5, the integrated higher Berry curvature is

$H:X\to GH,$6

and a boundary supports a single isolated Weyl point in parameter space. In the related $H:X\to GH,$7 family over $H:X\to GH,$8, the same number is expressed as a jump in boundary Chern number,

$H:X\to GH,$9

so the higher invariant is a Chern-number pump: during one cycle, a quantized Chern number $X$0 is pumped between bulk and boundary (Wen et al., 2021).

A complementary free-fermion formulation detects the same invariant from boundary scattering. For a semi-infinite gapless lead attached to a semi-infinite gapped $X$1 system, the boundary reflection matrix $X$2 defines a map $X$3. For a closed $X$4-manifold $X$5, the scattering invariant is

$X$6

In the continuum $X$7 Dirac model and in lattice realizations, this higher winding number equals the bulk higher Berry invariant; in the explicit examples,

$X$8

The approach is robust against disorder as long as the family remains gapped and, in the examples studied, diagnoses a quantized Chern-number pump without direct access to the bulk many-body wavefunction (Lo et al., 24 Feb 2026).

4. Tensor-network, wavefunction, and iMPS realizations

In one-dimensional tensor-network language, the higher Berry invariant is encoded by a gerbe rather than a line bundle. For parameterized MPS, the gerbe cocycle is extracted from the triple overlap

$X$9

and the higher Berry connection consists of $X=\mathrm{pt}$0-forms $X=\mathrm{pt}$1 on double overlaps and $X=\mathrm{pt}$2-forms $X=\mathrm{pt}$3 on patches, satisfying

$X=\mathrm{pt}$4

The global curvature is

$X=\mathrm{pt}$5

and the integral invariant is

$X=\mathrm{pt}$6

In this framework, constant-rank MPS cannot realize a nontrivial free integral higher Berry invariant; nonconstant-rank or essentially normal MPS are required for the nontrivial integral class (Ohyama et al., 2024).

A discrete numerical formulation replaces differential forms by simplicial data on a triangulated parameter space. For neighboring MPS, the dominant eigenvector $X=\mathrm{pt}$7 of the mixed transfer matrix plays the role of discrete transition data. On a triangle $X=\mathrm{pt}$8, one defines

$X=\mathrm{pt}$9

and on a tetrahedron $X\neq \mathrm{pt}$0,

$X\neq \mathrm{pt}$1

The integrated invariant is

$X\neq \mathrm{pt}$2

For the $X\neq \mathrm{pt}$3 example studied numerically, this invariant is $X\neq \mathrm{pt}$4 (Shiozaki et al., 2023).

Beyond one dimension, wavefunction-based constructions define a closed $X\neq \mathrm{pt}$5-form

$X\neq \mathrm{pt}$6

for locally parameterized short-range-entangled states, and in $X\neq \mathrm{pt}$7 an exactly solvable family over $X\neq \mathrm{pt}$8 satisfies

$X\neq \mathrm{pt}$9

(Sommer et al., 2024). In PEPS language, the $(D+2)$0 higher Berry phase is extracted from a quadruple inner product on fourfold overlaps, giving a Čech $(D+2)$1-cocycle

$(D+2)$2

with an explicit nontrivial family on $(D+2)$3 realizing the nontrivial $(D+2)$4 class in $(D+2)$5 (Ohyama et al., 2024).

A recent iMPS application makes the relation to familiar band-topological invariants explicit. Rewriting the $(D+2)$6 lattice Chern insulator as a family of translationally invariant infinite chains over $(D+2)$7, one computes a DDKS number

$(D+2)$8

The resulting phase diagram as a function of the Dirac mass is exactly congruent to the known phase diagram of the second Chern number $(D+2)$9, and with fixed orientation the paper states

$\Omega^{(D+2)}$0

This shows that higher Berry curvature can compute second Chern numbers in a manifestly quantized way (Heinsdorf et al., 24 Jun 2026).

5. Boundary conformal field theory and field-theoretic generalizations

A BCFT formulation introduces higher Berry geometry on a boundary conformal manifold, the moduli space of conformal boundary conditions connected by exactly marginal boundary deformations. The basic data are no longer pairwise overlaps but triple overlaps of boundary conditions, encoded in the phase of the disk three-point function of lightest boundary-condition-changing operators,

$\Omega^{(D+2)}$1

From this phase one defines a local $\Omega^{(D+2)}$2-form higher Berry connection

$\Omega^{(D+2)}$3

with curvature

$\Omega^{(D+2)}$4

The gauge law is that of a gerbe connection,

$\Omega^{(D+2)}$5

In compact-boson and WZW examples, $\Omega^{(D+2)}$6 coincides with the NS–NS $\Omega^{(D+2)}$7-field and $\Omega^{(D+2)}$8 with the Wess–Zumino $\Omega^{(D+2)}$9-form (Choi et al., 16 Jul 2025).

A related BCFT analysis makes the pumping interpretation fully explicit. For a two-flavor Dirac BCFT with $(d+2)$00-parametrized boundary conditions, the occupied Fock sea carries a total ordinary Berry curvature

$(d+2)$01

whose derivative gives the higher Berry curvature

$(d+2)$02

Its integral is quantized,

$(d+2)$03

Here the higher invariant measures a Chern-number pump in Fock space: multi-parameter spectral flow transports ordinary Berry curvature across the BCFT spectrum, and the same structure appears in entanglement Hamiltonians of parameterized gapped states (Wen, 16 Jul 2025).

In relativistic field theory, higher Berry curvature arises as the local response of parameterized invertible theories. For massive free Dirac fermions with spacetime-dependent mass parameters, the phase of the regulated partition function is expressed by an APS $(d+2)$04-invariant, and the local higher Berry curvature is

$(d+2)$05

This differential-form part need not exhaust the invariant: when $(d+2)$06, the theory may still carry a torsional Berry phase, a bordism invariant detected by exponentiated $(d+2)$07-invariants on nontrivial backgrounds (Choi et al., 2022).

6. Ambiguities, limitations, and neighboring notions

The phrase “higher Berry invariant” does not have a single fixed meaning across the literature. In the KS many-body framework it usually denotes the cohomology class $(d+2)$08, with a separate “KS number” for the integrated period. In gerbe-based MPS and BCFT formulations, closely related local data may instead be presented as a $(d+2)$09-form connection $(d+2)$10, a $(d+2)$11-form curvature $(d+2)$12 or $(d+2)$13, or a Čech cocycle such as $(d+2)$14. Encyclopedia treatments therefore require explicit attention to which layer—local form, cohomology class, or integrated integer—is meant in a given source (Wen et al., 2021, Ohyama et al., 2024, Choi et al., 16 Jul 2025).

Quantization and robustness also depend on hypotheses. The higher-form construction exists for interacting lattice systems quite generally, but quantization is argued most cleanly for short-range-entangled or invertible families, especially on spherical cycles (Kapustin et al., 2020). In the KS program, quantization on general closed oriented manifolds is expected but not proved in full generality (Wen et al., 2021). In BCFT, orbifold singularities can make $(d+2)$15 appear fractional unless one passes to the appropriate orbifold cohomology (Choi et al., 16 Jul 2025). In boundary scattering, the explicit detection formula is established for $(d+2)$16 gapped free-fermion families, while the interacting extension is stated as a conjecture rather than a theorem (Lo et al., 24 Feb 2026).

Several neighboring constructions are “higher” in a broader sense but are distinct from the KS higher Berry invariant of parametrized gapped families. The $(d+2)$17 Berry phase used for higher-order SPT phases is a quantized local-twist invariant for irreducible $(d+2)$18-site clusters, not a degree-$(d+2)$19 cohomological invariant of a family space (Araki et al., 2019). The three-point Bargmann invariant

$(d+2)$20

captures extrinsic geometry of projective-state manifolds beyond ordinary Berry curvature and quantum metric, but it is a different geometric program (Avdoshkin et al., 2022). Likewise, the Berry-curvature dipole

$(d+2)$21

and the quantum-metric-dipole terms governing second-order dc transport are gauge-invariant higher-order Berry-geometric response tensors, not quantized higher Berry invariants in the strict topological sense (Sodemann et al., 2015, Shibata, 21 Jun 2026).

Taken together, these works establish the higher Berry invariant as a family-level topological datum of gapped quantum matter. Its characteristic signatures are a closed higher-degree curvature form, a gerbe or higher-connection structure rather than an ordinary line bundle, quantized periods on suitable parameter cycles, and physical realizations as Chern-number pumps, anomalous boundary flow, scattering windings, and higher-dimensional characteristic classes (Kapustin et al., 2020, Wen et al., 2021, Lo et al., 24 Feb 2026)