Berry phase and the geometric phase

Intuition Beginner

A pendulum hung at the North Pole, set swinging in a fixed plane, would seem to anyone standing on the Earth to slowly rotate its plane of oscillation over the course of a day. The pendulum itself has no torque acting on it; the Earth simply turns beneath it. Foucault used this set-up in 1851 to demonstrate the Earth's rotation.

The same idea has a deeper twist. Imagine instead carrying a pendulum on a slow trip around a closed loop on the surface of a sphere: from the North Pole down to the equator, along a stretch of equator, then back up to the pole. When the pendulum returns to its starting point, its plane of oscillation has rotated by an angle. No force ever twisted it. The rotation is a signature of the curved geometry of the sphere itself.

The Berry phase is the quantum-mechanical version of this story. When the parameters that control a quantum system's Hamiltonian are slowly varied around a closed loop, the wavefunction returns to its starting eigenstate. But it does not return unchanged: it picks up an extra phase factor on top of the obvious one that comes from the energy of the state. This extra phase is purely geometric, depending only on the shape of the loop in parameter space and not on how slowly the loop was traced. Michael Berry pointed this out in 1984. The phase had been hiding in plain sight for half a century, absorbed quietly into arbitrary phase conventions, before anyone noticed it was a real, measurable thing.

A second, sharper example is the spin-1/2 particle. Rotate a spin-1/2 by a full turn around any axis, and its wavefunction comes back with a minus sign, not a plus sign. You have to rotate it by two full turns to get back to the original wavefunction. This sign flip is impossible to detect for a single spin in isolation, because an overall phase has no observable effect.

But if you split a beam of spin-1/2 particles in two, rotate one half by a full turn while leaving the other alone, then recombine the beams, you see a clear interference pattern that records the minus sign. This was first verified in neutron-interferometry experiments in the 1970s. The minus sign is a Berry phase: it is the geometric phase a spin picks up under a full rotation of its quantisation axis around the unit sphere of directions.

The unifying picture is that quantum states live in a curved space. The space of possible quantum states (modulo overall phase) has a natural notion of how to compare states at nearby points, and this comparison is not flat. When you slowly transport a state around a closed loop in some external parameter, the inability to consistently compare states at distant points shows up as a phase. The phase records the curvature enclosed by the loop.

The mathematical name for this structure is a gauge connection, and the Berry phase is its holonomy. Modern physics has found Berry phases everywhere: in molecules near electronic level crossings, in the quantised Hall conductance of two-dimensional electron gases, in topological insulators, in the optics of polarised light, and in protocols for robust quantum gates.

Visual Beginner

Picture the unit sphere of possible magnetic-field directions. A magnetic field of fixed strength can point anywhere on this sphere. As the field direction is slowly swept along a closed curve on the sphere, the curve sweeps out a piece of the sphere's surface. Berry's result for a spin-1/2 in such a slowly rotating field is that the extra phase the spin picks up is exactly minus half the area of that swept piece, measured in steradians.

A loop that encloses a small cap near the north pole gives a small phase. A loop that encloses the whole upper hemisphere gives a phase of minus pi. A loop that encloses the entire sphere is no loop at all and gives a phase of minus two pi, which is the same as zero for any object except a spin-1/2 — for spin-1/2, the half makes minus two pi land at minus the identity rather than at the identity.

The two side panels emphasise that the geometric phase is a feature of the curved geometry of the relevant parameter space, not a feature of the dynamics inside the loop. The pendulum and the spin-1/2 are running on entirely different physics, but the rotation of the pendulum's plane and the spin's Berry phase both equal the area enclosed by the loop on the sphere. The spin gets half the area because of its spinor character; the pendulum gets the full area because it is a classical vector being parallel-transported. Both are signatures of the underlying spherical geometry, not of any force acting on the system.

Worked example Beginner

Spin-1/2 carried around the equator. A spin-1/2 sits in the ground state of a magnetic field of fixed magnitude. The field starts pointing toward the north pole of the unit sphere, then is slowly tipped down to the equator, carried around the equator once, and tipped back up to the north pole. The path on the sphere is a closed loop consisting of two meridian arcs and one full equatorial circle, but the only piece that encloses area is the equatorial circle, which bounds the entire upper hemisphere.

Step 1. The area of the upper hemisphere on a unit sphere is two pi steradians (half of the full four pi). The loop on the sphere encloses this area.

Step 2. By Berry's half-solid-angle formula, the geometric phase the spin picks up is minus one half times two pi, which equals minus pi.

Step 3. A phase of minus pi multiplies the wavefunction by minus one. So the spin returns to the starting eigenstate but with a sign flip. This sign cannot be detected by measuring the spin's direction (the measured direction is the same), but it can be detected by interfering the spin with a reference spin that was held fixed: the interference pattern shifts by half a fringe, signalling the sign change.

What this tells us: the spin picks up a real, observable phase whose value depends only on the area enclosed by the loop on the sphere of field directions, not on how slowly the loop was traced. The minus-pi phase for a hemispheric loop is the simplest spin-1/2 Berry phase, and it is the topological reason a spin-1/2 must be rotated by four pi (two full turns) before its wavefunction comes back unchanged.

Check your understanding Beginner

Formal definition Intermediate+

Let $H(\mathbf{R})$ be a smooth family of self-adjoint Hamiltonians on a Hilbert space $\mathcal{H}$, parametrised by a point $\mathbf{R}$ in a smooth manifold $\mathcal{M}$ of control parameters. Assume that for each $\mathbf{R}\in \mathcal{M}$, the spectrum of $H(\mathbf{R})$ contains an isolated non-degenerate eigenvalue $E_n(\mathbf{R})$ with normalised eigenstate $|n(\mathbf{R})⟩$, separated by a uniform spectral gap $g>0$ from the rest of the spectrum.

The eigenstate $|n(\mathbf{R})⟩$ is determined by the eigenvalue equation $H(\mathbf{R})|n(\mathbf{R})⟩=E_n(\mathbf{R})|n(\mathbf{R})⟩$ and the normalisation $⟨n(\mathbf{R})|n(\mathbf{R})⟩=1$, up to a phase: any smooth function $\alpha :\mathcal{M}\to \mathbb{R}$ gives a new representative $|n^{\prime }(\mathbf{R})⟩=e^{i\alpha (\mathbf{R})}|n(\mathbf{R})⟩$ of the same physical state.

Berry connection. The Berry connection is the real-valued one-form on $\mathcal{M}$ defined by

$${\mathbf{A}}_n(\mathbf{R}):=i⟨n(\mathbf{R})|dn(\mathbf{R})⟩,$$

where $d$ denotes the exterior derivative in the parameter manifold $\mathcal{M}$. In local coordinates $\mathbf{R}=(R^1,\dots ,R^k)$, the component form is $A_{n,\mu }(\mathbf{R})=i⟨n(\mathbf{R})|{\partial }_{\mu }n(\mathbf{R})⟩$. The factor of $i$ makes ${\mathbf{A}}_n$ real: differentiating $⟨n|n⟩=1$ gives $⟨{\partial }_{\mu }n|n⟩+⟨n|{\partial }_{\mu }n⟩=0$, so $⟨n|{\partial }_{\mu }n⟩$ is purely imaginary, and multiplying by $i$ gives a real number.

Gauge transformation. Under a smooth phase redefinition $|n(\mathbf{R})⟩\to |n^{\prime }(\mathbf{R})⟩=e^{i\alpha (\mathbf{R})}|n(\mathbf{R})⟩$, the Berry connection transforms as

$${\mathbf{A}}_n\to {\mathbf{A}}_n^{\prime }={\mathbf{A}}_n-d\alpha .$$

This is the gauge transformation law of an abelian $U(1)$ connection. The connection itself is gauge-dependent.

Berry phase. The Berry phase along a smooth closed curve $C\subset \mathcal{M}$ is the line integral

$${\gamma }_n(C):={\oint }_C{\mathbf{A}}_n={\oint }_{C}i⟨n(\mathbf{R})|dn(\mathbf{R})⟩.$$

Under a smooth gauge transformation with $\alpha$ single-valued on $C$ (i.e. $\alpha$ returns to itself after going around the loop), the phase changes by $-{\oint }_{C}d\alpha =0$. The Berry phase is therefore gauge-invariant on closed loops with single-valued phase choice.

Total adiabatic phase. Combining with the dynamical phase from the adiabatic theorem (12.07.07), a state initialised in $|n(\mathbf{R}(0))⟩$ and evolved adiabatically along the curve $C$ traced by $\mathbf{R}(t)$ for $t\in [0,T]$ ends in the state

$$|\psi (T)⟩=e^{i{\varphi }_n^{\mathrm{dyn}}(T)+i{\gamma }_n(C)}|n(\mathbf{R}(T))⟩+O(1/T),$$

where ${\varphi }_n^{\mathrm{dyn}}(T)=-(1/\hslash ){\int }_0^T{E}_n(\mathbf{R}(t^{\prime }))d{t}^{\prime }$ is the dynamical phase. The dynamical phase depends on the energy and on the timing of the protocol; the geometric phase depends only on the path of $\mathbf{R}(t)$ in $\mathcal{M}$ and not on the parametrisation. For a closed loop $\mathbf{R}(T)=\mathbf{R}(0)$, both phases are defined modulo $2\pi$ and the total adiabatic phase is the observable phase that interferometry experiments measure.

Berry curvature. The Berry curvature is the gauge-invariant two-form

$$F_n(\mathbf{R}):=d{\mathbf{A}}_n(\mathbf{R}),$$

with component form $F_{n,\mu \nu }(\mathbf{R})={\partial }_{\mu }{A}_{n,\nu }-{\partial }_{\nu }{A}_{n,\mu }$. Under the gauge transformation ${\mathbf{A}}_n\to {\mathbf{A}}_n-d\alpha$, the curvature is unchanged because $d^2=0$. By Stokes's theorem, the Berry phase around a closed loop equals the curvature integrated over any oriented surface $S\subset \mathcal{M}$ bounded by the loop:

$${\gamma }_n(\partial S)={\oint }_{\partial S}{\mathbf{A}}_n={\int }_S{F}_n.$$

Spin-1/2 worked example. Consider a spin-1/2 in a magnetic field $\mathbf{B}=B\hat{\mathbf{n}}$ of fixed magnitude $B$ and slowly varying direction $\hat{\mathbf{n}}(\mathbf{R})$, with parameter space $\mathcal{M}=S^2$ (the unit sphere of directions). The Hamiltonian is $H=-\mu B\hat{\mathbf{n}}\cdot \mathbf{\sigma }$ with eigenvalues $\mp \mu B$ and eigenstates $|\pm \hat{\mathbf{n}}⟩$ aligned with or against the field. The ground state in spherical coordinates $\hat{\mathbf{n}}(\theta ,\varphi )=(\sin \theta \cos \varphi ,\sin \theta \sin \varphi ,\cos \theta )$ is

$$|n(\theta ,\varphi )⟩=\left(\begin{array}{c}\cos (\theta /2)\\ e^{i\varphi }\sin (\theta /2)\end{array}\right),$$

with the standard phase convention. Compute the Berry connection components:

$$A_{\theta }=i⟨n|{\partial }_{\theta }n⟩=0,A_{\varphi }=i⟨n|{\partial }_{\varphi }n⟩=-{\sin }^2(\theta /2)=-\frac{1-\cos \theta }{2}.$$

The Berry curvature is the two-form $F=dA=({\partial }_{\theta }{A}_{\varphi }-{\partial }_{\varphi }{A}_{\theta })d\theta \wedge d\varphi =-\frac{1}{2}\sin \theta d\theta \wedge d\varphi$, which is one half of the negative of the standard area form on the unit sphere.

For a closed loop $C$ bounding an oriented surface $S\subset S^2$ on the sphere of field directions, the Berry phase of the ground state is

$${\gamma }_{-}(C)={\int }_{S}F=-\frac{1}{2}{\int }_S\sin \theta d\theta d\varphi =-\frac{1}{2}\Omega (S),$$

where $\Omega (S)={\int }_S\sin \theta d\theta d\varphi$ is the solid angle subtended by $S$. This is Berry's celebrated half-solid-angle formula: the geometric phase of a spin-1/2 in a slowly rotating magnetic field is exactly minus half the solid angle enclosed by the loop on the sphere. The excited state $|+\hat{\mathbf{n}}⟩$ picks up the opposite sign, ${\gamma }_{+}(C)=+\Omega (S)/2$, as required by the relation ${\gamma }_{+}+{\gamma }_{-}=0\mathrm{mod}2\pi$ (the total phase of all states must be zero modulo a single Dirac monopole charge at the degeneracy point $\mathbf{B}=0$).

Aharonov-Bohm as a Berry phase. An electron travels along a closed loop in a region where the magnetic field ${\mathbf{B}}_{\mathrm{phys}}=0$ but the vector potential ${\mathbf{A}}_{\mathrm{em}}$ is non-zero (the electron loop encircles a flux tube but does not enter it). The wavefunction picks up the phase ${\gamma }_{\mathrm{AB}}=(e/\hslash ){\oint }_C{\mathbf{A}}_{\mathrm{em}}\cdot d\mathbf{r}=e\Phi /\hslash$ where $\Phi$ is the enclosed magnetic flux. In Berry's language, the parameter is the position of the electron, ${\mathbf{A}}_{\mathrm{em}}$ plays the role of the Berry connection (up to the coupling $e/\hslash$), and the Aharonov-Bohm phase is the Berry phase of the loop. The Aharonov-Bohm effect (1959) is, in retrospect, the first physical observation of a Berry phase, twenty-five years before Berry identified the general structure.

Geometric vs dynamical phase. Two key features distinguish the geometric phase from the dynamical phase. First, the geometric phase is independent of how slowly the loop is traced — doubling the protocol time has no effect, as long as the protocol remains adiabatic. The dynamical phase, by contrast, is proportional to the protocol time (through the energy times time integral). Second, the geometric phase is a property of the loop in parameter space, not of the state's intrinsic dynamics — it depends on how the Hamiltonian is steered through parameter space, not on what the Hamiltonian is at any one point. The dynamical phase depends on the energy spectrum and is local in time; the geometric phase depends on the global topology of the loop and is non-local.

Counterexamples to common slips Intermediate+

• The Berry phase is not zero just because the Hamiltonian is invariant. If $H(\mathbf{R})=H({\mathbf{R}}^{\prime })$ for $\mathbf{R}$ and ${\mathbf{R}}^{\prime }$ on the loop, the eigenstates at those two points are the same physical state but may have inequivalent phase representatives, and the loop integral $\oint {\mathbf{A}}_n$ records the cumulative phase mismatch.

• Gauge invariance requires a single-valued phase choice on the loop. If the eigenstate $|n(\mathbf{R})⟩$ cannot be smoothly defined on a single chart covering the loop — for example, the spin-1/2 ground state $|n(\theta ,\varphi )⟩$ is singular at $\theta =\pi$ where the $e^{i\varphi }\sin (\theta /2)$ component is undefined in the $\varphi$ phase — the loop integral splits into contributions on overlapping charts plus transition-function contributions on the chart overlaps. Wu and Yang (1976) gave the general bundle-theoretic resolution.

• The Berry phase is observable, but only relatively. An overall phase on the wavefunction has no observable consequence on its own. The Berry phase becomes observable when the state is interfered with a reference state, as in spin echo, neutron interferometry, or Mach-Zehnder optical setups. The phase shifts the interference pattern by an amount equal to the Berry phase difference between the two arms of the interferometer.

• The non-adiabatic version requires care. The Aharonov-Anandan generalisation gives a geometric phase for cyclic non-adiabatic evolution, but the integrand involves $⟨\psi |\dot{\psi }⟩$ rather than $⟨n|\dot{n}⟩$, with the difference depending on whether the wavefunction is the instantaneous eigenstate or the actual solution of the Schrodinger equation. The two phases coincide in the adiabatic limit and differ at finite protocol speed.

Key theorem with proof Intermediate+

Theorem (Berry's half-solid-angle formula). Let a spin-1/2 be in the ground state $|-\hat{\mathbf{n}}⟩$ of a magnetic field $\mathbf{B}(t)=B\hat{\mathbf{n}}(t)$ of fixed magnitude $B>0$ and slowly varying direction $\hat{\mathbf{n}}(t)\in S^2$, with the direction tracing a smooth closed loop $C\subset S^2$. Then the Berry phase accumulated over one full traversal of the loop is

$${\gamma }_{-}(C)=-\frac{1}{2}\Omega (C),$$

where $\Omega (C)$ is the solid angle subtended by any oriented surface $S\subset S^2$ bounded by $C$ (with orientation matching the direction of traversal).

Proof. Write $\hat{\mathbf{n}}$ in spherical coordinates $\hat{\mathbf{n}}(\theta ,\varphi )=(\sin \theta \cos \varphi ,\sin \theta \sin \varphi ,\cos \theta )$ on the unit sphere. The Hamiltonian $H(\hat{\mathbf{n}})=-\mu B\hat{\mathbf{n}}\cdot \mathbf{\sigma }$ has eigenvalues $\mp \mu B$ with eigenvectors

$$|-\hat{\mathbf{n}}⟩=\left(\begin{array}{c}\cos (\theta /2)\\ e^{i\varphi }\sin (\theta /2)\end{array}\right),|+\hat{\mathbf{n}}⟩=\left(\begin{array}{c}-e^{-i\varphi }\sin (\theta /2)\\ \cos (\theta /2)\end{array}\right),$$

in the standard $|\uparrow ⟩,|\downarrow ⟩$ basis of ${\sigma }_z$ eigenstates. (One may verify by direct matrix multiplication that these are eigenvectors with the stated eigenvalues.)

Compute the Berry connection on $S^2$ for the ground state $|-\hat{\mathbf{n}}⟩$. The component along ${\partial }_{\theta }$ is

$$A_{\theta }=i⟨-\hat{\mathbf{n}}|{\partial }_{\theta }|-\hat{\mathbf{n}}⟩=i\left[\cos (\theta /2)\cdot (-\frac{1}{2}\sin (\theta /2))+e^{-i\varphi }\sin (\theta /2)\cdot e^{i\varphi }(\frac{1}{2}\cos (\theta /2))\right]=0.$$

The component along ${\partial }_{\varphi }$ is

$$A_{\varphi }=i⟨-\hat{\mathbf{n}}|{\partial }_{\varphi }|-\hat{\mathbf{n}}⟩=i\left[\cos (\theta /2)\cdot 0+e^{-i\varphi }\sin (\theta /2)\cdot i{e}^{i\varphi }\sin (\theta /2)\right]=-{\sin }^2(\theta /2).$$

So the Berry connection one-form is ${\mathbf{A}}_{-}=-{\sin }^2(\theta /2)d\varphi$, which in the $1-\cos \theta =2{\sin }^2(\theta /2)$ identity equals $-\frac{1-\cos \theta }{2}d\varphi$.

The Berry curvature is the exterior derivative

$$F_{-}=d{\mathbf{A}}_{-}={\partial }_{\theta }(-{\sin }^2(\theta /2))d\theta \wedge d\varphi =-\sin (\theta /2)\cos (\theta /2)d\theta \wedge d\varphi =-\frac{1}{2}\sin \theta d\theta \wedge d\varphi .$$

This is minus one half times the standard area two-form $\sin \theta d\theta \wedge d\varphi$ on the unit sphere.

By Stokes's theorem on the parameter manifold $S^2$,

$${\gamma }_{-}(C)={\oint }_C{\mathbf{A}}_{-}={\int }_S{F}_{-}=-\frac{1}{2}{\int }_S\sin \theta d\theta d\varphi =-\frac{1}{2}\Omega (S),$$

where $\Omega (S)$ is the area of the oriented surface $S$ in the unit sphere's standard area measure — that is, the solid angle subtended at the origin by the surface. The result ${\gamma }_{-}(C)=-\Omega (C)/2$ is the half-solid-angle formula. $\square$

Remark (the monopole at the origin). The Berry curvature $F_{-}$ is a smooth two-form everywhere on $S^2$ except at the points where the chart breaks down. In the chart used above, the singularity is at $\theta =\pi$ (the south pole), where the phase $e^{i\varphi }$ in $|-\hat{\mathbf{n}}⟩$ is undefined. A second chart with $|-\hat{\mathbf{n}}{⟩}^{\prime }=(e^{-i\varphi }\cos (\theta /2),\sin (\theta /2){)}^{\top }$ covers the south pole but is singular at the north pole. The two chart conventions differ by the gauge transformation $|-\hat{\mathbf{n}}{⟩}^{\prime }=e^{-i\varphi }|-\hat{\mathbf{n}}⟩$, and the corresponding Berry curvatures are equal: $F_{-}^{\prime }=F_{-}$. The fact that no global smooth gauge choice exists on $S^2$ is the topological signature of a magnetic monopole sitting at the origin of parameter space $\mathbf{R}=\mathbf{B}=0$, where the two spin-1/2 levels become degenerate. Integrating the Berry curvature $F_{-}$ over the entire sphere gives the monopole's total magnetic flux

$${\int }_{S^2}{F}_{-}=-\frac{1}{2}{\int }_{S^2}\sin \theta d\theta d\varphi =-\frac{1}{2}\cdot 4\pi =-2\pi ,$$

which is exactly the Dirac quantisation condition for a monopole of charge $1/2$. The half-solid-angle formula and the monopole at the level-crossing are two faces of the same geometric structure.

Bridge. The Berry connection ${\mathbf{A}}_n$, its curvature $F_n$, and the resulting loop holonomy ${\gamma }_n(C)$ are the geometric refinement of the adiabatic theorem (12.07.07). They organise the leading geometric correction to the dynamical phase into a gauge-theoretic structure on the parameter manifold of the Hamiltonian. The Wilczek-Zee non-abelian generalisation for degenerate subspaces (Master tier below) replaces the $U(1)$ phase by a $U(d)$ unitary, with the connection becoming matrix-valued. The Aharonov-Bohm phase is a special case where the parameter is the electron's position and the connection is the electromagnetic vector potential. The TKNN formula for the integer quantum Hall conductance integrates the Berry curvature over the Brillouin zone to give a quantised Chern number. The Berry phase is therefore not just a geometric refinement; it is the unifying gauge-theoretic structure under which adiabatic dynamics, gauge couplings, topological insulators, and holonomic quantum gates are all special cases.

Exercises Intermediate+

Lean formalization Intermediate+

Mathlib's current functional-analysis and differential-geometry stack provides much of the prerequisite material for a Berry-phase formalisation but does not yet contain the finished structure. The core ingredients available are: the finite-dimensional spectral theorem for Hermitian matrices (Mathlib.LinearAlgebra.Matrix.Spectrum), smooth dependence of an isolated non-degenerate eigenvalue on a parameter through the analytic perturbation theory of resolvents (Mathlib.Analysis.NormedSpace.Spectrum), the resolvent and Riesz contour-integral representations of spectral projectors, smooth one-parameter families of bounded operators (Mathlib.Topology.ContinuousFunction.Bounded), the exterior algebra and exterior derivative for smooth differential forms on a manifold (Mathlib.Geometry.Manifold.MFDeriv, with cohomology infrastructure under active development), and Stokes's theorem for forms on manifolds with boundary at the level of differential-topology lemmas in Mathlib.Geometry.Manifold.Diffeomorph.

What is missing for a full Berry-phase formalisation is the synthesis of these pieces into the spectral-line-bundle picture: the smooth eigenstate map $\mathbf{R}\to |n(\mathbf{R})⟩$ as a section of a smooth complex line bundle on the parameter manifold, the Berry connection ${\mathbf{A}}_n=i⟨n|dn⟩$ as a $u(1)$-valued one-form on the parameter manifold, the Berry curvature $F_n=d{\mathbf{A}}_n$ as its exterior derivative, the gauge transformation ${\mathbf{A}}_n\to {\mathbf{A}}_n-d\alpha$ for smooth $\alpha :\mathcal{M}\to \mathbb{R}$, and the loop integral ${\gamma }_n={\oint }_C{\mathbf{A}}_n$ as a gauge-invariant real number modulo $2\pi$. The Chern-number integrality $C_1=(1/2\pi ){\int }_{S^2}F\in \mathbb{Z}$ would require the index theorem for the Dolbeault complex on a compact complex manifold, which is well beyond Mathlib's current cohomology infrastructure (Hodge theory, sheaf cohomology, and the de Rham theorem are at the early-development stage).

A natural first deliverable is the finite-dimensional Berry phase: for a smooth family $|n(\mathbf{R})⟩$ of normalised eigenvectors of an $N\times N$ Hermitian matrix $H(\mathbf{R})$ parametrised by $\mathbf{R}$ in a smooth manifold $\mathcal{M}$, define ${\mathbf{A}}_n(\mathbf{R})=i⟨n(\mathbf{R})|dn(\mathbf{R})⟩$ as a smooth real one-form, prove ${\oint }_C{\mathbf{A}}_n$ is invariant under any smooth gauge transformation $|n⟩\to e^{i\alpha (\mathbf{R})}|n⟩$ with $\alpha$ single-valued on $C$, and verify the spin-1/2 worked example yields ${\gamma }_{-}=-\Omega (C)/2$ for a loop $C$ on the unit sphere of magnetic-field directions enclosing solid angle $\Omega (C)$. This deliverable uses only finite-dimensional linear algebra (already in Mathlib), smooth maps between manifolds, and the spherical-area integration formula.

A more ambitious target is the Berry curvature formula and its monopole structure: for a non-degenerate eigenstate with isolated eigenvalue, the Berry curvature has the gap-resolved form (Exercise 4) with monopole-like singularities at level degeneracies, and the integral over a small sphere surrounding a generic level crossing of two eigenvalues equals $\pm 2\pi$ (the Dirac monopole charge of $\pm 1/2$ for the half-integer of the chern-number, the sign depending on which eigenvalue is being tracked). This requires the Riesz-projector representation and the smooth-perturbation theory of two-state level crossings, both at the level of Mathlib's current spectral-theory infrastructure.

The Wilczek-Zee non-abelian generalisation, the TKNN topological-band-invariant integrality, the K-theoretic classification of topological phases, and the Aharonov-Anandan non-adiabatic phase all sit at higher altitudes of differential geometry and algebraic topology and are well beyond Mathlib's current scope. lean_status: none reflects the gap. The structured proof in this unit's Key theorem is the target a future formalisation would aim at; Tyler's review attests intermediate-tier correctness in the meantime.

Advanced results Master

Pancharatnam phase: the optical precursor

Long before Berry's 1984 paper, Pancharatnam published in 1956 Generalised theory of interference and its applications in the Proceedings of the Indian Academy of Sciences a complete analysis of the geometric phase for polarised light. Pancharatnam was studying interference between two beams of light in arbitrary polarisation states, and he discovered that the relative phase between two beams depends not just on the optical path-length difference (the dynamical phase) but on a geometric quantity that records the polarisation history of the beams. For three different polarisation states $|p_1⟩,|p_2⟩,|p_3⟩$ on the Poincare sphere of polarisations, the closed-triangle phase $\arg ⟨p_1|p_2⟩⟨p_2|p_3⟩⟨p_3|p_1⟩$ is non-zero and equals half the area of the triangle on the Poincare sphere — exactly Berry's half-solid-angle formula thirty years early.

Berry himself credited Pancharatnam explicitly in the 1984 paper, noting that the geometric phase for two-state quantum systems had been derived in essentially modern form decades before. Pancharatnam's analysis used the Poincare-sphere representation of polarisation (which is mathematically identical to the Bloch sphere for a spin-1/2), and his triangle formula is a discrete version of the loop-integral formulation that Berry made general. The optics-quantum identification was made fully explicit by Ramaseshan and Nityananda (1986) and elaborated by Berry (1987) Journal of Modern Optics 34, 1401 in his own retrospective on the Pancharatnam connection.

The Pancharatnam phase has had a vigorous independent life in optics: it appears in the interference of polarised beams in cyclic polarisation rotators (Bhandari and Samuel 1988), in the dynamics of polarised photons in twisted optical fibres (Tomita and Chiao 1986), in the design of polarisation-encoded quantum communication protocols, and in the analysis of polarisation-sensitive imaging systems. The Pancharatnam-Berry phase optical elements (PBOE) are a class of subwavelength-engineered optical components whose phase response is determined by the geometric phase rather than by the conventional dynamical phase from optical path length, with applications in metasurface lenses, vortex beams, and structured light.

Wilczek-Zee non-abelian holonomy: degenerate subspaces

Wilczek and Zee's 1984 paper Appearance of gauge structure in simple dynamical systems in Physical Review Letters generalised Berry's construction to degenerate eigenspaces. When the relevant eigenvalue $E_n(\mathbf{R})$ is $d$-fold degenerate, the eigenstate is not determined up to a $U(1)$ phase but is an arbitrary basis of a $d$-dimensional subspace, determined only up to a $U(d)$ basis transformation. The Berry connection is therefore matrix-valued: in any local orthonormal basis $\{|n_a(\mathbf{R})⟩{\}}_{a=1}^d$ of the degenerate subspace, the connection is the $d\times d$ Hermitian matrix-valued one-form $A_{ab,\mu }(\mathbf{R})=i⟨n_a(\mathbf{R})|{\partial }_{\mu }{n}_b(\mathbf{R})⟩$.

Under a smooth gauge transformation $|n_a⟩\to {\sum }_b{U}_{ba}(\mathbf{R})|n_b⟩$ with $U(\mathbf{R})\in U(d)$, the connection transforms as $A_{\mu }\to U^{†}{A}_{\mu }U+i{U}^{†}{\partial }_{\mu }U$ (Exercise 9). The curvature $F_{\mu \nu }={\partial }_{\mu }{A}_{\nu }-{\partial }_{\nu }{A}_{\mu }-i[A_{\mu },A_{\nu }]$ is a covariant $d\times d$ matrix-valued two-form. The holonomy around a closed loop $C$ is the path-ordered exponential

$$\mathcal{P}\exp \left[i{\oint }_C{A}_{\mu }d{R}^{\mu }\right]\in U(d),$$

a unitary matrix in $U(d)$ that depends on the loop and on the connection. Under the gauge transformation, the holonomy transforms by conjugation: $W^{\prime }=U(0{)}^{†}WU(0)$, with eigenvalues and trace gauge-invariant.

The non-commutativity of the connection — the commutator $[A_{\mu },A_{\nu }]$ in the curvature — is the structural ingredient that makes Wilczek-Zee genuinely non-abelian: two loops in different orders generally give different holonomies, $\mathcal{P}\exp [i{\oint }_{C_1}A]\cdot \mathcal{P}\exp [i{\oint }_{C_2}A]\ne \mathcal{P}\exp [i{\oint }_{C_2}A]\cdot \mathcal{P}\exp [i{\oint }_{C_1}A]$. The associated mathematical structure is a $U(d)$-principal bundle on the parameter manifold $\mathcal{M}$ with a connection in the Lie algebra $u(d)$, and the path-ordered exponential is the parallel-transport map of the connection along the curve.

The Wilczek-Zee non-abelian Berry phase has a celebrated technological consequence: holonomic quantum computation (Zanardi-Rasetti 1999 Phys. Lett. A 264, 94). The construction takes a controllable Hamiltonian with a degenerate ground subspace large enough to encode the qubits of a quantum computer (typically $2^n$-fold degeneracy for $n$ qubits). Unitary gates on the qubits are implemented by traversing loops in the control parameter space; the gate is the non-abelian holonomy of the loop. The crucial property is that the gate depends only on the loop's geometric shape, not on the protocol speed, so the gate is intrinsically robust against timing errors as long as the protocol stays in the adiabatic regime. Universal sets of holonomic gates have been constructed for trapped ions (Duan-Cirac-Zoller 2001), nitrogen-vacancy centres in diamond (Zhang-Wang-Tan-Pan-Liu 2014), and superconducting qubits (Abdumalikov-Fink-Juliusson-Pechal-Berger-Wallraff-Filipp 2013).

The Wilczek-Zee connection also appears in the Born-Oppenheimer approximation for molecules with conical intersections. At a conical intersection of two electronic levels, the two-dimensional electronic eigenspace is degenerate at a single nuclear configuration, and the Berry connection on the surrounding nuclear configuration space is non-abelian $U(2)$. The molecular Aharonov-Bohm effect of Mead-Truhlar 1979 — a topological geometric-phase contribution to vibrational dynamics that affects molecular spectroscopy — is the holonomy of this non-abelian connection around the conical intersection.

Conical intersections and the Mead-Truhlar molecular Aharonov-Bohm effect

In the Born-Oppenheimer separation of electronic and nuclear motion in molecules, the electronic energy as a function of nuclear configuration $E_{\mathrm{el}}(\mathbf{R})$ generically has conical intersections — isolated points or sub-manifolds in nuclear configuration space where two electronic eigenvalues become degenerate. The locus of degeneracy has codimension 2 in the space of real Hamiltonians (two parameters must be tuned to bring two real eigenvalues together) and codimension 3 in the space of complex Hamiltonians. For real molecular Hamiltonians (the typical case for symmetric molecules at fixed nuclear configurations), conical intersections are isolated points in a two-parameter sub-manifold and are generic.

The mathematical phenomenon at a conical intersection is that the adiabatic real-valued electronic wavefunction $|n_{\mathrm{el}}(\mathbf{R})⟩$ chosen continuously must flip sign when the nuclear configuration is transported around a loop enclosing the conical intersection. This is the Longuet-Higgins sign theorem (1958), originally derived for the Jahn-Teller distortion of cyclic symmetric molecules with electronic degeneracy at the symmetric configuration. The sign flip is exactly the $\pi$ Berry phase one finds by applying Berry's formula to a real Hamiltonian (Exercise 6).

Mead and Truhlar 1979 J. Chem. Phys. 70, 2284 pointed out that the sign flip should be treated as a real, physically observable Berry phase that affects the nuclear quantum dynamics. The geometric-phase contribution to the nuclear-motion wavefunction produces a $\pi$ shift in interference patterns between trajectories that go around the conical intersection on opposite sides, with measurable consequences for molecular spectra (the photodissociation rates, the vibrational spacings, and the rotational fine structure of molecules with low-lying conical intersections). Modern computational chemistry includes the Berry-phase contribution in trajectory-surface-hopping and other non-adiabatic dynamics algorithms; ignoring the geometric phase leads to systematic errors in observed spectroscopic features near conical intersections.

The molecular Aharonov-Bohm effect is a precise analogue of the original Aharonov-Bohm effect: in the Aharonov-Bohm setup, the electron's wavefunction picks up a topological phase when transported around a flux tube it never enters; in the molecular Aharonov-Bohm setup, the nuclear wavefunction picks up a topological phase when transported around a conical intersection it never reaches. The flux tube and the conical intersection both act as defects in the parameter space that source a non-vanishing holonomy. The connection between the two effects was made fully explicit by Berry 1984 and Berry-Robbins 1993.

TKNN formula and integer quantum Hall effect

The integer quantum Hall effect was discovered by von Klitzing, Dorda, and Pepper in 1980 in a silicon MOSFET at high magnetic field and low temperature. The Hall conductance was observed to take only integer multiples of $e^2/h$, with the integer plateaus extending over wide ranges of magnetic field and electron density and remaining stable to a few parts in $10^9$ in modern measurements. The plateau quantisation is the most precise measurement of the fine-structure constant in physics and is the basis of the SI redefinition of the ampere via the quantum Hall plateau and the Josephson constant.

Thouless, Kohmoto, Nightingale, and den Nijs (TKNN) 1982 Phys. Rev. Lett. 49, 405 explained the integer quantisation. They computed the Hall conductance of a non-interacting electron gas in a two-dimensional periodic potential using the Kubo formula and found that for each filled Bloch band $n$, the contribution to the Hall conductance is

$${\sigma }_{xy}^{(n)}=-\frac{e^2}{h}\cdot \frac{1}{2\pi }{\int }_{\mathrm{BZ}}{F}_n(\mathbf{k})d{k}_{x}d{k}_y,$$

where $F_n(\mathbf{k})={\partial }_{k_x}{A}_{n,y}-{\partial }_{k_y}{A}_{n,x}$ is the Berry curvature of the $n$-th Bloch band over the Brillouin zone, computed from the Bloch states $|u_n(\mathbf{k})⟩$ at each $\mathbf{k}$. The integral is an integer — the first Chern number $C_1^{(n)}$ of the Bloch line bundle over the Brillouin-zone torus $T^2$. The Hall conductance is therefore quantised in integer multiples of $e^2/h$, with the integer equal to the sum of Chern numbers of the filled bands.

The topological invariance of the Chern number means the plateau is robust against any smooth perturbation that does not close the spectral gap. Disorder, electron-electron interactions, lattice imperfections — none of these change the integer Chern number as long as the gap stays open. This is the source of the plateau's robustness and its extraordinary measurement precision.

The TKNN formula was the first identification of a measurable transport coefficient with a topological invariant of a quantum band structure. Its significance was recognised slowly: Thouless received the 2016 Nobel Prize in Physics jointly with Haldane and Kosterlitz for the theoretical discoveries of topological phase transitions and topological phases of matter, with the TKNN formula as the principal contribution. The TKNN paper launched the modern field of topological band theory and was the conceptual ancestor of the entire programme of topological insulators, topological superconductors, and topological semimetals.

Karplus-Luttinger anomalous Hall effect: Bloch electrons in momentum space

The TKNN formula is the special case of a more general phenomenon: Berry curvature in momentum space modifies the semiclassical equations of motion of Bloch electrons in a way that produces anomalous transport coefficients. The original observation was Karplus and Luttinger 1954 Physical Review 95, 1154, who derived an additional contribution to the Hall conductance in ferromagnetic materials arising from the Berry curvature of the band structure even in the absence of an external magnetic field. The anomalous Hall conductance per filled band is ${\sigma }_{xy}^{(n),\mathrm{an}}=-(e^2/\hslash )\int F_n(\mathbf{k})d^{2}k/(2\pi {)}^2$, the same formula as the TKNN expression but with the Berry curvature now sourced by the spin-orbit coupling in the band structure rather than by an external orbital magnetic field.

The modern semiclassical equations of motion for a Bloch wave packet in an electric field $\mathbf{E}$ and an external magnetic field $\mathbf{B}$, including the Berry-curvature correction, are (Sundaram-Niu 1999 Physical Review B 59, 14915):

$$\dot{\mathbf{r}}=\frac{1}{\hslash }\frac{\partial E_n(\mathbf{k})}{\partial \mathbf{k}}-\dot{\mathbf{k}}\times {\mathbf{F}}_n(\mathbf{k}),$$

$$\hslash \dot{\mathbf{k}}=-e\mathbf{E}-e\dot{\mathbf{r}}\times \mathbf{B}.$$

The second term on the right-hand side of the first equation is the anomalous velocity from the Berry curvature, perpendicular to both $\dot{\mathbf{k}}$ and the curvature vector ${\mathbf{F}}_n$. In the presence of an external electric field $\mathbf{E}$, $\dot{\mathbf{k}}\propto \mathbf{E}$, so the anomalous velocity is $\propto \mathbf{E}\times {\mathbf{F}}_n$, perpendicular to the electric field — a Hall-like response with the Berry curvature replacing the magnetic field. Integrating over the occupied states gives the intrinsic anomalous Hall conductance.

The Karplus-Luttinger anomalous Hall effect was rediscovered and reframed in the topological language by Jungwirth-Niu-MacDonald 2002 Phys. Rev. Lett. 88, 207208 and Onoda-Nagaosa 2002 Journal of the Physical Society of Japan 71, 19. It is now understood as a generic property of band structures with non-vanishing Berry curvature — including those of ferromagnetic conductors with spin-orbit coupling, topological semimetals (Weyl semimetals exhibit a particularly large intrinsic anomalous Hall effect from the Berry monopole at the Weyl point), and chiral magnets.

The companion anomalous Nernst effect (a transverse thermal-gradient-driven voltage) and spin Hall effect (a transverse spin current driven by an electric field) are also Berry-curvature responses in the band structure, with the same topological-invariant character as the intrinsic Hall conductance.

Topological insulators: Z2 invariants and beyond

The integer Chern number classifies two-dimensional band structures that break time-reversal symmetry — the integer quantum Hall effect. For time-reversal-symmetric two-dimensional band structures, the Chern number must vanish (time reversal flips the Berry curvature so the integral over the Brillouin zone is zero by symmetry). But Kane and Mele 2005 Phys. Rev. Lett. 95, 146802 showed that a more refined ${\mathbb{Z}}_2$-valued invariant can be defined for time-reversal-invariant two-dimensional band structures, classifying them into two distinct topological phases: an ordinary band insulator (no protected edge states) and a quantum spin Hall insulator (protected counter-propagating edge states with opposite spin).

The Kane-Mele ${\mathbb{Z}}_2$ invariant is computed from the time-reversal-invariant points of the Brillouin zone using the Pfaffian of the sewing matrix of time-reversal partners. The construction was experimentally verified in HgTe/CdTe quantum wells (Konig et al. 2007 Science 318, 766) confirming the BHZ model (Bernevig-Hughes-Zhang 2006 Science 314, 1757) and inaugurating the modern era of topological insulators.

The three-dimensional generalisation (Fu-Kane-Mele 2007 Phys. Rev. Lett. 98, 106803) gave the four ${\mathbb{Z}}_2$ invariants $({\nu }_0;{\nu }_1,{\nu }_2,{\nu }_3)$ that classify three-dimensional time-reversal-symmetric band insulators into strong and weak topological insulators. The strong topological insulator (the ${\nu }_0=1$ case) has a single odd number of Dirac cones on every surface, and the surface states form a "topologically protected metal" — robust to disorder, edge geometry, and interactions, as long as the bulk gap and time-reversal symmetry are preserved.

The K-theoretic classification of free-fermion topological phases by Schnyder-Ryu-Furusaki-Ludwig 2008 Phys. Rev. B 78, 195125 and Kitaev 2009 AIP Conf. Proc. 1134, 22 systematised the framework: for each of the 10 Altland-Zirnbauer symmetry classes, the topological invariants in each spatial dimension form a periodic pattern (the periodic table of topological phases) determined by the real and complex K-theory groups of the appropriate classifying spaces. The Berry curvature and its generalisations (Pfaffian invariants, Chern-Simons forms in odd dimensions, mod-2 invariants in symplectic-symmetric classes) provide concrete representatives of the abstract K-theoretic classes.

The non-interacting classification has been extended in many directions: topological crystalline insulators (Fu 2011) protected by point-group symmetries; higher-order topological insulators (Benalcazar-Bernevig-Hughes 2017) with protected corner or hinge states rather than surface states; symmetry-protected topological phases in interacting systems (Chen-Gu-Liu-Wen 2013, the cohomology classification); topological superconductors with Majorana boundary modes (Kitaev 2001, the one-dimensional p-wave chain). All of these classifications rest on the Berry-phase apparatus, in particular on the topological invariance of the band-structure Chern numbers and their generalisations.

Aharonov-Anandan generalisation: non-adiabatic cyclic phase

The Aharonov-Anandan 1987 Phys. Rev. Lett. 58, 1593 paper removed the adiabatic-evolution restriction from the Berry phase. Aharonov and Anandan considered an arbitrary cyclic solution $|\psi (t)⟩$ of the time-dependent Schrodinger equation, with $|\psi (T)⟩=e^{i\chi }|\psi (0)⟩$ for some real $\chi$ but with the protocol not necessarily slow. They defined the dynamical phase ${\varphi }^{\mathrm{dyn}}=-(1/\hslash ){\int }_0^T⟨\psi |H|\psi ⟩dt$ and the Aharonov-Anandan geometric phase $\beta =\chi -{\varphi }^{\mathrm{dyn}}$.

Aharonov and Anandan proved that $\beta$ depends only on the closed curve $C$ traced by the state in the projective Hilbert space ${\mathbb{CP}}^N=({\mathbb{C}}^N\setminus \{0\})/{\mathbb{C}}^{\ast }$ (the space of physical states modulo overall phase), and not on the parametrisation or the speed of the evolution. The geometric phase has the form $\beta ={\oint }_{\tilde{C}}i⟨\tilde{\psi }|d\tilde{\psi }⟩$, with $\tilde{C}$ a lift of $C$ to the unit sphere $S^{2N-1}\subset {\mathbb{C}}^N$ of normalised states (Exercise 7). The integrand is the natural Hermitian connection on the tautological line bundle over ${\mathbb{CP}}^N$, the Hopf bundle in the case $N=2$ that gives the projective space ${\mathbb{CP}}^1\cong S^2$.

The Aharonov-Anandan phase reduces to the Berry phase in the adiabatic limit (where the curve $C$ is the loop in parameter space pulled back through the eigenstate map). For non-adiabatic cyclic evolution, the Aharonov-Anandan phase is the more general geometric quantity — any closed curve in projective Hilbert space has a geometric phase, whether or not it corresponds to following an instantaneous eigenstate.

The Aharonov-Anandan phase has been measured in many experimental settings: nuclear magnetic resonance spin-echo experiments with cyclic radio-frequency pulses (Suter-Mueller-Pines 1988), neutron interferometry with arbitrary spin manipulation (Wagh-Rakhecha 1994), photon polarisation cyclic experiments (Tomita-Chiao 1986 saw a related geometric phase in twisted optical fibre), and most recently in superconducting qubit Mach-Zehnder interferometry where the geometric phase from a cyclic Hamiltonian drive is interfered with a reference dynamics (Leek et al. 2007 Science 318, 1889).

The Anandan-Aharonov 1990 Phys. Rev. Lett. 65, 1697 follow-up paper gave the Hilbert-space-metric interpretation of the geometric phase: the geometric phase is the area swept on ${\mathbb{CP}}^N$ in the Fubini-Study metric, divided by an appropriate factor. The Fubini-Study metric is the natural Riemannian metric on projective Hilbert space induced from the Hermitian inner product on the underlying Hilbert space, and the area form is the imaginary part of the Hermitian inner product on tangent vectors at each point.

Holonomic quantum computation: robust quantum gates

Zanardi and Rasetti 1999 Phys. Lett. A 264, 94 proposed holonomic quantum computation as a paradigm in which unitary gates on a quantum computer are implemented by traversing loops in the parameter space of a controllable Hamiltonian with a degenerate ground subspace. The gates are the Wilczek-Zee non-abelian holonomies of the loops; they depend only on the loops' geometric shapes and on the Berry connection, not on the protocol speeds.

The key advantage is intrinsic robustness to timing errors: a holonomic gate executed by a loop in parameter space yields the same unitary regardless of how the loop is parameterised in time, as long as the protocol remains adiabatic. Pulse-shape errors that affect the protocol speed do not affect the gate. This is in stark contrast to dynamical gates (the standard quantum-circuit paradigm) where the unitary depends on the integrated Rabi-frequency-times-pulse-duration, and any error in either the Rabi frequency or the pulse duration directly translates into a gate error.

Universal sets of holonomic gates have been theoretically constructed for a wide variety of platforms: trapped ions (Duan-Cirac-Zoller 2001 Science 292, 1695), nitrogen-vacancy centres in diamond (Solinas-Zanardi-Zanghi-Rossi 2003), superconducting qubits (Faoro-Siewert-Fazio 2003), and ultracold atoms in optical lattices (Pachos-Beige 2004). Experimental demonstrations have been carried out for trapped ions (Toyoda-Uchida-Noguchi-Haze-Urabe 2013, see Acta Physica Polonica), NV centres in diamond (Zhang-Wang-Tan-Pan-Liu 2014 Nature 514, 72), and superconducting qubits (Abdumalikov et al. 2013 Nature 496, 482).

The holonomic paradigm has been extended to non-adiabatic holonomic gates (Sjoqvist 2012, Xu-Tong-Long-Wang 2012) that achieve the same geometric character of the gates without requiring the slow adiabatic limit, by carefully designing the protocol so that the dynamical phase contribution cancels. This removes the slow-protocol-time bottleneck of the original adiabatic holonomic gates while preserving the timing-error robustness.

A related paradigm is decoherence-free subspaces and noiseless subsystems (Lidar-Chuang-Whaley 1998), where the degenerate subspace is also protected against environmental decoherence, providing both timing robustness and decoherence protection. The combination of holonomic gates and decoherence-free subspaces is one of the most promising routes to fault-tolerant quantum computation.

Synthesis

The Berry phase is the geometric refinement of adiabatic quantum dynamics that emerges naturally from the structure of the Hilbert space. Mathematically, it is the holonomy of a $U(1)$ connection on a complex line bundle over the parameter manifold of a Hamiltonian, with the connection being ${\mathbf{A}}_n=i⟨n|dn⟩$, the curvature being $F_n=d{\mathbf{A}}_n$, and the phase around a closed loop being the line integral of the connection or the surface integral of the curvature.

Physically, the Berry phase appears in every system whose Hamiltonian depends on parameters that can be slowly varied around a closed loop. The spin-1/2 in a rotating magnetic field gives the prototype half-solid-angle formula, with the spinor sign flip under a full rotation as the topological signature. The Aharonov-Bohm phase is the Berry phase of an electron transported around a flux tube. The molecular Aharonov-Bohm effect at conical intersections gives a $\pi$ Berry phase that modifies molecular spectra. The integer quantum Hall conductance is the Chern number of the Berry curvature integrated over the Brillouin zone, an integer topological invariant of the band structure. Topological insulators, topological semimetals, and topological superconductors are all classified by Berry-curvature-derived invariants. Holonomic quantum computation uses non-abelian Berry phases as intrinsically robust quantum gates.

The Berry phase unifies adiabatic dynamics with gauge theory and topology. It revealed that quantum mechanics has a geometric structure that had been hiding in plain sight for half a century, and the structure has turned out to be the unifying language for some of the most important developments in condensed-matter physics, quantum optics, quantum information, and molecular physics of the past forty years.

Full proof set Master

Proposition 1 (gauge invariance of the Berry phase on closed loops). Let $|n(\mathbf{R})⟩$ be a smooth family of normalised eigenstates of a Hamiltonian $H(\mathbf{R})$ on a smooth parameter manifold $\mathcal{M}$, with isolated non-degenerate eigenvalue $E_n(\mathbf{R})$. Let $C\subset \mathcal{M}$ be a smooth closed curve, and let $\alpha :\mathcal{M}\to \mathbb{R}$ be a smooth function on a neighbourhood of $C$ that is single-valued on $C$. Define the gauge-transformed eigenstate $|n^{\prime }(\mathbf{R})⟩=e^{i\alpha (\mathbf{R})}|n(\mathbf{R})⟩$. Then ${\oint }_C{\mathbf{A}}_n^{\prime }={\oint }_C{\mathbf{A}}_n$, where ${\mathbf{A}}_n^{\prime }=i⟨n^{\prime }|d{n}^{\prime }⟩$ and ${\mathbf{A}}_n=i⟨n|dn⟩$.

Proof. Compute ${\mathbf{A}}_n^{\prime }$ from the definition: $d|n^{\prime }⟩=e^{i\alpha }d|n⟩+i(d\alpha )e^{i\alpha }|n⟩$, so $⟨n^{\prime }|d{n}^{\prime }⟩=⟨n|dn⟩+id\alpha$. Therefore ${\mathbf{A}}_n^{\prime }=i⟨n^{\prime }|d{n}^{\prime }⟩=i⟨n|dn⟩-d\alpha ={\mathbf{A}}_n-d\alpha$.

The loop integral is ${\oint }_C{\mathbf{A}}_n^{\prime }={\oint }_C{\mathbf{A}}_n-{\oint }_{C}d\alpha$. By the fundamental theorem of calculus on the closed curve, ${\oint }_{C}d\alpha =\alpha (\mathbf{R}(T))-\alpha (\mathbf{R}(0))=0$ since $\alpha$ is single-valued on $C$ and $\mathbf{R}(T)=\mathbf{R}(0)$. Therefore ${\oint }_C{\mathbf{A}}_n^{\prime }={\oint }_C{\mathbf{A}}_n$. $\square$

Remark. If $\alpha$ is not single-valued on $C$ (i.e. $\alpha$ acquires a non-zero winding number around the loop), then ${\oint }_{C}d\alpha =2\pi n_w$ for some integer $n_w$ (the winding number). The Berry phase changes by $-2\pi n_w$, which is a multiple of $2\pi$, so the exponentiated Berry phase $e^{i{\gamma }_n(C)}=e^{i{\oint }_C{\mathbf{A}}_n}$ is genuinely gauge-invariant modulo any such redefinition. The phase itself is gauge-invariant only modulo $2\pi$; the exponentiated phase is strictly gauge-invariant.

Proposition 2 (gauge invariance of the Berry curvature). Under the gauge transformation ${\mathbf{A}}_n\to {\mathbf{A}}_n-d\alpha$, the Berry curvature $F_n=d{\mathbf{A}}_n$ is unchanged.

Proof. $F_n^{\prime }=d{\mathbf{A}}_n^{\prime }=d({\mathbf{A}}_n-d\alpha )=d{\mathbf{A}}_n-d^2\alpha =d{\mathbf{A}}_n=F_n$, using the exterior-derivative identity $d^2=0$. $\square$

Proposition 3 (Stokes connection between Berry phase and Berry curvature). Let $C$ be a smooth closed oriented curve in $\mathcal{M}$ bounding an oriented surface $S\subset \mathcal{M}$. Then ${\oint }_C{\mathbf{A}}_n={\int }_S{F}_n$.

Proof. This is Stokes's theorem applied to the smooth 1-form ${\mathbf{A}}_n$ on the manifold $\mathcal{M}$, with the boundary $\partial S=C$ inheriting the orientation from $S$. The integral identity is ${\oint }_{\partial S}\omega ={\int }_{S}d\omega$ for any smooth 1-form $\omega$ and 2-chain $S$ with smooth boundary. Applied to $\omega ={\mathbf{A}}_n$ with $d\omega =F_n$, this gives ${\oint }_C{\mathbf{A}}_n={\int }_S{F}_n$. $\square$

Proposition 4 (the gap-resolved Berry curvature formula). For a smooth family of non-degenerate eigenstates $|n(\mathbf{R})⟩$ of $H(\mathbf{R})$ on a parameter manifold $\mathcal{M}$, with all other eigenstates $|m(\mathbf{R})⟩,m\ne n$ also smooth and the eigenvalue $E_n$ isolated by a uniform gap, the Berry curvature has the form

$$F_{n,\mu \nu }(\mathbf{R})=i\sum _{m\ne n}\frac{⟨n|{\partial }_{\mu }H|m⟩⟨m|{\partial }_{\nu }H|n⟩-⟨n|{\partial }_{\nu }H|m⟩⟨m|{\partial }_{\mu }H|n⟩}{(E_n-E_m{)}^2}.$$

Proof. Worked through in Exercise 4. The key step is differentiating $H|n⟩=E_n|n⟩$ to get $⟨m|{\partial }_{\mu }n⟩=⟨m|{\partial }_{\mu }H|n⟩/(E_n-E_m)$ for $m\ne n$, and substituting into the antisymmetric expression $F_{n,\mu \nu }=i[⟨{\partial }_{\mu }n|{\partial }_{\nu }n⟩-⟨{\partial }_{\nu }n|{\partial }_{\mu }n⟩]$ via a complete-set resolution. The symmetric in-state contribution cancels under antisymmetrisation. $\square$

Proposition 5 (integrality of the Chern number). Let $|n(\mathbf{R})⟩$ be a smooth family of normalised eigenstates of a Hamiltonian $H(\mathbf{R})$ on a closed (compact without boundary) oriented two-dimensional manifold $\mathcal{M}$. Then

$$C_1=\frac{1}{2\pi }{\int }_{\mathcal{M}}{F}_n\in \mathbb{Z}.$$

Proof sketch. Choose two open charts $U_1,U_2$ that cover $\mathcal{M}$, on each of which $|n(\mathbf{R})⟩$ can be defined as a smooth single-valued section. On the overlap $U_1\cap U_2$, the two sections differ by a smooth transition function $|n^{(1)}⟩=g_{12}(\mathbf{R})|n^{(2)}⟩$ with $|g_{12}|=1$, so $g_{12}=e^{i{\varphi }_{12}}$ for some smooth ${\varphi }_{12}:U_1\cap U_2\to \mathbb{R}$ (locally; globally on a multiply-connected overlap, ${\varphi }_{12}$ may not be single-valued). The connections on the two charts satisfy ${\mathbf{A}}_n^{(1)}={\mathbf{A}}_n^{(2)}-d{\varphi }_{12}$.

Apply Stokes's theorem on each chart separately and integrate over the boundary $\partial U_1=-\partial U_2$ (with the orientation flip). The interior surface integrals reassemble into the total integral ${\int }_{\mathcal{M}}{F}_n$ on the left-hand side and into ${\oint }_{\partial U_1}({\mathbf{A}}_n^{(2)}-{\mathbf{A}}_n^{(1)})={\oint }_{\partial U_1}d{\varphi }_{12}$ on the right-hand side. The line integral ${\oint }_{\partial U_1}d{\varphi }_{12}$ equals $2\pi$ times the winding number of the transition function $g_{12}$ around the boundary curve, which is an integer.

Therefore $(1/2\pi ){\int }_{\mathcal{M}}{F}_n=$ winding number $\in \mathbb{Z}$. This is the integer Chern number of the complex line bundle defined by the eigenstate family $|n(\mathbf{R})⟩$. The argument generalises with appropriate technical care to higher genus and to higher-dimensional manifolds (where the higher Chern numbers and the Chern-Simons forms enter). $\square$

Proposition 6 (Wilczek-Zee gauge transformation law for non-abelian connection). For a $d$-fold degenerate eigenspace with smooth basis $\{|n_a(\mathbf{R})⟩{\}}_{a=1}^d$ and matrix-valued connection $A_{ab,\mu }=i⟨n_a|{\partial }_{\mu }{n}_b⟩$, under the smooth gauge transformation $|n_a⟩\to {\sum }_b{U}_{ba}(\mathbf{R})|n_b⟩$ with $U(\mathbf{R})\in U(d)$, the connection transforms as $A_{\mu }\to U^{†}{A}_{\mu }U+i{U}^{†}{\partial }_{\mu }U$.

Proof. Worked through in Exercise 9. The key step is expanding the product-rule derivative ${\partial }_{\mu }(U_{db}|n_d⟩)=({\partial }_{\mu }{U}_{db})|n_d⟩+U_{db}{\partial }_{\mu }|n_d⟩$ inside the inner product. The first piece gives $i{U}^{†}{\partial }_{\mu }U$; the second piece gives $U^{†}{A}_{\mu }U$. $\square$

The structural import of Propositions 1-6 is that the Berry-phase apparatus is a complete differential-geometric package: the Berry connection is a $U(1)$ connection on a complex line bundle, its curvature is gauge-invariant, the loop integral is the holonomy, and the integral over a closed surface is an integer Chern number. The Wilczek-Zee non-abelian generalisation upgrades the bundle to rank $d$ and the connection to $U(d)$-valued, with the same conceptual structure but matrix-valued objects throughout. All of these structures arise naturally from quantum-mechanical adiabatic dynamics and from the geometry of the projective Hilbert space — they are not imposed on quantum mechanics but emerge from it.

Connections Master

• Adiabatic theorem 12.07.07. The Berry phase is the geometric phase that emerges in the strict adiabatic limit. The adiabatic theorem identifies the spectral subspace as the slow degree of freedom and produces the $O(1/T)$ error bound for the slow protocol; the Berry phase is the leading geometric correction to the dynamical phase that survives in this limit. The Wilczek-Zee non-abelian generalisation is the corresponding extension for degenerate spectral subspaces, with the connection becoming matrix-valued.

• Angular momentum operators 12.05.01. The spin-1/2 Berry phase is the prototype example, with the half-solid-angle formula $\gamma =-\Omega /2$ being the topological signature of the $SU(2)$ spinor representation. The same calculation extends to higher spin $j$ with the Berry phase scaling as $-m\Omega$ for the $|j,m⟩$ eigenstate of $\hat{\mathbf{n}}\cdot \mathbf{J}$. The spinor sign flip under a full rotation, $e^{-i\pi \hat{\mathbf{n}}\cdot \mathbf{\sigma }}=-I$, is the Berry-monopole holonomy.

• Aharonov-Bohm effect [10.06]. The first observed Berry phase, twenty-five years before Berry's general identification. The electron's wavefunction around a flux tube it never enters picks up a phase $e\Phi /\hslash$ from the loop integral of the electromagnetic vector potential, which in Berry's language is the connection. The Aharonov-Bohm effect made manifest that the gauge potential, not just the field, has direct physical consequences.

• Topological insulators and quantum Hall effect. The integer Hall conductance is the Chern number of the Berry curvature integrated over the Brillouin zone (TKNN 1982), and the ${\mathbb{Z}}_2$ invariants of time-reversal-symmetric topological insulators (Kane-Mele 2005, Bernevig-Hughes-Zhang 2006, Fu-Kane-Mele 2007) are refinements of the Berry-curvature integral. All band-topology invariants are derived from the Berry-phase apparatus.

• Holonomic and adiabatic quantum computation 12.17.11 pending. Both paradigms use the Berry-phase apparatus as the computational primitive. Adiabatic quantum computation (Farhi-Goldstone-Gutmann-Sipser 2000) uses the spectral gap and the adiabatic theorem; holonomic quantum computation (Zanardi-Rasetti 1999) uses the non-abelian Berry holonomy as gate primitive with intrinsic timing-error robustness. The two paradigms are deeply related and both equivalent in power to the gate-based circuit model.

• Differential geometry of fibre bundles [03]. The Berry connection is a $U(1)$ connection on a complex line bundle over parameter space, with the curvature as the field strength and the Chern number as the topological invariant. The Wilczek-Zee non-abelian Berry phase is a $U(d)$ connection on a rank-$d$ complex vector bundle. The mathematical framework is the same as for gauge theories of fundamental physics (electromagnetism is $U(1)$, weak interactions are $SU(2)$, strong interactions are $SU(3)$, all on principal bundles over spacetime). The Berry phase and the gauge field of fundamental physics are two physical realisations of the same differential-geometric structure.

• Conical intersections and molecular dynamics [12.09]. At a conical intersection of two electronic Born-Oppenheimer levels, the adiabatic real wavefunction acquires a $\pi$ Berry phase around any loop encircling the intersection. The Mead-Truhlar molecular Aharonov-Bohm effect (1979) makes this phase observable in molecular spectra and dynamics. Modern computational chemistry near conical intersections must include the geometric phase contribution.

• Optics and the Pancharatnam phase [Pancharatnam 1956]. Berry's spin-1/2 half-solid-angle formula was anticipated by thirty years in Pancharatnam's analysis of polarised-light interference on the Poincare sphere. The Pancharatnam-Berry phase is the basis of modern polarisation-based optical metasurfaces, and the optical implementation of the geometric phase has independent experimental traditions in twisted optical fibres (Tomita-Chiao 1986), polarisation rotators (Bhandari-Samuel 1988), and structured-light vortex beams.

• Statistical mechanics and quantum thermodynamics. Berry-phase contributions to the partition function and to non-equilibrium response functions appear in slow-driven quantum thermodynamic protocols, with the geometric phase contributing to the work and the entropy generation. Geometric-phase generalisations of fluctuation theorems and of the Jarzynski equality have been developed (Sagawa-Ueda 2008, Sinitsyn-Nemenman 2007 on geometric pumping).

Historical & philosophical context Master

The discovery of the geometric phase is one of the most striking cases in twentieth-century physics where a fundamental structural feature of the formalism remained unnoticed for decades because everyone routinely absorbed it into arbitrary phase conventions. The Berry phase was implicit in the Born-Fock 1928 adiabatic theorem — the formula ${\gamma }_n^{\mathrm{geom}}=i\int ⟨n|dn/ds⟩ds$ appears in Born and Fock's derivation, but they did not single it out as a separate, observable contribution and treated it as a phase choice. For fifty-six years, every quantum mechanics textbook simply omitted the geometric phase, treating it as one of the many phases that drop out of physical predictions.

There were earlier hints of the geometric phase that were also unrecognised. Pancharatnam (1956) gave a complete analysis of the geometric phase for polarised-light interference, including the half-solid-angle formula for the Poincare-sphere triangle, but his paper was published in the Proceedings of the Indian Academy of Sciences and went largely unread by the physics mainstream. Longuet-Higgins (1958) showed that adiabatic real molecular wavefunctions must flip sign around a conical intersection, the Longuet-Higgins sign theorem, but framed it as a curious mathematical fact rather than a fundamental geometric phase. Mead and Truhlar (1979) recognised the sign as a real Berry-style phase with observable spectroscopic consequences — the molecular Aharonov-Bohm effect — but Mead's 1979 paper was titled the "molecular Aharonov-Bohm" because the connection to a general geometric-phase principle had not yet been articulated.

The Aharonov-Bohm effect itself (1959) was, in retrospect, the first observed Berry phase, but for twenty-five years it was framed as a peculiarity of the electromagnetic gauge potential rather than as an instance of a general geometric structure. The connection was made explicit only when Berry's 1984 paper showed that the gauge structure of the Berry phase and the gauge structure of electromagnetism are two physical realisations of the same mathematical object — a $U(1)$ connection on a principal bundle.

Berry's 1984 paper Quantal phase factors accompanying adiabatic changes in the Proceedings of the Royal Society A was the catalyst. Berry's central contributions were three: (1) he identified the geometric phase ${\gamma }_n^{\mathrm{geom}}$ as a gauge-invariant observable quantity that survives the strict adiabatic limit, not a phase convention to be absorbed; (2) he derived the gap-resolved Berry-curvature formula and demonstrated that the curvature has monopole-like singularities at level degeneracies; (3) he gave the spin-1/2-in-rotating-field worked example with the half-solid-angle formula, which made the geometric character of the phase visible and concrete. Berry's paper was rapidly recognised as a major reframing of adiabatic quantum dynamics, and the field of geometric-phase research exploded.

Simon 1983 Holonomy, the quantum adiabatic theorem, and Berry's phase in Physical Review Letters — published one year before Berry's paper, in response to a preprint that Simon saw early — gave the bundle-theoretic reformulation. Simon recognised that the Berry phase is the holonomy of the natural Hermitian connection on the spectral line bundle, and that Berry's calculation is the prototype example of a much more general construction. Simon's paper provided the differential-geometric language that has become standard.

Wilczek and Zee 1984 Appearance of gauge structure in simple dynamical systems in Physical Review Letters generalised Berry's $U(1)$ connection to degenerate eigenspaces, with the matrix-valued non-abelian connection $A_{ab,\mu }$, its $U(d)$ gauge transformation, and the path-ordered exponential holonomy. The Wilczek-Zee paper recognised that the structural language of fibre bundles, connections, and gauge transformations — long familiar to mathematical physics in the context of Yang-Mills theory — applies equally to the geometry of quantum adiabatic dynamics. This was the bridge that connected adiabatic quantum mechanics to the gauge theories of high-energy physics.

Aharonov-Anandan 1987 Phase change during a cyclic quantum evolution in Physical Review Letters removed the adiabatic-evolution restriction. Aharonov and Anandan showed that any cyclic quantum evolution has a geometric phase associated with the closed curve in projective Hilbert space, with the integrand being the natural Hermitian connection of the tautological line bundle. The Aharonov-Anandan phase reduces to the Berry phase in the adiabatic limit, but the more general framework is the projective-Hilbert-space picture.

Mead-Truhlar molecular Aharonov-Bohm (1979). Mead and Truhlar identified the topological geometric phase contribution to nuclear dynamics in molecules near conical intersections, predicting a $\pi$ Berry phase around any loop encircling a conical intersection. The prediction has been experimentally confirmed by sensitive spectroscopic measurements of molecules with low-lying conical intersections (the H + H2 system, the methyl cation, the cyclopentadienyl radical) and is now a standard correction in molecular quantum dynamics calculations.

TKNN 1982 and the integer quantum Hall effect. Thouless, Kohmoto, Nightingale, and den Nijs identified the integer Hall conductance with the integer Chern number of the Bloch line bundle over the Brillouin zone, launching the modern field of topological band theory. Thouless received the 2016 Nobel Prize in Physics for this and related work on topological phase transitions.

Topological insulators (2005-2007). Kane-Mele 2005, Bernevig-Hughes-Zhang 2006, and Fu-Kane-Mele 2007 introduced the ${\mathbb{Z}}_2$ topological invariants for time-reversal-symmetric band insulators, generalising the Chern-number classification to include the topological insulators. The HgTe/CdTe experimental confirmation by Konig et al. 2007 launched the modern era of topological materials.

Periodic table of topological phases (2008-2009). Schnyder-Ryu-Furusaki-Ludwig 2008 and Kitaev 2009 organised the topological invariants of free-fermion systems into the periodic table of topological phases, indexed by Altland-Zirnbauer symmetry class and spatial dimension, with the entries determined by the K-theory of the appropriate classifying spaces. The Berry phase is the underlying geometric quantity from which all of these invariants are derived.

Holonomic quantum computation (Zanardi-Rasetti 1999). Zanardi and Rasetti proposed using the Wilczek-Zee non-abelian Berry holonomy as the primitive for quantum-gate implementation, with intrinsic timing-error robustness as the chief advantage. Experimental demonstrations in trapped ions, NV centres in diamond, and superconducting qubits have followed, with the non-adiabatic variants of Sjoqvist 2012 and Xu-Tong-Long-Wang 2012 removing the slow-protocol-time bottleneck.

The philosophical content of the Berry phase is the recognition that quantum mechanics has a rich geometric structure that had been hiding in plain sight. The geometric phase emerges from the structural features of the Hilbert space — the eigenstate is defined only up to a phase, the parameter manifold of the Hamiltonian carries a natural connection, the holonomy of the connection is observable — and these features are present in every quantum system. The Berry phase is not an exotic correction added to a standard theory; it is a fundamental geometric feature of quantum mechanics that we had been ignoring.

The broader lesson is that gauge structure is everywhere in quantum mechanics, not just in the formal description of fundamental forces. Every quantum system with a controllable Hamiltonian and a parameter manifold carries a Berry connection on its spectral line bundle, with a gauge-invariant curvature and a topologically classified Chern-number structure. This unification of the gauge structure of fundamental physics with the gauge structure of adiabatic quantum dynamics is one of the most important conceptual unifications of late-twentieth-century theoretical physics, with reverberations that continue to drive research in topological matter, quantum information, molecular physics, and quantum optics.

Bibliography Master

Primary literature and historical sources:

@article{Pancharatnam1956,
  author = {Pancharatnam, S.},
  title = {Generalised theory of interference and its applications},
  journal = {Proceedings of the Indian Academy of Sciences A},
  volume = {44},
  year = {1956},
  pages = {247--262},
}

@article{AharonovBohm1959,
  author = {Aharonov, Y. and Bohm, D.},
  title = {Significance of electromagnetic potentials in quantum theory},
  journal = {Physical Review},
  volume = {115},
  year = {1959},
  pages = {485--491},
}

@article{LonguetHiggins1958,
  author = {Longuet-Higgins, H. C.},
  title = {Studies of the {Jahn-Teller} effect. {II}. The dynamical problem},
  journal = {Proceedings of the Royal Society A},
  volume = {244},
  year = {1958},
  pages = {1--16},
}

@article{MeadTruhlar1979,
  author = {Mead, C. A. and Truhlar, D. G.},
  title = {On the determination of {Born-Oppenheimer} nuclear motion wave functions including complications due to conical intersections and identical nuclei},
  journal = {Journal of Chemical Physics},
  volume = {70},
  year = {1979},
  pages = {2284--2296},
}

@article{Berry1984,
  author = {Berry, M. V.},
  title = {Quantal phase factors accompanying adiabatic changes},
  journal = {Proceedings of the Royal Society A},
  volume = {392},
  year = {1984},
  pages = {45--57},
}

@article{Simon1983,
  author = {Simon, B.},
  title = {Holonomy, the quantum adiabatic theorem, and {Berry's} phase},
  journal = {Physical Review Letters},
  volume = {51},
  year = {1983},
  pages = {2167--2170},
}

@article{WilczekZee1984,
  author = {Wilczek, F. and Zee, A.},
  title = {Appearance of gauge structure in simple dynamical systems},
  journal = {Physical Review Letters},
  volume = {52},
  year = {1984},
  pages = {2111--2114},
}

@article{TKNN1982,
  author = {Thouless, D. J. and Kohmoto, M. and Nightingale, M. P. and den Nijs, M.},
  title = {Quantized {Hall} conductance in a two-dimensional periodic potential},
  journal = {Physical Review Letters},
  volume = {49},
  year = {1982},
  pages = {405--408},
}

@article{AharonovAnandan1987,
  author = {Aharonov, Y. and Anandan, J.},
  title = {Phase change during a cyclic quantum evolution},
  journal = {Physical Review Letters},
  volume = {58},
  year = {1987},
  pages = {1593--1596},
}

@article{ZanardiRasetti1999,
  author = {Zanardi, P. and Rasetti, M.},
  title = {Holonomic quantum computation},
  journal = {Physics Letters A},
  volume = {264},
  year = {1999},
  pages = {94--99},
}

@article{KarplusLuttinger1954,
  author = {Karplus, R. and Luttinger, J. M.},
  title = {{Hall} effect in ferromagnetics},
  journal = {Physical Review},
  volume = {95},
  year = {1954},
  pages = {1154--1160},
}

@article{KaneMele2005,
  author = {Kane, C. L. and Mele, E. J.},
  title = {{$\mathbb{Z}_2$} topological order and the quantum spin {Hall} effect},
  journal = {Physical Review Letters},
  volume = {95},
  year = {2005},
  pages = {146802},
}

@article{BernevigHughesZhang2006,
  author = {Bernevig, B. A. and Hughes, T. L. and Zhang, S.-C.},
  title = {Quantum spin {Hall} effect and topological phase transition in {HgTe} quantum wells},
  journal = {Science},
  volume = {314},
  year = {2006},
  pages = {1757--1761},
}

@article{FuKaneMele2007,
  author = {Fu, L. and Kane, C. L. and Mele, E. J.},
  title = {Topological insulators in three dimensions},
  journal = {Physical Review Letters},
  volume = {98},
  year = {2007},
  pages = {106803},
}

@article{SundaramNiu1999,
  author = {Sundaram, G. and Niu, Q.},
  title = {Wave-packet dynamics in slowly perturbed crystals: gradient corrections and {Berry-phase} effects},
  journal = {Physical Review B},
  volume = {59},
  year = {1999},
  pages = {14915--14925},
}

@article{Berry1987,
  author = {Berry, M. V.},
  title = {The adiabatic phase and {Pancharatnam's} phase for polarized light},
  journal = {Journal of Modern Optics},
  volume = {34},
  year = {1987},
  pages = {1401--1407},
}

@article{TomitaChiao1986,
  author = {Tomita, A. and Chiao, R. Y.},
  title = {Observation of {Berry's} topological phase by use of an optical fiber},
  journal = {Physical Review Letters},
  volume = {57},
  year = {1986},
  pages = {937--940},
}

Textbooks and reprint volumes:

@book{ShapereWilczek1989,
  editor = {Shapere, A. and Wilczek, F.},
  title = {Geometric Phases in Physics},
  publisher = {World Scientific},
  year = {1989},
}

@book{Griffiths2018,
  author = {Griffiths, D. J. and Schroeter, D. F.},
  title = {Introduction to Quantum Mechanics},
  edition = {3},
  publisher = {Cambridge University Press},
  year = {2018},
}

@book{SakuraiNapolitano2017,
  author = {Sakurai, J. J. and Napolitano, J.},
  title = {Modern Quantum Mechanics},
  edition = {2},
  publisher = {Cambridge University Press},
  year = {2017},
}

@book{Bohm2003,
  author = {Bohm, A. and Mostafazadeh, A. and Koizumi, H. and Niu, Q. and Zwanziger, J.},
  title = {The Geometric Phase in Quantum Systems},
  publisher = {Springer},
  year = {2003},
}

@book{Nakahara2003,
  author = {Nakahara, M.},
  title = {Geometry, Topology, and Physics},
  edition = {2},
  publisher = {Institute of Physics Publishing},
  year = {2003},
}

@book{BernevigHughes2013,
  author = {Bernevig, B. A. and Hughes, T. L.},
  title = {Topological Insulators and Topological Superconductors},
  publisher = {Princeton University Press},
  year = {2013},
}

Mathematical foundations and rigorous treatments:

@article{Kohmoto1985,
  author = {Kohmoto, M.},
  title = {Topological invariant and the quantization of the {Hall} conductance},
  journal = {Annals of Physics},
  volume = {160},
  year = {1985},
  pages = {343--354},
}

@article{Schnyder2008,
  author = {Schnyder, A. P. and Ryu, S. and Furusaki, A. and Ludwig, A. W. W.},
  title = {Classification of topological insulators and superconductors in three spatial dimensions},
  journal = {Physical Review B},
  volume = {78},
  year = {2008},
  pages = {195125},
}

@article{Kitaev2009,
  author = {Kitaev, A.},
  title = {Periodic table for topological insulators and superconductors},
  journal = {AIP Conference Proceedings},
  volume = {1134},
  year = {2009},
  pages = {22--30},
}

@article{HasanKane2010,
  author = {Hasan, M. Z. and Kane, C. L.},
  title = {Colloquium: topological insulators},
  journal = {Reviews of Modern Physics},
  volume = {82},
  year = {2010},
  pages = {3045--3067},
}

@article{Xiao2010,
  author = {Xiao, D. and Chang, M.-C. and Niu, Q.},
  title = {{Berry} phase effects on electronic properties},
  journal = {Reviews of Modern Physics},
  volume = {82},
  year = {2010},
  pages = {1959--2007},
}