Berry Phase and Curvature in Adiabatic Evolution

Welcome back to our module on Geometric and Topological Methods. In the last lesson, we formally introduced the quantum adiabatic theorem and established the conditions for its validity. We saw that for a system evolving slowly and without level crossings, its state tracks the instantaneous eigenstate of the Hamiltonian, but with two important phase factors: a dynamical phase and a mysterious "geometric" phase, which we wrote as:

γ_{n} (t) = i \int_{0}^{t} ⟨ ψ_{n} (t^{'}) ∣ \frac{d}{d t ^{'}} ∣ ψ_{n} (t^{'})⟩ d t^{'}

Today, we will unravel this term. You will learn that this is the celebrated Berry phase, a profound concept that links quantum evolution to the geometry of a system's parameter space. Our goal is to compute this phase and understand its physical interpretation through its local counterpart, the Berry curvature. This lesson will directly address your stated discomfort with line integrals by showing how they arise naturally when we re-express this time-dependent phase in a more geometric language.

From Time to Geometry: The Berry Connection

The expression for $γ_{n} (t)$ appears to depend on the duration of the evolution, $t$ . However, Michael Berry's crucial insight was that this dependence is illusory. The phase is purely geometric. The key is to recognize that the instantaneous eigenstate $∣ n (t)⟩$ changes with time only because the parameters of the Hamiltonian, which we can denote by a vector $R (t)$ , are changing. Applying the chain rule allows us to switch from a derivative with respect to time to a derivative with respect to the parameters:

\frac{d}{d t} ∣ n (R (t))⟩ = (\frac{d R}{d t} \cdot \nabla_{R}) ∣ n (R (t))⟩

Here, $\nabla_{R}$ is the gradient operator in the parameter space. Substituting this into the integral for the Berry phase gives:

γ_{n} (t) = \int_{0}^{t} i ⟨ n (R (t^{'})) ∣ \nabla_{R} ∣ n (R (t^{'}))⟩ \cdot \frac{d R}{d t ^{'}} d t^{'}

This expression has the form of a line integral. The term $(d R / d t^{'}) d t^{'}$ represents an infinitesimal displacement $d R$ along the path $C$ that the system traces in its parameter space. The integral is therefore over the path itself, not explicitly over time:

γ_{n} (C) = \int_{C} i ⟨ n (R) ∣ \nabla_{R} ∣ n (R)⟩ \cdot d R

The integral now clearly depends only on the path $C$ , not on how fast or slow it was traversed (as long as it was slow enough for the adiabatic approximation to hold).

This structure strongly suggests an analogy with electromagnetism. We can define a vector field in the parameter space, called the Berry connection or Berry potential:

A_{n} (R) = i ⟨ n (R) ∣ \nabla_{R} ∣ n (R)⟩

The Berry phase is then simply the line integral of the Berry connection along the path in parameter space:

γ_{n} (C) = \int_{C} A_{n} (R) \cdot d R

This is formally identical to how the magnetic part of the Aharonov-Bohm phase is calculated from the magnetic vector potential $A$ .

The following video provides a concise and visually clear derivation of the Berry phase and introduces the concept of the Berry connection.

Deriving Berry phase, Berry connection & Berry curvature

This video, "Deriving Berry phase, Berry connection & Berry curvature" from the channel Professor Nano, walks through the derivation we just outlined. It shows how the time-dependent Schrödinger equation leads to an extra phase term, which is then transformed into a path-independent geometric phase.

Watch the segment from the derivation. Pay close attention to how the initial guess for the wavefunction is shown to be incomplete, necessitating the introduction of the Berry phase \gamma. The key moment is the transition from a time integral to a line integral over the parameter space \mathbf{R}.

Gauge Invariance: When is the Berry Phase Physical?

There's a subtlety we must address. The instantaneous eigenstates $∣ n (R)⟩$ are determined by the Schrödinger equation only up to an arbitrary phase factor. We are free to redefine our eigenstates at every point in parameter space with a local phase transformation, known as a gauge transformation:

∣ n^{'} (R)⟩ = e^{i ζ (R)} ∣ n (R)⟩

where $ζ (R)$ is a real, smooth function. How does this affect the Berry connection and the Berry phase? A straightforward calculation shows that the new Berry connection, $A_{n}^{'}$ , is related to the old one by:

A_{n}^{'} (R) = A_{n} (R) - \nabla_{R} ζ (R)

This is precisely how the magnetic vector potential transforms under a gauge transformation in electromagnetism. Consequently, the Berry phase for an arbitrary path $C$ from $R_{i}$ to $R_{f}$ is not gauge-invariant:

γ_{n}^{'} (C) = γ_{n} (C) - [ζ (R_{f}) - ζ (R_{i})]

Since we can choose $ζ (R)$ arbitrarily, this means the Berry phase for an open path has no intrinsic physical meaning; it can be changed at will by a different choice of phase convention for the wavefunctions.

The situation changes dramatically if the path is a closed loop, where $R_{f} = R_{i}$ . In this case, for the wavefunction to be single-valued, the phase factor $e^{i ζ (R)}$ must also be single-valued, which implies that $ζ (R_{f}) - ζ (R_{i}) = 2 π m$ for some integer $m$ . This means that for a closed loop, the Berry phase is gauge-invariant up to integer multiples of $2 π$ . Since a phase is only physically meaningful modulo $2 π$ , we arrive at a profound conclusion: the Berry phase accumulated over a closed loop is a physical observable.

The following excerpt provides a detailed derivation of this crucial property.

L17.2 Berry's phase and Berry's connection

This lecture from MIT OpenCourseWare, "Berry's phase and Berry's connection," provides a careful and explicit derivation of how the Berry connection and Berry phase transform under a gauge change.

Please watch the section from the gauge transformation. Follow the calculation that shows how \mathbf{A}_n transforms and how this leads to the non-invariance for an open path. The key takeaway is at the end, where the lecturer concludes that for a closed path, the extra term vanishes and the phase becomes a physically meaningful, gauge-invariant quantity.

Berry Curvature: The Field Behind the Phase

The analogy with electromagnetism goes deeper. Just as the physically observable magnetic field $B$ can be derived from the gauge-dependent vector potential $A$ via $B = \nabla \times A$ , we can define a local, gauge-invariant quantity from the Berry connection. This is the Berry curvature, $Ω_{n}$ , defined as the curl of the Berry connection in parameter space:

Ω_{n} (R) = \nabla_{R} \times A_{n} (R)

Since the curl of a gradient is always zero ( $\nabla \times (\nabla ζ) = 0$ ), the Berry curvature is manifestly gauge-invariant. It is a local property of the parameter space at point $R$ , reflecting the intrinsic geometry of the manifold of eigenstates.

Using Stokes' theorem, we can now express the Berry phase for a closed loop $C$ as the flux of the Berry curvature through any surface $S$ that is bounded by the loop:

γ_{n} (C) = \oint_{C} A_{n} \cdot d R = \iint_{S} Ω_{n} \cdot d S

This provides a powerful geometric interpretation: the Berry phase is the total "flux" of Berry curvature enclosed by the adiabatic loop.

The following resource expands on this powerful analogy.

[PDF] Berry Phase Effects on Electronic Properties - NTNU Physics

This review article provides a concise summary of the Berry curvature and its properties.

Please read the subsection titled Berry curvature. It formally defines the Berry curvature tensor \Omega_{n \mu\nu} and shows how Stokes' theorem connects it to the Berry phase. Critically, it also provides an alternative formula for the curvature: \Omega_{n, \mu \nu}(\mathbf{R}) = i \sum_{n' \neq n} \frac{\langle n | \frac{\partial H}{\partial R_\mu} | n' \rangle \langle n' | \frac{\partial H}{\partial R_\nu} | n \rangle - (\mu \leftrightarrow \nu)}{(\varepsilon_n(\mathbf{R}) - \varepsilon_{n'}(\mathbf{R}))^2} This formula is extremely useful, both for practical calculations and for physical insight. It shows that the curvature at a point is generated by the "virtual" transitions to other energy levels, weighted by the energy denominators. You can see immediately that the curvature will be large where energy levels get close to each other.

A Canonical Example: Spin in a Magnetic Field

Let's solidify these abstract concepts with the most important example: a spin-1/2 particle in a slowly rotating magnetic field $B (t)$ . The Hamiltonian is $H (t) = - γ B (t) \cdot S$ . The parameter space here is the space of possible field directions, which can be represented by points on a sphere. Let's say the field traces a closed loop on this sphere. What is the Berry phase acquired by the ground state (spin-up along $B$ )?

3 Berry phases and curvatures

This textbook chapter by David Vanderbilt provides a wonderfully clear calculation for this exact problem.

First, read the section An example. This sets up the problem of a spin in a magnetic field and computes the phase for a specific triangular path. Next, read the section Berry curvature and Chern theorem. This is the core of the example. It calculates the Berry curvature for the spinor ground state and finds the remarkable result that it is uniform over the sphere of directions. It then uses Stokes' theorem to conclude that the Berry phase for any loop is simply proportional to the solid angle enclosed by that loop on the sphere.

The result from this example is a cornerstone of the field:

γ_{spin-up} = - \frac{1}{2} \times (Solid Angle enclosed by the loop)

If the magnetic field is adiabatically rotated through a full $2 π$ rotation and returns to its start (e.g., along the equator), the solid angle enclosed is $2 π$ . The Berry phase is thus $- π$ . This means the final state is $e^{- iπ} ∣ ψ (0)⟩ = - ∣ ψ (0)⟩$ . The wavefunction of a spin-1/2 particle flips its sign after a $2 π$ rotation, a purely geometric and observable quantum effect! The source of this curvature is a "magnetic monopole" in parameter space, located at the degeneracy point $B = 0$ .

Berryology in Condensed Matter

Your own research interests lie in condensed matter physics, which is arguably the field where Berry's ideas have had the most profound impact. In crystalline solids, the electron momentum $k$ acts as a parameter for the Bloch Hamiltonian $H (k)$ . The parameter space is the Brillouin zone. We can therefore define a Berry connection $A_{n} (k)$ and Berry curvature $Ω_{n} (k)$ for each energy band $n$ .

This seemingly abstract concept has concrete physical consequences. For instance, the Berry curvature acts as an effective magnetic field in momentum space, deflecting electrons and giving rise to the anomalous Hall effect in ferromagnets.

3 Berry phases and curvatures

The same chapter provides an introduction to applying these concepts to the Brillouin zone.

Read the section Berryology of the Brillouin zone. This will show you how to map the concepts of Berry phase, connection, and curvature onto the Bloch states of a crystal. Pay attention to the important distinction of using the cell-periodic part of the Bloch function, |u_{n\mathbf{k}}\rangle, to define these quantities.

Conclusion

In this lesson, we have demystified the geometric phase by giving it a name, a formula, and a physical interpretation. We have seen how the abstract geometry of quantum state space can have direct, observable consequences.

Key Takeaways:

The Berry phase is a geometric phase acquired by a quantum system during a cyclic adiabatic evolution. It is calculated as the line integral of the Berry connection $A_{n} (R) = i ⟨ n (R) ∣ \nabla_{R} ∣ n (R)⟩$ around a closed path in parameter space.
While the Berry connection is gauge-dependent, the Berry phase for a closed loop is gauge-invariant (mod $2 π$ ) and is a physical observable.
The Berry curvature, $Ω_{n} = \nabla_{R} \times A_{n}$ , is a local, gauge-invariant quantity analogous to a magnetic field in parameter space. The Berry phase is the flux of the Berry curvature through the loop.
Degeneracy points in the parameter space act as sources or "monopoles" of Berry curvature.
In condensed matter physics, the Brillouin zone serves as the parameter space, and the Berry curvature of the Bloch bands governs a variety of electronic transport phenomena.

In our next lesson, we will build upon these ideas to distinguish between properties that are merely geometric and those that are topological. We will see that integrating the Berry curvature over a closed manifold (like the entire Brillouin zone) can yield quantized integers, known as topological invariants, which lead to extraordinarily robust physical phenomena.

Recommended resources