Can the Berry Connection be derived from a metric?

Question

The Berry Connection is

$$A_\mu(R)=-i \langle \Psi(R) |\partial_\mu \Psi(R) \rangle$$ which allows us to parallel transport a state indexed by $R$. We can integrate the Berry Connection to get the Berry Phase, and we can differentiate the Berry Connection to get the Berry Curvature.

Can the Berry Connection be derived from a metric? As a prototypical example, I'm thinking about how the Christoffel Symbols in General Relativity (GR) can be derived from the metric tensor. Also, I think for every connection, there exists a metric for which the connection is a Levi-Civita connection. However, is there a natural and physical metric that induces the Berry Connection?

In this presentation, the great and powerful Haldane relates "quantum distance" to the Berry Curvature, but it doesn't look like you can derive the Berry Curvature from the quantum distance in the same way curvature and the metric are related in GR.

This post imported from StackExchange Physics at 2015-11-01 18:09 (UTC), posted by SE-user ChickenGod

Comments

Also, I think for every connection, there exists a metric for which the connection is a Levi-Civita connection. This is not the case; see eg mathoverflow.net/questions/54434/…

This post imported from StackExchange Physics at 2015-11-01 18:09 (UTC), posted by SE-user Christoph

---

Ok cool thanks. So if that's not true, my question still holds: Does a metric exist such that the Berry Connection is a Levi-Civita connection?

This post imported from StackExchange Physics at 2015-11-01 18:09 (UTC), posted by SE-user ChickenGod

---

@Christoph Thanks for a fantastic link. I thought I'd read most of Bill Thurston's answers on MO: Bill Thurston RIP and thanks for the truly dazzling work that the rest of us love to behold.

This post imported from StackExchange Physics at 2015-11-01 18:09 (UTC), posted by SE-user WetSavannaAnimal aka Rod Vance

Answer 1

The answer is positive, except that the Berry connection being an Abelian connection, and the corresponding metric is not a metric on the tangent bundle as in the Riemannian case, but rather a metric on a line bundle , i.e., a one dimensional metric.

This line bundle was defined in In Barry Simon's seminal work , where he proved that the Berry phase is the holonomy of a (connection of) the Hermitian line bundle given by: $\{R, |\Psi (R)\rangle \} \in (\mathcal{M}, C^\times )$ subject to the constraint:

$$H(R) |\Psi(R) \rangle = E(R) |\Psi (R)\rangle$$

Where $\mathcal{M}$ is the parameter space of the Hamiltonian $H(R)$. The line bundle is aligned at every point of $\mathcal{M}$ along the eigenvector $|\Psi (R)\rangle$ of the Schroedinger equation. This bundle possesses a metric on the space of sections which allows computing scalar products between two sections $x$ and $y$. ($x$ and $y$ are locally nonvanishing complex functions on $\mathcal{M}$):

$$(x,y)(R) = \bar{x}(R) e^{-\langle \Psi(R) |\Psi (R)\rangle } y(R)$$ This scalar product is invariant in the transition between patches of the manifold $\mathcal{M}$.

Now it is easy to show that the Berry connection is compatible with this metric, (just like the Levi-Civita connection is compatible with the Riemannian metric:

$$\partial_{\mu} (x,y) = (D_{\mu} x,y) + (x, D_{\mu} y)$$ Where: $D_{\mu}$ is the covariant derivative corresponding to the Berry connection $$D_{\mu} = \partial_{\mu}+iA_{\mu}$$

This post imported from StackExchange Physics at 2015-11-01 18:09 (UTC), posted by SE-user David Bar Moshe