Let's have simple scalar $\Phi$ action involves spontaneously symmetry breaking in a form $$ \tag 1 S = \int d^{4}x\left( |\partial_{\mu}\psi|^{2} + \psi^{2}|\partial_{\mu}\theta |^{2} - \frac{\lambda}{4} (\psi^{2} - v^{2})^{2}\right) $$ It is usual higgs-like action where complex scalar field is represented in a form $\Phi = \psi e^{i\theta}$. E.o.m. generated by $(1)$ implies string-like solutions $$ \tag 2 \Phi = vf(r)e^{in\varphi}, \quad f(0) = 0, \quad f(\infty ) = 1 $$ ($r, \varphi$ denote polar coordinates). From substituting $(2)$ in $(1)$ and calculating of stress-energy tensor follows that strings have nonzero tension, thus they have to radiate.

One says (see, for example, this article, p.5, section "Dual representation...") that by using duality relation $$ \tag 3 \psi^{2}\partial_{\mu}\theta = \frac{1}{2}f_{a}\epsilon_{\mu \nu \alpha \beta}\partial^{\nu}B^{\lambda \rho} $$ one provides analytical description of radiation of $\theta $ bosons by string-like solutions $(2)$.

The question: I completely don't understand how action generated by $(3)$ (it is called Kalb-Ramond action) describes radiation of string-like solution $(3)$. Can you explain this?


This is a standard particle-vortex duality transformation. The idea is there exists a "dual" description of a superfluid, in which the bosons (to be precise, the Goldstone mode) is described by a 2-form(tensor) gauge field, and the vortices are now string-like "matter" charged under the gauge field. The effective action for phase fluctuations (i.e. Goldstone mode) is just

$ L=\frac{1}{2}\rho^2 (\partial_\mu\theta)^2$

Here $\rho$ is the superfluid density, roughly your $f$ at infinity. $\theta$ can be divided into two parts: first the smooth fluctuations, and the singular ones (vortices). We will write $\theta=\eta + \theta_v$. $\eta$ satisfies $[\partial_x,\partial_y]\eta=0$.

The next thing is to rewrite the Lagrangian using Hubbard-Stratonovich transformation, by introducing an auxiliary vector field $\zeta_\mu$:

$L=\partial_\mu \theta \zeta^\mu -\frac{1}{\rho^2} \zeta^2=\partial_\mu \eta\zeta^\mu+\partial_\mu \theta_v\zeta^\mu -\frac{1}{\rho^2} \zeta^2$.

Since $\eta$ is a smooth function, the functional integration over $\eta$ can be easily performed, yielding a constraint $\partial_\mu \zeta^\mu=0$ which can be resolved by writing $\zeta^\mu=\varepsilon^{\mu\nu\lambda\rho}\partial_\nu B_{\lambda\rho}$. Notice that $B$ is a 2-form gauge field, since this expression has exactly the gauge redundancy. Then the Lagrangian can be casted into a gauge theory of 2-form tensor gauge field coupled to string-like sources (i.e. the $\theta_v$).

| cite | improve this answer | |
  • $\begingroup$ Thank you! If I understand correctly, you've introduced $\zeta$ field by formal transformation $$ \tag 1 \int D \rho D \theta e^{i\int d^{4}x\left[... + \frac{1}{2}\rho^{2}(\partial_{\mu}\theta )^{2}\right]} \sim \int D \theta D \rho D \zeta e^{i \int d^{4}x\left[... + \partial_{\mu}\theta \zeta^{\mu} - \frac{1}{\rho^{2}}\zeta^{2}\right]}? $$ If yes, how to integrate over $\eta$ here? Action $(1)$ depends on $\eta$ only linear, so corresponding path integral isn't gaussian. $\endgroup$ – Name YYY Jul 1 '15 at 12:52
  • 1
    $\begingroup$ Integrating over some linear action is even easier than Gaussian: it's just $\int dx e^{ikx}=\delta(k)$. $\endgroup$ – Meng Cheng Jul 1 '15 at 14:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.