Zero Modes, Klein Factors, and Compactification

Oscillator modes reproduce local power laws, but a bosonized theory on a finite ring is incomplete without charge and current zero modes, compact identifications, boundary twists, and Klein factors. These data determine which states and vertex operators actually exist. They also resolve apparent disagreements between formulas that use different compactification conventions.

Required background. The Abelian Bosonization Dictionary fixes the local fields and vertex normalization. Helpful background. Free Bosons and Vertex Operators supplies the compact-boson state–operator correspondence.

Compact sectors on a ring

Use the convention

$$[\phi(x),\partial_y\theta(y)]=i\pi\delta(x-y), \qquad \rho-\rho_0=-\frac1\pi\partial_x\phi.$$

On a ring of circumference $L$, integer excess particle number $N$ and current quantum number $J$ impose

$$\phi(x+L)=\phi(x)-\pi N, \qquad \theta(x+L)=\theta(x)+\pi J.$$

A useful decomposition is

$$\phi(x)=\phi_0-\frac{\pi N x}{L}+\phi_{\mathrm{osc}}(x), \qquad \theta(x)=\theta_0+\frac{\pi J x}{L}+\theta_{\mathrm{osc}}(x).$$

The zero-mode commutators can be chosen as $[\phi_0,J]=i$ and $[N,\theta_0]=i$, equivalently $[\theta_0,N]=-i$, consistently with the field algebra after treating the periodic delta function. In particular, the factor $e^{+i\theta_0}$ in the fermion annihilation vertex lowers $N$ by one. Substitution into the Luttinger Hamiltonian yields

$$E_{N,J}-E_{0,0} =\frac{\pi u}{2L}\left(\frac{N^2}{K}+KJ^2\right) +\frac{2\pi u}{L}\sum_{q>0}q\,n_q,$$

where the last expression uses dimensionless positive oscillator mode numbers. The $1/L$ spectrum is a direct finite-size test of $u$ and $K$, central to Haldane’s formulation of the Luttinger liquid Haldane 1981, pp. 2585–2609.

Compactification is an operator statement

Because local physical operators contain integer-charge vertices, constant shifts of $\phi$ and $\theta$ are identified only when every allowed operator is unchanged. In the spinless fermion convention, $e^{\pm2i\phi}$ and $e^{\pm2i\theta}$ are local even-fermion operators, whereas $e^{-i(r\phi-\theta)}$ changes fermion parity. Saying merely “$\phi$ has period $\pi$” is therefore insufficient: one must also state whether odd-fermion operators and twisted sectors are included.

A rescaling $\Phi=\phi/\sqrt{\pi K}$ changes the displayed compactification radius and moves $K$ into the vertex charges. It does not change the spectrum or dimensions. The invariant comparison consists of the field commutator, the identification lattice, the Hamiltonian, and the physical vertex lattice taken together.

Selection rules and boundary twists

For periodic microscopic fermions, right- and left-moving occupation changes $N_R,N_L\in\mathbb Z$ give

$$N=N_R+N_L, \qquad J=N_R-N_L,$$

so $N$ and $J$ have the same parity. An inserted flux or a switch between periodic and antiperiodic fermions shifts the allowed current lattice. In bosonic systems the microscopic selection rule can differ. These are global constraints, not corrections that vanish in the thermodynamic limit: they control finite-size degeneracies, persistent currents, and which correlation functions are nonzero.

Threading a dimensionless twist $\alpha=\Phi/\Phi_0$ shifts $J\mapsto J-2\alpha$ in a common charge-one convention. The ground state follows the minimum over allowed $J$, producing piecewise parabolic flux dependence. The period is fixed by the microscopic charge and sector selection, not by treating $J$ as continuous.

Klein factors as zero-mode translators

For multiple species $a$, write

$$\psi_{r,a}=\frac{\eta_{r,a}}{\sqrt{2\pi a_0}} e^{-i(r\phi_a-\theta_a)}, \qquad \{\eta_{r,a},\eta_{r',b}\}=2\delta_{rr'}\delta_{ab}.$$

The exponential changes the appropriate bosonic charge; $\eta_{r,a}$ ensures anticommutation across independent species. In a finite Hilbert space it is often clearer to use unitary ladder operators $F_{r,a}$ that translate $N_{r,a}$ and carry prescribed anticommutation phases. Replacing every Klein factor by the number $1$ is safe only in a fixed calculation where their product is a constant and no competing process has a different species ordering. It fails for tunneling, superconducting terms, or impurity operators that change relative particle numbers. Constructive finite-size treatments are developed in von Delft and Schoeller 1998, §§ 5–7.

A practical consistency check

For any proposed bosonization convention, verify four statements:

1. integrating $-\partial_x\phi/\pi$ gives the declared integer charge;
2. the fermion vertex translates the correct zero modes and anticommutes with other species;
3. its winding around the ring reproduces the microscopic boundary condition;
4. the zero-mode energy agrees with the compressibility and charge stiffness.

Local correlators alone test none of the last three. Conversely, differences in displayed radius or signs are conventional when all four physical statements coincide.

Exercises

1. Minimize the zero-mode energy in the presence of a twist $J\mapsto J-2\alpha$ when $N$ is even and hence $J$ is even.

Solution

Write $J=2m$ with $m\in\mathbb Z$. The current contribution is $(2\pi uK/L)(m-\alpha)^2$. The ground state chooses the integer $m$ nearest to $\alpha$, so adjacent branches cross at half-integer $\alpha$. The minimized energy is periodic under $\alpha\mapsto\alpha+1$.

2. Show that the parity rule follows from integer $N_R,N_L$.

Solution

$N+J=2N_R$ and $N-J=2N_L$ are even. Thus $N$ and $J$ have the same parity. Conversely, if $N$ and $J$ share parity, $(N\pm J)/2$ are integers and define allowed $N_R,N_L$.

References

• Haldane, F. D. M. “‘Luttinger Liquid Theory’ of One-Dimensional Quantum Fluids. I. Properties of the Luttinger Model and Their Extension to the General 1D Interacting Spinless Fermi Gas.” Journal of Physics C: Solid State Physics 14 (1981): 2585–2609. DOI.
• von Delft, J., and H. Schoeller. “Bosonization for Beginners—Refermionization for Experts.” Annalen der Physik 7 (1998): 225–305. DOI.