Integrable Spherical Brane Model at Large $N$

TL;DR

This work tests the Lukyanov-Zamolodchikov conjecture for the boundary free energy of the integrable spherical brane model by performing a large-$N$ expansion. The authors formulate the boundary constraint with a Lagrange multiplier, solve the leading order saddle point to obtain a renormalized energy $\mathcal{E}_{lead}$ with $E_0 = 4\pi E^*$, and then compute the next-to-leading order correction via a one-loop determinant, carefully renormalizing UV divergences. The subleading analysis yields a finite $1/E^*$ correction that, together with the leading result, reproduces the conjectured coefficients $C_k$ for $k\ge1$ and aligns with the RG running of $g_0$. The results reinforce the integrable boundary structure of the model and illustrate how large-$N$ methods illuminate the interplay between dissipation, boundary RG flows, and boundary integrability in two dimensions.

Abstract

We study one of the simplest integrable two-dimensional quantum field theories with a boundary: $N$ free non-compact scalars in the bulk, constrained non-linearly on the boundary to lie on an $(N-1)$-sphere of radius $1/\sqrt{g}$. The $N=1$ case reduces to the single-channel Kondo problem, for $N=2$ the model describes dissipative Coulomb charging in quantum dots, and larger $N$ is analogous to higher-spin impurity or multi-channel scenarios. Adding a boundary magnetic field -- a linear boundary coupling to the scalars -- enriches the model's structure while preserving integrability. Lukyanov and Zamolodchikov (2004) conjectured an expansion for the boundary free energy on the infinite half-cylinder in powers of the magnetic field. Using large-$N$ saddle-point techniques, we confirm their conjecture to next-to-leading order in $1/N$. Renormalization of the subleading solution turns out to be highly instructive, and we connect it to the RG running of $g$ studied by Giombi and Khanchandani (2020).