arXiv:2609.21846
Sine-Gordon Model with Bosonic Tensor Networks: Continuum Matching and Soliton Scattering
Florian Hechenberger, Tommaso Rainaldi, Felix Ringer
Abstract
We use bosonic tensor networks to connect the lattice sine-Gordon model in the Hamiltonian formulation quantitatively to its continuum theory. Matching the lattice vertex operator to its conformal normalization at the free-boson ultraviolet fixed point yields the exact relation between the bare lattice coupling and the renormalized continuum mass parameter. The soliton mass then approaches Zamolodchikov's exact continuum prediction throughout the studied parameter range, without adjustable parameters. Using uniform matrix product states and a quasiparticle ansatz, we also recover the relativistic soliton dispersion and the two lightest breather masses at the percent level. We simulate real-time collisions of Gaussian soliton-antisoliton wave packets near a reflectionless point and extract the Wigner spatial displacement. We compare this displacement with the exact continuum prediction obtained from the momentum derivative of the transmission phase, recovering its characteristic rapidity dependence. Our bosonic simulations provide a foundation for nonintegrable extensions, offer lessons for renormalization in other Hamiltonian lattice field theories, including gauge theories, and provide benchmarks for continuous-variable quantum simulations.
AI Summary
The paper asks how quantitatively a bosonic Hamiltonian lattice simulation of sine-Gordon theory can reproduce the continuum spectrum and soliton scattering once the operator normalization is fixed independently of numerical observables. Matching free-boson ultraviolet vertex correlators fixes the lattice-to-CFT normalization constant to $C=4e^{\gamma_E}$. This gives $m_R^2=m_{\rm bare}^2(a/C)^{\beta^2/(4\pi)}$ and specifies the bare-coupling trajectory for taking the continuum limit. Combined with Zamolodchikov’s mass–coupling relation, it supplies a soliton-mass prediction without fitted parameters. Uniform bosonic MPS and a quasiparticle ansatz yield soliton masses approaching this prediction for $\beta=1$–$3$, together with restoration of relativistic dispersion. At $a=0.3$, local cutoff $d=90$ and bond dimension $\chi=64$, the first two breather masses agree with continuum predictions within about $1\%$; several third-breather estimates deviate substantially. Two-site TDVP simulations collide Gaussian soliton–antisoliton packets at the continuum reflectionless coupling $\beta^2=2\pi$. Charge-density centroids, compared with independent single-soliton reference runs, give Wigner spatial displacements. Four runs at two momenta and lattice spacings $a=1,0.75$ reproduce transmission and the predicted rapidity dependence, but exceed the momentum-averaged continuum displacements by approximately $7$–$15\%$. Packet averaging changes the prediction by roughly $4\%$. The stated limitations include lattice, Fock-space and bond-dimension truncations, finite evolution time, and sensitivity to packet preparation and centroid-analysis windows. Scattering lacks a controlled continuum extrapolation; gauge-dependent packet offsets and residual excitations complicate interpretation. Nonintegrable extensions are proposed rather than demonstrated.
Error found: The lattice-to-continuum matching has an exact leading normalization constant, but its displayed operator relation requires a continuum-limit qualification because finite-cutoff corrections were dropped. The soliton-mass results lack independent Fock-cutoff and bond-dimension convergence sequences, leaving numerical truncation control incompletely demonstrated. The conclusion's claim of scattering-displacement deviations below 10% conflicts with table entries showing approximately 15.0% and 10.4%, while the lower-momentum runs show no resolved improvement with reduced spacing. The breather discussion omits third-breather discrepancies of approximately 24.3% at beta=1.5 and 14.8% at beta=2.0 while highlighting only the comparable discrepancy at beta=1.0.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 13 pages, 5 figures