arXiv:2609.28634
Non-stabilizerness and entanglement in $(2+1)$-dimensional SU(2) lattice gauge theory using tensor networks
Raghav G. Jha, Jaber I. Taher, Muhammad Asaduzzaman, Goksu C. Toga, Bojko N. Bakalov, Alexander F. Kemper
Abstract
We study non-stabilizerness (magic) in the ground state of $(2+1)$-dimensional $\mathrm{SU}(2)$ Hamiltonian lattice gauge theory with matter, formulated in the dressed-site basis in the hardcore-gluon truncation and restricted to the zero baryon-number sector. Using matrix product states, we compute three facets of magic: the second-order stabilizer Rényi entropy (SRE) $M_2$, its non-local component $M_2^{\rm NL}$, estimated by the SRE $M_2(\tildeψ)$ of the computational-basis Schmidt representative $\tildeψ$, and a lower bound in terms of the anti-flatness $F$ of the entanglement spectrum. We also prove a stronger form of the sandwich relation: $-\log_2(1-4F)\le M_2^{\rm NL}\le M_2$; the lower bound rests on a stronger inequality that we obtain for arbitrary Schmidt bases and rank, with an optimal constant. Combined with recent results, this yields the unconditional bracket $\max\{-\log_2(1-4F),\,M_2(\tildeψ)-4\}\le M_2^{\rm NL}\le M_2(\tildeψ)$. We emphasize a structural distinction that makes the non-local quantities the physically preferred diagnostics: whereas the full SRE depends on the (non-unique) encoding of the gauge-invariant local Hilbert space into qubits, the non-local magic, its estimate $M_2(\tildeψ)$, and the anti-flatness are invariant under site-local re-encodings and are therefore intrinsic to the state and bipartition. By varying the gauge coupling for two values of the fermion mass on lattices up to $8\times 8$ ($\sim 288$ qubits) with bond dimension up to $128$, we find that the non-local magic, as estimated by $M_2(\tildeψ)$, furnishes a sharper and more bond-dimension-friendly probe of the gauge-matter delocalization crossover compared to the full SRE or the gauge-invariant entanglement entropy, retaining a clear signal at bond dimensions well below those needed to converge the ground state itself.
AI Summary
The paper asks whether magic provides an encoding-independent diagnostic of gauge–matter delocalization in $(2+1)$-dimensional SU(2) lattice gauge theory. It studies ground states with staggered matter in the zero-baryon sector, using a gauge-invariant dressed-site formulation with the hardcore-gluon cutoff $j\leq1/2$. Two-site DMRG calculations use MPS bond dimensions up to $128$ on open lattices up to $8\times8$ (288 encoded qubits), at masses $m=1$ and $0.2$. The diagnostics are the full second-order stabilizer Rényi entropy $M_2$, the SRE $M_2(\tilde\psi)$ of an ordered computational-basis Schmidt representative, and entanglement-spectrum anti-flatness $F=\operatorname{Tr}\rho_A^3-(\operatorname{Tr}\rho_A^2)^2$. A general matrix-inequality proof establishes $4F\leq1-W$, where $W=2^{-M_2}$, with optimal coefficient $4$, yielding $-\log_2(1-4F)\leq M_2^{\rm NL}$. Combined with existing results, this gives $\max\{-\log_2(1-4F),M_2(\tilde\psi)-4\}\leq M_2^{\rm NL}\leq M_2(\tilde\psi)$. Unlike full SRE, these spectral diagnostics are invariant under site-local re-encodings. Numerically, $M_2(\tilde\psi)$ develops a broad maximum near the previously identified crossover $g^2\sim1$, with less bond-dimension sensitivity than entanglement entropy or full SRE. It remains below $1.9$ bits throughout; the anti-flatness lower bound is universally capped at approximately $0.79$ bits. The plotted Schmidt-representative SRE is an upper bound, not an exact evaluation of non-local magic; equality remains conjectural for general spectra. The largest weak-coupling calculations are not ground-state converged. No representation-cutoff scaling or continuum limit is attempted, and extending beyond the hardcore truncation remains necessary.
Central claim may not hold: The claim that the Schmidt-reference SRE reliably diagnoses the ground-state crossover before the ground state converges is not validated on the larger lattices, where its peak location still shifts with bond dimension and no converged reference is supplied.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- v2: 13+15 pp, v1: 10+14 pp. Comments always welcome!