Loop Optimization of Tensor Network Renormalization: Algorithms and Applications

Loop optimization for tensor network renormalization (loop-TNR) is a real-space renormalization group algorithm suitable for studying 1+1D critical systems. While the original proposal by Yang et al. focused on classical models, we extend this algorithm with new techniques to enable accurate and efficient extraction of conformal data from critical quantum models. Benchmark results are provided for a number of quantum models, including ones described by non-minimal or non-unitary conformal field theories, showcasing both the strengths and limitations of loop-TNR. We discuss the subtle issue of non-analytic finite size effect in quantum lattice models and its impact on loop-TNR, and propose the use of virtual-space transfer-matrix to circumvent it, using the XY model as a demonstration. We then generalize loop-TNR to fermionic systems by incorporating Grassmann numbers, and benchmark the generalized algorithm on the t-V model. Next, we demonstrate a non-trivial application of loop-TNR by studying the 1D domain wall between untwisted and twisted 2D lattice gauge theories of finite groups G. We numerically study such domain walls for G = Z_N (with N