method-peps

To simulate 2D quantum lattice systems using the Projected Entangled Pair State (PEPS) ansatz, use a standardized workflow for environment contraction via CTMRG, ground-state optimization through imaginary-time evolution, and automatic differentiation to calculate observables accurately.

The best way to contract tensor networks for 2D frustrated magnets is the Corner Transfer Matrix Renormalization Group (CTMRG) method, which correctly calibrates environment bond dimensions and convergence thresholds to prevent incorrect environment contraction or convergence failures.

iPEPS method selection works by providing a decision framework to choose between PEPS, iPEPS, and classical partition-function contraction based on the physical system, allowing you to set up and converge simulations for square-lattice Heisenberg models.

You need chi-convergence and D-extrapolation for ground-state optimization when rigorous verification is required, ensuring that environment bond dimensions and tensor dimensions are correctly calibrated to achieve target accuracy and match established quantum Monte Carlo results.

Yes, you can use automatic differentiation for ground-state optimization in tensor networks, which is supported alongside Simple/Full Update methods within the workflow to manage the complex pipeline of environment contraction and observable calculation.

CTMRG environment contraction fails during tensor network simulation due to common pitfalls like incorrect environment contraction or convergence failures, which can be prevented by following a structured, expert-curated workflow that calibrates convergence thresholds for target accuracy.