TL;DR
This paper introduces an improved version of Block Product Density Matrix Embedding Theory (BPDMET) by incorporating "spin-state superposition" into the impurity cluster. This enhancement leads to more accurate calculations of ground state energies and correlation functions for strongly correlated spin systems, particularly for the challenging Kitaev-Heisenberg model.
What Problem Does This Paper Address?
Calculating the properties of strongly correlated quantum many-body systems is notoriously difficult due to the exponential growth of the Hilbert space with system size. Exact diagonalization becomes computationally infeasible for anything but small systems. Embedding theories, like Density Matrix Embedding Theory (DMET) and its spin-lattice variant Block Product DMET (BPDMET), offer a way to tackle this by treating a small "impurity" cluster exactly and approximating the surrounding "environment." However, the accuracy of these methods can be limited by how well the impurity's wave function captures correlations, especially in systems with strong entanglement. This paper aims to improve the accuracy of BPDMET by enhancing its ability to represent the impurity's quantum state.
Methods
The core of this work is the extension of the BPDMET method. BPDMET approximates the environment using "block product states," which are themselves products of states within individual bath clusters. The key innovation here is the introduction of spin-state superposition within the impurity cluster. Instead of using a fixed basis for the impurity, the authors allow the impurity states to be linear combinations of these basis states, determined by an optimized unitary matrix. This allows the impurity to better capture complex correlations.
The optimization process involves iteratively optimizing both the block product states of the environment and the unitary matrix describing the impurity superposition. This is done by minimizing a variational energy. For calculating excited states and spectral functions, the authors employ a method involving diagonalization in the "tangent space" of the BPDMET ansatz. They also investigate different cluster shapes for the impurity and environment, which can significantly impact the results.
Datasets
Not discussed in the paper. The paper focuses on theoretical models and uses established models (Heisenberg and Kitaev-Heisenberg) as benchmarks.
Key Insights and Ideas
- Spin-state superposition is crucial for accuracy: Allowing the impurity cluster to exist in a superposition of states significantly improves the accuracy of BPDMET, especially in challenging intermediate phases of matter.
- Cluster shape matters: The choice of cluster geometry for the impurity and surrounding bath significantly influences the accuracy of BPDMET, particularly for detecting phase transitions and describing exotic states like spin liquids.
- Variational vs. BPDMET energy: For ground state properties, the non-variational BPDMET energy (derived from expectation values) is found to be more sensitive to improvements than the variational energy. However, for spectral functions, the variational energy is essential for obtaining accurate results.
- BPDMET as a Tensor Network State: The BPDMET ansatz can be naturally represented as a tensor network state, providing a link to other powerful computational techniques.
Key Results
- Improved energies: The introduction of spin-state optimization leads to substantial improvements in the calculated ground state energies for both the J1-J2 Heisenberg model and the Kitaev-Heisenberg model, particularly in the intermediate paramagnetic phase. For example, in the 40-spin Heisenberg model, significant energy improvements were observed.
- Accurate phase transitions: The method correctly identifies the expected phase transitions in the J1-J2 Heisenberg model at J2/J1 ≈ 0.4 and J2/J1 ≈ 0.62. For the Kitaev-Heisenberg model, while phase transitions are detected, the description of the spin liquid phase remains challenging, though improved with spin-state optimization.
- Spectral functions: The tangent space method, when combined with the variational energy, can accurately reproduce spectral functions for systems where BPDMET can find the exact ground state. However, accuracy degrades when the ground state is not perfectly captured.
Limitations and Future Work
- Spin liquid description: While improved, BPDMET still struggles to accurately describe the spin liquid phase in the Kitaev-Heisenberg model, especially with smaller cluster sizes.
- Computational cost: While BPDMET offers linear scaling with system size, increasing the impurity cluster size can lead to exponential scaling in computational time.
- Non-variational energy: The BPDMET energy is not variational, meaning calculated energies are not guaranteed to be upper bounds to the true ground state energy.
- Future work: The authors suggest exploring different ansatzes for bath states and further investigating the connection to tensor network states. They also highlight the potential for BPDMET to study larger systems than exact diagonalization allows. The method's ability to calculate spectral functions opens avenues for studying excited states and dynamic properties.
Summary updated