Xiaodong Cao,Yi Lu,E. Miles Stoudenmire,Olivier Parcollet
Abstract:We present an approach to tame the growth of entanglement during time evolution by tensor network methods. It combines time evolution in the complex plane with a perturbative and controlled reconstruction of correlation functions on the real-time axis. We benchmark our approach on the single impurity Anderson model. Compared to purely real-time evolution, the complex time evolution significantly reduces the required bond dimension to obtain the spectral function. Notably, our approach yields self-energy results with high precision at low frequencies, comparable to numerical renormalization group (NRG) results, and it successfully captures the exponentially small Kondo energy scale.
What problem does this paper attempt to address?
This paper aims to solve the problem of entanglement growth during the time - evolution process in quantum many - body systems, especially the difficulties encountered in calculating the real - frequency spectral function. Specifically, the author proposes a method to reduce the growth of entanglement by performing time - evolution in the complex plane and combining a controlled perturbation technique to reconstruct the correlation function on the real axis. This method is particularly suitable for calculating the self - energy at low frequencies, and its accuracy can be comparable to that of the numerical renormalization group (NRG), while being able to capture very small Kondo energy scales.
### Problems the paper attempts to solve:
1. **Entanglement growth problem**: When using the tensor network method for long - time evolution, the entanglement in the system will increase rapidly, resulting in a significant increase in the required bond dimension, which makes the calculation very expensive and difficult to handle.
2. **Accuracy problem in low - frequency self - energy calculation**: Existing tensor - network - based quantum impurity solvers cannot achieve the same accuracy as the numerical renormalization group (NRG) at low frequencies, especially when dealing with large multi - orbital systems.
3. **Difficulties in calculating long - time dynamics**: In real - time evolution, due to the growth of entanglement, the calculation of long - time dynamics becomes very difficult, especially in the low - frequency region.
### Solutions:
The author proposes a two - step method:
1. **Time - evolution in the complex plane**: By choosing an appropriate time contour in the complex plane, high - energy excitations can be suppressed to a certain extent, thereby reducing the growth of entanglement. In this way, even in long - time evolution, a smaller bond dimension can be used to maintain the efficiency and accuracy of the calculation.
2. **Perturbative reconstruction on the real axis**: Using the results of time - evolution in the complex plane, the perturbative expansion technique is used to reconstruct it onto the real axis. This method can maintain high accuracy while avoiding the high computational cost of direct real - time evolution.
### Experimental verification:
The author benchmarked this method on the single - impurity Anderson model (SIAM), and the results show:
- **Calculation of the spectral function**: The complex - time - evolution method can accurately reproduce the entire spectrum, including the low - energy Kondo resonance and the high - energy Hubbard satellite structure, with a smaller bond dimension (for example, \(\chi\sim20\)).
- **Accuracy of self - energy at low frequencies**: The accuracy of self - energy calculation at low frequencies by the complex - time - evolution method is very high, comparable to the NRG results.
- **Extraction of the Kondo energy scale**: The Kondo energy scale can be accurately extracted from the spectrum obtained by complex - time - evolution, and the result is consistent with previous DDMRG studies.
### Conclusion:
This method provides an effective means to significantly improve the computational efficiency and accuracy when calculating the real - frequency spectral function of quantum many - body systems, especially in the low - frequency region. This provides new possibilities for dealing with complex quantum impurity models and may be further applied to the research of multi - orbital quantum impurity models in the future.