Table of Contents
Fetching ...

Numerical study of boson mixtures with multi-component continuous matrix product states

Wei Tang, Benoît Tuybens, Jutho Haegeman

Abstract

The continuous matrix product state (cMPS) ansatz is a promising numerical tool for studying quantum many-body systems in continuous space. Although it provides a clean framework that allows one to directly simulate continuous systems, the optimization of cMPS is known to be a very challenging task, especially in the case of multi-component systems. In this work, we have developed an improved optimization scheme for multi-component cMPS that enables simulations of bosonic quantum mixtures with substantially larger bond dimensions than previous works. We benchmark our method on the two-component Lieb-Liniger model, obtaining numerical results that agree well with analytical predictions. Our work paves the way for further numerical studies of quantum mixture systems using the cMPS ansatz.

Numerical study of boson mixtures with multi-component continuous matrix product states

Abstract

The continuous matrix product state (cMPS) ansatz is a promising numerical tool for studying quantum many-body systems in continuous space. Although it provides a clean framework that allows one to directly simulate continuous systems, the optimization of cMPS is known to be a very challenging task, especially in the case of multi-component systems. In this work, we have developed an improved optimization scheme for multi-component cMPS that enables simulations of bosonic quantum mixtures with substantially larger bond dimensions than previous works. We benchmark our method on the two-component Lieb-Liniger model, obtaining numerical results that agree well with analytical predictions. Our work paves the way for further numerical studies of quantum mixture systems using the cMPS ansatz.
Paper Structure (16 sections, 36 equations, 5 figures)

This paper contains 16 sections, 36 equations, 5 figures.

Figures (5)

  • Figure 1: Half-infinite entanglement entropy in the cMPS ground state approximation, plotted with respect to the logarithm of the cMPS correlation length $\xi$ (as naturally obtained for different bond dimensions), for $c_{12} = -0.6, -0.3, 0.0, 0.3, 0.6$ (shown in different colors). The bond dimensions of the cMPS data are $\chi = 8, 16, 32, 64$. The entanglement entropy data for different values of $c_{12}$ are vertically shifted by a constant so that the data points are collapsed onto a single line. The dashed line represents the scaling $S_{\mathrm{EE}} = (C/6) \log(\xi) + S_0$, where $C=2$ is the central charge and $S_0$ is a constant.
  • Figure 2: The cMPS prediction for the Luttinger parameter $K_\pm$ and the sound velocity $v_\pm$ in the charge and spin sectors, respectively. All the results are obtained with bond dimension $\chi = 32$. The dashed lines represent the analytical results obtained from the weak-coupling expansion results \ref{['eq:weak-coupling-approximation-sound-velocity']} and \ref{['eq:weak-coupling-approximation-luttinger-parameter']}.
  • Figure 3: The cMPS results for the correlation functions $C_+(x)$ and $C_-(x)$. For each choice of $c_{12}$, the correlation functions are plotted with respect to the ratio between $x$ and the correlation length $\xi$ of the system. The cMPS results are obtained with bond dimension $\chi=32$. The dashed lines represent the scaling predicted by the Luttinger liquid theory, where the values of $K_+$ and $K_-$ are taken from the numerical results obtained previously.
  • Figure 4: The ground-state particle density $\rho$ as a function of the inter-species interaction strength $c_{12}$. We plot the results with respect to different functions of $c_{12}$, in regard to the TG limit and the weak-coupling limit, respectively. The cMPS results are obtained with bond dimension $\chi=32$. The dashed lines represent the analytical predictions in the two different limits.
  • Figure 5: A comparison of the optimization performance of our optimization scheme with and without preparation steps. As a comparison, we also plot the energy curve of the ordinary L-BFGS optimization scheme in the diagonal gauge. The bond dimension of the cMPS is $\chi=8$, and the Hamiltonian parameters are $c = 10.0$, $c_{12} = -7.0$, and $\mu = 0.0$.