拓扑不变量, 学术

陈数Chern number的计算(Kubo公式,附Python代码)

另外几篇关于计算陈数的方法:

本篇通过Kubo公式计算陈数。这个方法跟高效法、Wilson loop方法一样,不需要对波函数求导,因此数值计算过程不依赖于波函数相位。

从定义出发,贝里联络(Berry connection)公式为:

An(R)=i⟨n(R)|∂∂R|n(R)⟩A_n(\bm{R})=i\langle n(\bm{R})|\frac{\partial}{\partial \bm{R}}|n(\bm{R})\rangle

贝里曲率(Berry curvature)公式:

Ωn(R)=∇R×An(R)\Omega_n(\bm{R})= \nabla_ {\bm{R}} \times A_n(\bm{R})

展开写为[1]:

Ωμνn(R)=∂∂RμAνn(R)−∂∂RνAμn(R)=i[∂∂Rμ⟨n(R)|∂∂Rν|n(R)⟩−(ν↔μ)]\begin{aligned}\Omega_{\mu\nu}^{n}(\bm{R}) &=\frac{\partial}{\partial R_\mu} A_\nu^n(\bm{R})- \frac{\partial}{\partial R_\nu} A_\mu^n(\bm{R}) \\&= i[\frac{\partial}{\partial R_{\mu}}\langle n(\bm{R})|\frac{\partial}{\partial R_{\nu}}|n(\bm{R})\rangle-(\nu \leftrightarrow \mu)] \end{aligned}

进一步推导(由“向同学”提供):

继续推导需要用到以下等式:

⟨n′|∂H∂R|n⟩=⟨n′|∂∂R|n⟩(En−En′)forn′≠n\langle n'|\frac{\partial H}{\partial \bm{R}}|n\rangle = \langle n'| \frac{\partial}{\partial \bm{R}} |n\rangle(E_n-E_{n'})\quad\text{for}\quad n'\neq n

该等式的证明如下[2]。

薛定谔方程:H|n⟩=En|n⟩H|n\rangle =E_n|n\rangle。

两边求导:∂H∂R|n⟩+H∂|n⟩∂R=∂E∂R|n⟩+En∂|n⟩∂R\frac{\partial H}{\partial \bm{R}}|n\rangle+H\frac{ \partial |n\rangle}{ \partial \bm{R} }=\frac{\partial E}{\partial \bm{R} }|n\rangle+E_n\frac{ \partial |n\rangle }{ \partial \bm{R} }。

两边左乘⟨n′|\langle n'|: ⟨n′|∂H∂R|n⟩+⟨n′|H∂|n⟩∂R=⟨n′|∂En∂R|n⟩+⟨n′|En∂|n⟩∂R\langle n'| \frac{\partial H}{\partial \bm{R}}|n\rangle+ \langle n'| H\frac{ \partial |n\rangle}{ \partial \bm{R} }= \langle n'| \frac{\partial E_n}{\partial \bm{R} }|n\rangle+ \langle n'| E_n\frac{ \partial |n\rangle }{ \partial \bm{R} } 。

整理后: ⟨n′|∂H∂R|n⟩+En′⟨n′|∂∂R|n⟩=En⟨n′|∂∂R|n⟩\langle n'| \frac{\partial H}{\partial \bm{R}}|n\rangle+ E_{n'}\langle n'|\frac{ \partial }{ \partial \bm{R} }|n\rangle=E_n \langle n'|\frac{ \partial }{ \partial \bm{R} }|n\rangle 。其中,考虑了 n′≠nn'\neq n ,因此⟨n′|n⟩=0\langle n'| n \rangle = 0。

继续整理,得到: ⟨n′|∂H∂R|n⟩=⟨n′|∂∂R|n⟩(En−En′)\langle n'|\frac{\partial H}{\partial \bm{R}}|n\rangle = \langle n'| \frac{\partial}{\partial \bm{R}} |n\rangle(E_n-E_{n'}) 。证毕!

接下来继续推导贝里曲率公式,考虑贝里曲率中的第一项。

中间插入单位算符 ∑n′|n′⟩⟨n′|\sum\limits_{n'} |n'\rangle \langle n'|,得到: ∑n′⟨∂∂Rμn|n′⟩⟨n′|∂∂Rν|n⟩\sum\limits_{n'} \langle \frac{\partial}{\partial R_{\mu}} n| n'\rangle \langle n'| \frac{\partial}{\partial R_{\nu}}|n\rangle。由于贝里曲率后面还有一项(ν↔μ)(\nu \leftrightarrow \mu) ,对于n′=nn' =n的情况两项相减为零,因此这一项可以直接写为: ∑n′≠n⟨∂∂Rμn|n′⟩⟨n′|∂∂Rν|n⟩\sum\limits_{n'\neq n} \langle \frac{\partial}{\partial R_{\mu}} n| n'\rangle \langle n'| \frac{\partial}{\partial R_{\nu}}|n\rangle。

代入上面被证明过的表达式,得到: ∑n′≠n⟨n|∂H∂Rμ|n′⟩⟨n′|∂H∂Rν|n⟩(En−En′)2\sum\limits_{n'\neq n} \frac{ \langle n|\frac{\partial H}{\partial R_{\mu}}|n'\rangle \langle n'|\frac{\partial H}{\partial R_{\nu}}|n\rangle }{(E_n-E_{n'})^2}

贝里曲率公式:

Ωμνn(R)=i∑n′≠n⟨n|∂H∂Rμ|n′⟩⟨n′|∂H∂Rν|n⟩−(ν↔μ)(En−En′)2\Omega_{\mu\nu}^{n}(\bm{R}) = i \sum\limits_{n'\neq n} \frac{ \langle n|\frac{\partial H}{\partial R_{\mu}}|n'\rangle \langle n'|\frac{\partial H}{\partial R_{\nu}}|n\rangle-(\nu \leftrightarrow \mu) }{(E_n-E_{n'})^2}

以这篇“陈数Chern number的计算(定义法,附Python/Matlab代码)”中的哈密顿量为例子,计算陈数,Python代码如下(说明:这里只有两条带,所以没有求和。如果存在多条能带,需要对所有其他能带求和):

"""
This code is supported by the website: https://www.guanjihuan.com
The newest version of this code is on the web page: https://www.guanjihuan.com/archives/16148
"""

import numpy as np
from math import *
import time


def hamiltonian(kx, ky):  # one QAH model with Chern number = 2
    t1 = 1.0
    t2 = 1.0
    t3 = 0.5
    m = -1.0
    matrix = np.zeros((2, 2), dtype=complex)
    matrix[0, 1] = 2*t1*cos(kx)-1j*2*t1*cos(ky)
    matrix[1, 0] = 2*t1*cos(kx)+1j*2*t1*cos(ky)
    matrix[0, 0] = m+2*t3*sin(kx)+2*t3*sin(ky)+2*t2*cos(kx+ky)
    matrix[1, 1] = -(m+2*t3*sin(kx)+2*t3*sin(ky)+2*t2*cos(kx+ky))
    return matrix


def main():
    start_time = time.time()
    n = 200       # integration
    delta = 1e-6  # derivation
    chern_number = 0 
    for kx in np.arange(-pi, pi, 2*pi/n):
        for ky in np.arange(-pi, pi,2*pi/n):
            H = hamiltonian(kx, ky)
            eigenvalue, eigenvector = np.linalg.eig(H)
            vector_0 = eigenvector[:, np.argsort(np.real(eigenvalue))[0]]
            vector_1 = eigenvector[:, np.argsort(np.real(eigenvalue))[1]]
            eigenvalue = np.sort(np.real(eigenvalue))

            H_delta_kx = hamiltonian(kx+delta, ky)-hamiltonian(kx, ky) 
            H_delta_ky = hamiltonian(kx, ky+delta)-hamiltonian(kx, ky)

            berry_curvature = 1j*(np.dot(np.dot(np.dot(np.dot(np.dot(vector_0.transpose().conj(), H_delta_kx/delta), vector_1), vector_1.transpose().conj()), H_delta_ky/delta), vector_0)- np.dot(np.dot(np.dot(np.dot(np.dot(vector_0.transpose().conj(), H_delta_ky/delta), vector_1), vector_1.transpose().conj()), H_delta_kx/delta), vector_0))/(eigenvalue[0]-eigenvalue[1])**2

            chern_number = chern_number + berry_curvature*(2*pi/n)**2
    chern_number = chern_number/(2*pi)
    print('Chern number = ', chern_number)
    end_time = time.time()
    print('运行时间(min)=', (end_time-start_time)/60)


if __name__ == '__main__':
    main()

计算结果:Chern number = (-1.9999999999798186+1.9718230813235673e-19j)

补充:Kubo公式也可以推广到简并的情况,计算方法可以看这篇综述文献的公式(73):First-principle calculations of the Berry curvature of Bloch states for charge and spin transport of electrons。

参考资料:

[1] Quantized Hall Conductance in a Two-Dimensional Periodic Potential

[2] Berry phase effects on electronic properties

[3] https://phyx.readthedocs.io/en/latest/TI/Lecture%20notes/2.html

[4] https://mp.weixin.qq.com/s/VJxeLOIyBoJQ1-x-P4Pwtg

评论功能已关闭,以下留言为历史存档,仅供查看。