前提背景
Affleck, Kennedy, Lieb, and Tasaki(AKLT)模型:
AKLT模型:是一维的延伸量子(海森堡自旋模型)
通过构造,AKLT哈密顿量的基态是带有连接每个相邻位点对的单价键的价键固体。在图形上,这可以表示为:
在这里,实心点表示自旋1/2,并处于单重态。连接自旋1/2的线是表示单线态的价键。椭圆是投影算子,它们将两个自旋1/2一起“绑定”为一个自旋1,突出自旋0或单重态子空间,并仅保留自旋1或三重态子空间。符号“ +”,“ 0”和“-”标记了标准自旋1基态(
S
z
S^{z}
Sz运算符)。
正文:
对于一个具有𝑛个自由度的量子多体系统:
问题:一共
d
N
d^N
dN个元素这是一个巨大无比的张量计算机根本无法存储
解决方案:将其化为一个近似的特殊的张量网络表达,下面以MPS态表示
矩阵乘积态有两种不同的MPS形式,在本文中,我们将运用最大纠缠态来制备周期边界条件的MPS(PBC-MPS)如图所示:
注:其中每个粒子各自的取值情况数为2,分别为0,1
a
i
b
i
a_{i}b_{i}
aibi这两个粒子是两个不相关的粒子
1
2
(
∣
01
⟩
−
∣
10
⟩
)
b
i
a
i
+
1
\frac{1}{\sqrt{2}}(|01\rangle-|10\rangle)_{b_{i} a_{i+1}}
21(∣01⟩−∣10⟩)biai+1
第一步
我们对这个纠缠做一个变换:
1
2
(
∣
01
⟩
−
∣
10
⟩
)
b
i
a
i
+
1
=
1
2
[
0
1
−
1
0
]
b
i
a
i
+
1
(
∣
00
⟩
+
∣
11
⟩
)
b
i
a
i
+
1
\frac{1}{\sqrt{2}}(|01\rangle-|10\rangle)_{b_{i} a_{i+1}}=\frac{1}{\sqrt{2}}\left[\begin{array}{cc}0 & 1 \\ -1 & 0\end{array}\right]_{b_{i} a_{i+1}}(|00\rangle+|11\rangle)_{b_{i} a_{i+1}}
21(∣01⟩−∣10⟩)biai+1=21[0−110]biai+1(∣00⟩+∣11⟩)biai+1
证明:
①方法一
因为:
∣
0
⟩
=
(
1
,
0
)
T
,
∣
1
⟩
=
(
0
,
1
)
T
|0\rangle = (1,0)^T,|1\rangle = (0,1)^T
∣0⟩=(1,0)T,∣1⟩=(0,1)T
所以
∣
01
⟩
=
(
0
,
1
,
0
,
0
)
T
,
∣
10
⟩
=
(
0
,
0
,
1
,
0
)
T
,
∣
00
⟩
=
(
1
,
0
,
0
,
0
)
T
,
∣
11
⟩
=
(
0
,
0
,
0
,
1
)
T
|01\rangle = (0,1,0,0)^T,|10\rangle = (0,0,1,0)^T,|00\rangle = (1,0,0,0)^T,|11\rangle = (0,0,0,1)^T
∣01⟩=(0,1,0,0)T,∣10⟩=(0,0,1,0)T,∣00⟩=(1,0,0,0)T,∣11⟩=(0,0,0,1)T
∣
01
⟩
−
∣
10
⟩
=
(
0
,
1
,
−
1
,
0
)
T
,
∣
00
⟩
+
∣
11
⟩
=
(
1
,
0
,
0
,
1
)
T
|01\rangle-|10\rangle = (0,1,-1,0)^T,|00\rangle+|11\rangle = (1,0,0,1)^T
∣01⟩−∣10⟩=(0,1,−1,0)T,∣00⟩+∣11⟩=(1,0,0,1)T
将
∣
01
⟩
−
∣
10
⟩
=
(
0
,
1
,
−
1
,
0
)
T
|01\rangle-|10\rangle = (0,1,-1,0)^T
∣01⟩−∣10⟩=(0,1,−1,0)Treshape为矩阵
[
0
1
−
1
0
]
\left[ \begin{matrix} 0 & 1 \\ -1 & 0 \end{matrix} \right]
[0−110]将
∣
00
⟩
+
∣
11
⟩
=
(
1
,
0
,
0
,
1
)
T
|00\rangle+|11\rangle = (1,0,0,1)^T
∣00⟩+∣11⟩=(1,0,0,1)Treshape为矩阵
[
1
0
0
1
]
\left[ \begin{matrix} 1 & 0 \\ 0 & 1 \end{matrix} \right]
[1001]因为:
[
0
1
−
1
0
]
×
[
1
0
0
1
]
=
[
0
1
−
1
0
]
\left[ \begin{matrix} 0 & 1 \\ -1 & 0 \end{matrix} \right] \times \left[ \begin{matrix} 1 & 0 \\ 0 & 1 \end{matrix} \right] = \left[ \begin{matrix} 0 & 1 \\ -1 & 0 \end{matrix} \right]
[0−110]×[1001]=[0−110]将
[
0
1
−
1
0
]
\left[ \begin{matrix} 0 & 1 \\ -1 & 0 \end{matrix} \right]
[0−110]reshape为向量即为
(
0
,
1
,
−
1
,
0
)
T
=
∣
01
⟩
−
∣
10
⟩
(0,1,-1,0)^T = |01\rangle-|10\rangle
(0,1,−1,0)T=∣01⟩−∣10⟩
②方法二
由①可知
∣
00
⟩
+
∣
11
⟩
=
(
1
,
0
,
0
,
1
)
T
|00\rangle+|11\rangle = (1,0,0,1)^T
∣00⟩+∣11⟩=(1,0,0,1)Treshape出来的是一个单位阵,与任何矩阵乘积不改变原来矩阵形式
令
M
b
i
a
i
+
1
=
1
2
[
0
1
−
1
0
]
b
i
a
i
+
1
M_{b_{i} a_{i+1}} = \frac{1}{\sqrt{2}}\left[\begin{array}{cc}0 & 1 \\ -1 & 0\end{array}\right]_{b_{i} a_{i+1}}
Mbiai+1=21[0−110]biai+1
则原式
=
M
b
i
a
i
+
1
(
∣
00
⟩
+
∣
11
⟩
)
b
i
a
i
+
1
=M_{b_{i} a_{i+1}}(|00\rangle+|11\rangle)_{b_{i} a_{i+1}}
=Mbiai+1(∣00⟩+∣11⟩)biai+1
因为
∣
00
⟩
+
∣
11
⟩
|00\rangle+|11\rangle
∣00⟩+∣11⟩为单位阵,对于它的作用相当与对
b
i
a
i
+
1
b_{i} a_{i+1}
biai+1的作用
则对于整个模型而言即为:
∑
b
i
a
i
+
1
M
b
i
a
i
+
1
∣
b
i
a
i
+
1
⟩
\sum_{b_{i} a_{i+1}} M_{b_{i} a_{i+1}}\left|b_{i} a_{i+1}\right\rangle
∑biai+1Mbiai+1∣biai+1⟩
第二步
注:再次声明——这两个粒子是两个不相关的粒子。各自的取值情况数为2,因此总的希尔伯特空间维度为4。但是我们可以将它们投影到一个更小的空间上去。
①:定义投影算符
(
σ
i
∈
{
+
1
,
0
,
−
1
}
)
\left(\sigma_{i} \in\{+1,0,-1\}\right)
(σi∈{+1,0,−1}) :
P
i
=
∑
a
i
b
i
σ
i
N
a
i
b
i
σ
i
∣
σ
i
⟩
⟨
a
i
b
i
∣
\begin{array}{l} P_{i}=\sum_{a_{i} b_{i} \sigma_{i}} N_{a_{i} b_{i}}^{\sigma_{i}}\left|\sigma_{i}\right\rangle\left\langle a_{i} b_{i}\right| \end{array}
Pi=∑aibiσiNaibiσi∣σi⟩⟨aibi∣其中:
N
(
+
1
)
=
[
1
0
0
0
]
,
N
(
0
)
=
[
0
1
2
1
2
0
]
,
N
(
−
1
)
=
[
0
0
0
1
]
\begin{array}{l} N^{(+1)}=\left[\begin{array}{ll} 1 & 0 \\ 0 & 0 \end{array}\right], \quad N^{(0)}=\left[\begin{array}{cc} 0 & \frac{1}{\sqrt{2}} \\ \frac{1}{\sqrt{2}} & 0 \end{array}\right], \quad N^{(-1)}=\left[\begin{array}{ll} 0 & 0 \\ 0 & 1 \end{array}\right] \end{array}
N(+1)=[1000],N(0)=[021210],N(−1)=[0001]
②:将投影算符
P
i
P_{i}
Pi展开观察:
P
i
=
∑
a
i
b
i
σ
i
N
a
i
b
i
σ
i
∣
σ
i
⟩
⟨
a
i
b
i
∣
P_{i}=\sum_{a_{i} b_{i} \sigma_{i}} N_{a_{i} b_{i}}^{\sigma_{i}}\left|\sigma_{i}\right\rangle\left\langle a_{i} b_{i}\right|
Pi=∑aibiσiNaibiσi∣σi⟩⟨aibi∣
=
N
00
(
+
1
)
∣
+
1
⟩
⟨
00
∣
+
N
01
(
0
)
∣
0
⟩
⟨
01
∣
+
N
10
(
0
)
∣
0
⟩
⟨
10
∣
+
N
11
(
−
1
)
∣
−
1
⟩
⟨
11
∣
=N_{00}^{(+1)}|+1\rangle\langle 00|+N_{01}^{(0)}| 0\rangle\langle 01|+N_{10}^{(0)}| 0\rangle\langle 10|+N_{11}^{(-1)}|-1\rangle\langle 11|
=N00(+1)∣+1⟩⟨00∣+N01(0)∣0⟩⟨01∣+N10(0)∣0⟩⟨10∣+N11(−1)∣−1⟩⟨11∣
=
[
1
0
0
0
]
∣
+
1
⟩
⟨
00
∣
+
[
0
1
2
1
2
0
]
∣
0
⟩
⟨
01
∣
+
[
0
1
2
1
2
0
]
∣
0
⟩
⟨
10
∣
+
[
0
0
0
1
]
∣
−
1
⟩
⟨
11
∣
=\left[\begin{array}{cc}1 & 0 \\ 0 & 0\end{array}\right]|+1\rangle\langle 00|+\left[\begin{array}{cc}0 & \frac{1}{\sqrt{2}} \\ \frac{1}{\sqrt{2}} & 0\end{array}\right]| 0\rangle\langle 01|+\left[\begin{array}{cc}0 & \frac{1}{\sqrt{2}} \\ \frac{1}{\sqrt{2}} & 0\end{array}\right]| 0\rangle\langle 10|\\+\left[\begin{array}{cc}0 & 0 \\ 0 & 1\end{array}\right]|-1\rangle\langle 11|
=[1000]∣+1⟩⟨00∣+[021210]∣0⟩⟨01∣+[021210]∣0⟩⟨10∣+[0001]∣−1⟩⟨11∣
③:将投影算符
P
i
P_{i}
Pi作用在粒子上
∣
Ψ
⟩
=
P
1
P
2
⋯
P
N
∑
b
1
a
2
M
b
1
a
2
∣
b
1
a
2
⟩
∑
b
2
a
3
M
b
2
a
3
∣
b
2
a
3
⟩
⋯
∑
b
N
a
1
M
b
N
a
1
∣
b
N
a
1
⟩
|\Psi\rangle=P_{1} P_{2} \cdots P_{N} \sum_{b_{1} a_{2}} M_{b_{1} a_{2}}\left|b_{1} a_{2}\right\rangle\sum_{b_{2} a_{3}} M_{b_{2} a_{3}}| b_{2} a_{3}\rangle \cdots \sum_{b_{N} a_{1}} M_{b_{N} a_{1}}\left|b_{N} a_{1}\right\rangle
∣Ψ⟩=P1P2⋯PN∑b1a2Mb1a2∣b1a2⟩∑b2a3Mb2a3∣b2a3⟩⋯∑bNa1MbNa1∣bNa1⟩
=
∑
a
1
b
1
σ
1
N
a
1
b
1
σ
1
∣
σ
1
⟩
⟨
a
1
b
1
∣
∑
a
2
b
2
σ
2
N
a
2
b
2
σ
2
∣
σ
2
⟩
⟨
a
2
b
2
∣
⋯
∑
a
N
b
N
σ
N
N
a
N
b
N
σ
N
∣
σ
N
⟩
⟨
a
N
b
N
∣
∑
b
1
a
2
M
b
1
a
2
∣
b
1
a
2
⟩
∑
b
2
a
3
M
b
2
a
3
∣
b
2
a
3
⟩
⋯
∑
b
N
a
1
M
b
N
a
1
∣
b
N
a
1
⟩
=\sum_{a_{1} b_{1} \sigma_{1}} N_{a_{1} b_{1}}^{\sigma_{1}}|\sigma_{1}\rangle\langle a_{1} b_{1}| \sum_{a_{2} b_{2} \sigma_{2}} N_{a_{2} b_{2}}^{\sigma_{2}}|\sigma_{2}\rangle\langle a_{2} b_{2}| \cdots \sum_{a_{N} b_{N} \sigma_{N}} N_{a_{N} b_{N}}^{\sigma_{N}}| \sigma_{N}\rangle \langle a_{N} b_{ N} | \\ \sum_{b_{1} a_{2}} M_{b_{1} a_{2}}| b_{1} a_{2}\rangle \sum_{b_{2} a_{3}} M_{b_{2} a_{3}}| b_{2} a_{3}\rangle \cdots \sum_{b_{N} a_{1}} M_{b_{N} a_{1}}|b_{N} a_{1}\rangle
=∑a1b1σ1Na1b1σ1∣σ1⟩⟨a1b1∣∑a2b2σ2Na2b2σ2∣σ2⟩⟨a2b2∣⋯∑aNbNσNNaNbNσN∣σN⟩⟨aNbN∣∑b1a2Mb1a2∣b1a2⟩∑b2a3Mb2a3∣b2a3⟩⋯∑bNa1MbNa1∣bNa1⟩
注:
⟨
a
1
b
1
∣
⟨
a
2
b
2
∣
⋯
⟨
a
N
b
N
∣
与
∣
b
1
a
2
⟩
∣
b
2
a
3
⟩
⋯
∣
b
N
a
1
⟩
\langle a_{1} b_{1}|\langle a_{2} b_{2}|\cdots\langle a_{N} b_{N}|与| b_{1} a_{2}\rangle| b_{2} a_{3}\rangle\cdots|b_{N} a_{1}\rangle
⟨a1b1∣⟨a2b2∣⋯⟨aNbN∣与∣b1a2⟩∣b2a3⟩⋯∣bNa1⟩计算为
⟨
a
1
b
1
a
2
b
2
⋯
a
N
b
N
∣
b
1
a
2
b
2
a
3
⋯
b
N
a
1
⟩
\langle a_{1} b_{1}a_{2} b_{2}\cdots a_{N} b_{N}| b_{1} a_{2}b_{2} a_{3}\cdots b_{N} a_{1}\rangle
⟨a1b1a2b2⋯aNbN∣b1a2b2a3⋯bNa1⟩(其中:
⟨
a
N
∣
a
N
⟩
\langle a_{N} | a_{N}\rangle
⟨aN∣aN⟩ =
⟨
b
N
∣
b
N
⟩
\langle b_{N} | b_{N}\rangle
⟨bN∣bN⟩ =1)
所以其最后的值为1
=
∑
a
i
b
i
σ
i
(
N
a
1
b
1
σ
1
M
b
1
a
2
)
⋯
(
N
a
N
b
N
σ
N
M
b
N
a
1
)
∣
σ
1
,
⋯
,
σ
N
⟩
=\sum_{a_{i} b_{i} \sigma_{i}}\left(N_{a_{1} b_{1}}^{\sigma_{1}} M_{b_{1} a_{2}}\right) \cdots\left(N_{a_{N} b_{N}}^{\sigma_{N}} M_{b_{N} a_{1}}\right)\left|\sigma_{1}, \cdots, \sigma_{N}\right\rangle
=∑aibiσi(Na1b1σ1Mb1a2)⋯(NaNbNσNMbNa1)∣σ1,⋯,σN⟩
∣
Ψ
⟩
=
∑
a
i
b
i
σ
i
(
N
a
1
b
1
σ
1
M
b
1
a
2
)
⋯
(
N
a
N
b
N
σ
N
M
b
N
a
1
)
∣
σ
1
,
⋯
,
σ
N
⟩
\begin{array}{c}|\Psi\rangle=\sum_{a_{i} b_{i} \sigma_{i}}\left(N_{a_{1} b_{1}}^{\sigma_{1}} M_{b_{1} a_{2}}\right) \cdots\left(N_{a_{N} b_{N}}^{\sigma_{N}} M_{b_{N} a_{1}}\right)\left|\sigma_{1}, \cdots, \sigma_{N}\right\rangle\end{array}
∣Ψ⟩=∑aibiσi(Na1b1σ1Mb1a2)⋯(NaNbNσNMbNa1)∣σ1,⋯,σN⟩令
N
a
1
b
1
σ
1
M
b
1
a
2
=
A
a
1
a
2
σ
1
,
⋯
,
N
a
N
b
N
σ
N
M
b
N
a
1
=
A
a
N
a
1
σ
1
\begin{array}{c} N_{a_{1} b_{1}}^{\sigma_{1}} M_{b_{1} a_{2}}=A_{a_{1} a_{2}}^{\sigma_{1}}, \cdots, N_{a_{N} b_{N}}^{\sigma_{N}} M_{b_{N} a_{1}}=A_{a_{N} a_{1}}^{\sigma_{1}} \end{array}
Na1b1σ1Mb1a2=Aa1a2σ1,⋯,NaNbNσNMbNa1=AaNa1σ1
∣
Ψ
⟩
=
∑
a
i
b
i
σ
i
(
A
a
1
a
2
σ
1
)
⋯
(
A
a
N
a
1
σ
1
)
∣
σ
1
,
⋯
,
σ
N
⟩
\begin{array}{c} |\Psi\rangle=\sum_{a_{i} b_{i} \sigma_{i}}\left(A_{a_{1} a_{2}}^{\sigma_{1}}\right) \cdots\left(A_{a_{N} a_{1}}^{\sigma_{1}}\right)\left|\sigma_{1}, \cdots, \sigma_{N}\right\rangle\end{array}
∣Ψ⟩=∑aibiσi(Aa1a2σ1)⋯(AaNa1σ1)∣σ1,⋯,σN⟩
第三步
由第二步中
A
a
1
a
2
σ
1
=
N
a
1
b
1
σ
1
M
b
1
a
2
,
⋯
,
A
a
N
a
1
σ
1
=
N
a
N
b
N
σ
N
M
b
N
a
1
\begin{array}{c} A_{a_{1} a_{2}}^{\sigma_{1}}=N_{a_{1} b_{1}}^{\sigma_{1}} M_{b_{1} a_{2}}, \cdots, A_{a_{N} a_{1}}^{\sigma_{1}}=N_{a_{N} b_{N}}^{\sigma_{N}} M_{b_{N} a_{1}} \end{array}
Aa1a2σ1=Na1b1σ1Mb1a2,⋯,AaNa1σ1=NaNbNσNMbNa1我们来看看算符
A
A
A的值:
A
(
+
1
)
=
N
(
+
1
)
M
=
[
1
0
0
0
]
[
0
1
2
−
1
2
0
]
=
[
0
1
2
0
0
]
A^{(+1)} = N^{(+1)}M= \left[\begin{array}{cc}1 & 0 \\ 0 & 0\end{array}\right] \left[\begin{array}{cc}0 & \frac{1}{\sqrt{2}} \\ -\frac{1}{\sqrt{2}} & 0\end{array}\right] = \left[\begin{array}{cc}0 & \frac{1}{\sqrt{2}} \\ 0 & 0\end{array}\right]
A(+1)=N(+1)M=[1000][0−21210]=[00210]
A
(
0
)
=
N
(
0
)
M
=
[
0
1
2
1
2
0
]
[
0
1
2
−
1
2
0
]
=
[
−
1
2
0
0
1
2
]
A^{(0)} =N^{(0)}M= \left[\begin{array}{cc}0 & \frac{1}{\sqrt{2}} \\ \frac{1}{\sqrt{2}} & 0\end{array}\right] \left[\begin{array}{cc}0 & \frac{1}{\sqrt{2}} \\ -\frac{1}{\sqrt{2}} & 0\end{array}\right] = \left[\begin{array}{cc}-\frac{1}{2} & 0 \\ 0 & \frac{1}{2}\end{array}\right]
A(0)=N(0)M=[021210][0−21210]=[−210021]
A
(
−
1
)
=
N
(
−
1
)
M
=
[
0
0
0
1
]
[
0
1
2
−
1
2
0
]
=
[
0
0
−
1
2
0
]
A^{(-1)} =N^{(-1)}M= \left[\begin{array}{cc}0 & 0 \\ 0 & 1\end{array}\right] \left[\begin{array}{cc}0 & \frac{1}{\sqrt{2}} \\ -\frac{1}{\sqrt{2}} & 0\end{array}\right] = \left[\begin{array}{cc}0 & 0 \\ -\frac{1}{\sqrt{2}} & 0\end{array}\right]
A(−1)=N(−1)M=[0001][0−21210]=[0−2100]
矩阵
A
a
i
a
i
+
1
σ
i
A_{a_{i} a_{i+1}}^{\sigma_{i}}
Aaiai+1σi的值取决于
M
M
M的值,即最终MPS中各张量的关联关系与每一个纠缠态的纠缠特性有关系,而这个特性就是由
M
M
M矩阵决定的。
注:
σ
i
\sigma_{i}
σi是开放的indices,在一些文献中称为“fixed collection of physical indices” 。
而
a
i
a_i
ai很明显,是粒子的编号,在这里叫做“connecting bond indices”。