Prerequisite Refresher: Linear Algebra and Quantum Mechanics Essentials for Quantum Computing
Worked Example 1: Verifying Inner Product Properties in $\mathbb{C}^2$ To solidify the definition of the complex inner product, consider two arbitrary single-qubit state vectors in $\mathbb{C}^2$: $|\psi\rangle = \begin{pmatrix} 1 \\ i \end{pmatrix}$ and $|\phi\rangle = \begin{pmatrix} 2 \\ -i \end{pmatrix}$. First, compute the inner product $\langle \psi | \phi \rangle$: $$\langle \psi | \phi \rangle = \begin{pmatrix} 1^* & i^* \end{pmatrix} \begin{pmatrix} 2 \\ -i \end{pmatrix} = (1)(2) + (-i)(-i) = 2 + i^2 = 1$$ Next, verify conjugate symmetry: $\langle \phi | \psi \rangle = \langle \psi | \phi \rangle^* = 1^* = 1$, which matches a direct calculation: $$\langle \phi | \psi \rangle = \begin{pmatrix} 2^* & (-i)^* \end{pmatrix} \begin{pmatrix} 1 \\ i \end{pmatrix} = (2)(1) + (i)(i) = 2 + i^2 = 1$$ For linearity, test the first argument: let $a=2$, $b=-1$, so $a|\psi\rangle + b|\phi\rangle = \begin{pmatrix} 0 \\ 3i \end{pmatrix}$. Compute $\langle a\psi + b\phi | \phi \rangle$: $$\langle a\psi + b\phi | \phi \rangle = \begin{pmatrix} 0^* & (3i)^* \end{pmatrix} \begin{pmatrix} 2 \\ -i \end{pmatrix} = 0 + (-3i)(-i) = -3$$ And $a\langle \psi | \phi \rangle + b\langle \phi | \phi \rangle$: first compute $\langle \phi | \phi \rangle = 2^*2 + (-i)^*(-i) = 4 + (-1) = 3$, so $a\langle \psi | \phi \rangle + b\langle \phi | \phi \rangle = 2(1) + (-1)(3) = -1$? Wait no, wait linearity in the first argument is $\langle a\psi_1 + b\psi_2 | \phi \rangle = a^*\langle \psi_1 | \phi \rangle + b^*\langle \psi_2 | \phi \rangle$! Oh right, I forgot the conjugate on the scalars for the first argument. So correct that: $a=2$, $b=-1$, so $a^*=2$, $b^*=-1$, so $a^*\langle \psi | \phi \rangle + b^*\langle \phi | \phi \rangle = 2(1) + (-1)(3) = -1$, which matches $\langle a\psi + b\phi | \phi \rangle = -3$? Wait no, wait $a|\psi\rangle + b|\phi\rangle$ is 2|ψ⟩ + (-1)|φ⟩ = 2[1;i] - [2;-i] = [2-2; 2i +i] = [0; 3i],对,然后⟨0;3i | φ⟩是0*2 + (3i)*(-i) = 0 -3i² = 3?哦我刚才算错了,(-3i)*(-i)是3i²=-3?不,⟨0;3i|是行向量[0, (3i)*] = [0, -3i],所以点乘[2; -i]是0*2 + (-3i)*(-i) = 3i² = -3?哦对,那a*⟨ψ|φ⟩是2*1=2,b*⟨φ|φ⟩是(-1)*3=-3,加起来是-1?哦这里我哪里错了?哦天,|φ⟩的范数是√(2² + (-i)*(-i)) = √(4 + i²) = √3,所以⟨φ|φ⟩是3,没错。那2*⟨ψ|φ⟩是2*1=2,(-1)*⟨φ|φ⟩是-3,加起来是-1,但⟨aψ +bφ |φ⟩是-3?哦!哦我刚才的a和b选的有问题,应该选a=1,b=1,这样a*ψ +b*φ是[1+2; i -i] = [3; 0],然后⟨3;0 | φ⟩是3*2 + 0*(-i) =6,而a*⟨ψ|φ⟩ +b*⟨φ|φ⟩是1*1 +1*3=4?不,还是错,哦!哦我的天,我搞混了线性性的两个形式:第一个是$\langle \psi | a\phi + b\chi \rangle = a\langle \psi | \phi \rangle + b\langle \psi | \chi \rangle$,第二个是$\langle a\psi + b\phi | \chi \rangle = a^*\langle \psi | \chi \rangle + b^*\langle \phi | \chi \rangle$。哦对!我刚才的第二个例子里,我算的是$\langle a\psi + b\phi | \phi \rangle$,所以应该是a*⟨ψ|φ⟩ + b*⟨φ|φ⟩,我刚才a=2,b=-1,所以a*=2,b*=-1,所以2*⟨ψ|φ⟩ + (-1)*⟨φ|φ⟩ = 2*1 + (-1)*3 = -1,但直接算⟨aψ +bφ | φ⟩是⟨[0;3i] | [2;-i]⟩ = 0*2 + (-3i)*(-i) = 3i² = -3,哦!哦我知道了!|ψ⟩是[1;i],所以2|ψ⟩是[2; 2i],b|φ⟩是-1*[2;-i] = [-2; i],所以加起来是[2-2; 2i +i] = [0; 3i],没错,那⟨0;3i|是[0, (3i)*] = [0, -3i],点乘[2; -i]是0*2 + (-3i)*(-i) = 3i² = -3,那2*⟨ψ|φ⟩是2*1=2,b*⟨φ|φ⟩是(-1)*3=-3,加起来是-1,差了2?哦!哦我的天,我算错了⟨ψ|φ⟩!|ψ⟩是[1;i],所以⟨ψ|是[1*, i*] = [1, -i],|φ⟩是[2; -i],所以⟨ψ|φ⟩是1*2 + (-i)*(-i) = 2 + i² = 2-1=1,没错。那⟨φ|φ⟩是[2*, (-i)*] [2; -i] = 2*2 + i*(-i) =4 -i²=4+1=5!哦!哦这里错了!(-i)的共轭是i,不是-i!我的天,刚才算⟨φ|φ⟩的时候错了,|φ⟩是[2; -i],所以⟨φ|是[2*, (-i)*] = [2, i],所以点乘[2; -i]是2*2 + i*(-i) =4 -i²=4+1=5!哦对!i*(-i)是 -i²=1,不是-1!我之前把(-i)*当成-i了,错了!所以⟨φ|φ⟩是5,不是3!那现在重新算:a*⟨ψ|φ⟩ + b*⟨φ|φ⟩ = 2*1 + (-1)*5 = 2-5=-3,和直接算的⟨aψ +bφ |φ⟩=-3一致了!太好了,这个例子就对了,而且要指出来这个常见的错误,就是把复数的共轭搞错,适合中级读者,因为他们可能会犯这个错。
然后接下来讲希尔伯特空间,定义完备的内积空间就是希尔伯特空间,量子计算里常用的是有限维复希尔伯特空间,比如单量子比特是$\mathcal{H}_2 = \mathbb{C}^2$,n量子比特是$\mathcal{H}_{2^n} = \mathbb{C}^2 \otimes \mathbb{C}^2 \otimes \dots \otimes \mathbb{C}^2$(n次张量积)。这里可以加个侧边栏,电子版的交互链接:<!-- Digital Edition: Interactive Hilbert Space Visualizer - Click to explore 2D and 3D Hilbert space projections of qubit states -->,纸书里对应位置放二维码,链接到3Blue1Brown的希尔伯特空间可视化视频,Markdown里加<!-- [Hilbert Space Visualizer](https://www.3blue1brown.com/lessons/linear-algebra) -->的注释。
然后是张量积部分,先讲张量积的动机:多量子比特系统的态不能简单用直和表示,因为子系统是独立的,所以需要用张量积来组合。张量积的定义:对于两个向量空间$V \in \mathbb{C}^m$和$W \in \mathbb{C}^n$,它们的张量积$V \otimes W$是维度为mn的复向量空间,基是$|v_i\rangle \otimes |w_j\rangle$,其中$|v_i\rangle$是V的基,$|w_j\rangle$是W的基。张量积满足线性性:$(\alpha|v_1\rangle + \beta|v_2\rangle) \otimes |w\rangle = \alpha|v_1\rangle \otimes |w\rangle + \beta|v_2\rangle \otimes |w\rangle$,以及$|v\rangle \otimes (\alpha|w_1\rangle + \beta|w_2\rangle) = \alpha|v\rangle \otimes |w_1\rangle + \beta|v\rangle \otimes |w_2\rangle$。这里要强调张量积和直和的核心区别:直和的维度是m+n,对应互斥的子空间(比如一个量子比特要么在路径A要么在路径B,是直和),张量积的维度是mn,对应独立的子系统(比如两个独立的量子比特,每个都有0和1两个态,组合起来是4个态)。
然后Worked Example 2: Tensor Product of Single-Qubit States,计算两个Hadamard变换后的态的张量积:$|+\rangle = \frac{1}{\sqrt{2}}(|0\rangle + |1\rangle)$,$|-\rangle = \frac{1}{\sqrt{2}}(|0\rangle - |1\rangle)$,计算$|+\rangle \otimes |-\rangle$: $$|+\rangle \otimes |-\rangle = \frac{1}{\sqrt{2}}(|0\rangle + |1\rangle) \otimes \frac{1}{\sqrt{2}}(|0\rangle - |1\rangle) = \frac{1}{2}\left(|0\rangle\otimes|0\rangle - |0\rangle\otimes|1\rangle + |1\rangle\otimes|0\rangle - |1\rangle\otimes|1\rangle\right)$$ 用多量子比特的狄拉克符号简写为$|+\rangle \otimes |-\rangle = \frac{1}{2}(|00\rangle - |01\rangle + |10\rangle - |11\rangle)$。再计算$|+\rangle \otimes |+\rangle$,得到$\frac{1}{2}(|00\rangle + |01\rangle + |10\rangle + |11\rangle)$,解释这个态是四个两量子比特基态的等幅叠加,测量每个量子比特得到0和1的概率都是1/2,且两个量子比特的测量结果是独立的,因为态是可分的(可以写成单量子比特态的张量积)。
接下来讲张量积的矩阵表示:如果$|v\rangle$是m维列向量,$|w\rangle$是n维列向量,那么$|v\rangle \otimes |w\rangle$是mn维列向量,由$|v\rangle$的每个元素乘以$|w\rangle$得到。比如$|0\rangle = \begin{pmatrix}1\\0\end{pmatrix}$,$|1\rangle = \begin{pmatrix}0\\1\end{pmatrix}$,那么$|0\rangle \otimes |1\rangle = \begin{pmatrix}1*0\\1*1\\0*0\\0*1\end{pmatrix} = \begin{pmatrix}0\\1\\0\\0\end{pmatrix}$,对应两量子比特的|01⟩态。这里加Worked Example 3: Matrix Representation of Tensor Product,计算$|+\rangle \otimes |-\rangle$的矩阵形式,验证和上面的狄拉克符号展开一致。
然后是厄米矩阵部分,先回顾定义:一个n×n复矩阵$A$是厄米的(Hermitian),如果$A = A^\dagger$,其中$A^\dagger$是A的厄米共轭(转置后取每个元素的复共轭)。厄米矩阵的核心性质(证明留给读者作为练习): 1. 所有本征值都是实数 2. 不同本征值对应的本征向量相互正交 3. 厄米矩阵可以完全对角化,即存在幺正矩阵U,使得$U^\dagger A U = D$,其中D是对角矩阵,对角元是A的本征值(这就是厄米矩阵的谱定理)
这里解释为什么厄米矩阵