Quantum Mechanics
Done Right中曾说物理学家喜欢在内积的定义中采用第二个位置的加性和齐性,原来是因为物理学中习惯用右矢
Shankar_ch.1
1.6 notes
名词对应:
- Unitary(酉算子)— 等距同构
- Hermitian — self-adjoint(自伴)
Dirac记号非常直观,比如算子的矩阵 $\mathcal{M}(\Omega)$,按照线性代数中的定义,其中元素满足:
$$
\Omega v_j = \sum_i \mathcal{M}(\Omega)_{ij} v_i
$$
在物理里全是正交归一基,且不顾忌无限维向量空间与有限维不同的问题,这样直接考虑$\Omega\Ket{j}=\Ket{j’}$,系数$\mathcal{M}(\Omega)_{ij}$ 的含义就是 $\ket{j’}$ 在基上的投影分量之一,即 $\langle i|j’\rangle=\bra{i}\Omega\ket{j}$ ,得到一个极其直观的写法:
$$
\mathcal{M}(\Omega)_{ij}=\bra{i}\Omega\ket{j}
$$
此外,$\mathcal{M}(\Omega^\dagger)_{ij}=\bra{j}\Omega\ket{i}^\ast$
Ex 1.7.1
采用Dirac记号之后比线代中的证明看起来更舒服
$$
\sum_i(\Omega\Lambda)_{ii}=\sum_i\bra{i}\Omega\Lambda\ket{i}=\sum_j\sum_i\bra{i}\Omega\ket{j}\bra{j}\Lambda\ket{i}
$$
$$
=\sum_j\sum_i\bra{j}\Lambda\ket{i}\bra{i}\Omega\ket{j}=\sum_j\bra{j}\Lambda\Omega\ket{j}=\sum_j(\Lambda\Omega)_{jj}
$$
P33的中括号即复谱定理。
P36 定理10表述正确但证明有误
证明中都没有用到自伴的条件,证明若正确,对于所有算子都有对角化结论,显然荒谬。
漏洞在选走$\lambda_1$之后其正交补空间不一定仍是$\Omega$的不变子空间,这样$\Omega$不能限制在$U^\perp$上成为$\Omega\vert_{U^\perp}$。
$E(\lambda_i,\Omega)\subseteq G(\lambda_i,\Omega)$,但代数重数是$\text{dim}G(\lambda_i,\Omega)$,例如$\lambda_i$重数为3,但$\text{dim}E(\lambda_i,\Omega)=1$,选走一个本征向量$\ket{\lambda_i}$后,可能再找不到别的本征向量了(剩下两维都是由非本征向量的两个广义本征向量组成),所以该证明递推有问题。
Ex 1.8.8
(1) $(M^i)^2=I$ 作用于 $\forall$本征向量$\ket{v}$,$(\lambda^2-1)\ket{v}=\ket{0}$
(2)$\text{Tr}(M^iM^j)=\text{Tr}(M^jM^i)$,且Tr是线性运算,用齐性
(3) 由于本征值就是$\pm1$,又由于迹$=0$,$1$和$-1$个数相等,所以只能偶数维
Ex 1.8.9
原式即 $\sum\limits_\alpha m_\alpha(\omega r_\alpha^2-r_\alpha(r_\alpha\cdot\omega))$ 代入$M_{ij}$即可
(1)平行$\iff \ket{\omega}$为$M$的本征向量
(2) $M_{ij}=M_{ji}$ (元素都是实的)
(3) 维数3,加之$M$自伴,必有$E_1\oplus E_2\oplus E_3=\mathbb{V}$
(4)本征值全都一样,随意取三个正交的向量都可
P45
对$\Omega$ indifferent原因:该空间不管怎么选三个线性无关限量,$\Omega$作用下都是乘以$\lambda_i$,仍然对角
Ex 1.9.2
定义为良好的—实部虚部均收敛(收敛半径均为无穷),取由本征向量构成的基,对应对角矩阵每个元素都是$e^{i\omega_i}$,易知是等距同构
Ex 1.9.3
取上面的那个基,$\text{det} U=\prod\limits_j e^{i\omega_j}=e^{i\sum\omega_i}=e^{i\text{Tr}(H)}$
P62 顶部
含参变量积分,对于x求导,穿过积分号只对$\delta$函数求导。
更“对偶”的理解参考《重温微积分》
式1.10.22用分部积分法可得
Ex 1.10.1
$$
\int\delta(ax)d(ax)=1=|a|\int\delta(ax)
dx=|a|\int\frac{\delta(x)}{|a|}dx
$$
$$
\int\left(\delta(ax)-\frac{\delta(x)}{|a|}\right)dx=0
$$
对于任意上下限的积分都是,且有连续等条件可得括号中被积函数等于$0$
Ex 1.10.2
于$x_i$处展开
$$
\begin{align*}
f(x)&=f(x_i)+f’(x_i)(x-x_i)+\dots
\newline
\newline
&=f’(x_i)(x-x_i)+\dots
\newline
\newline
\delta(f(x))&=\delta(f’(x_i)(x-x_i)+\dots)
\newline
\newline
&\approx\delta(f’(x_i)(x-x_i))
\newline
\newline
&=\frac{\delta(x-x_i)}{|f’(x_i)|}
\end{align*}
$$
Ex 1.10.3
$$
\int_\Omega \frac{d}{dx}\theta(x-x’)dx=d\theta(x-x’)=
\begin{cases}
1 (x’\in \Omega) \\
0 (x’\notin \Omega) \\
\end{cases}
\newline
\newline
\therefore
\int_\Omega \left(\frac{d}{dx}\theta(x-x’)-\delta(x-x’)\right)dx=0
$$
恒成立
P64 式(1.10.27)省略了$(x)$,(1.10.28)下面的无标号式子也是。可能是作者习惯
P69 $\bra{k}X\ket{k’}$(即在$k$基中$X$的矩阵表示的一个元素)的计算:
$$
\bra{k}X\ket{k’}=\int\bra{k}X\ket{x}\langle x|k’\rangle dx=\int x \langle k|x\rangle \langle x|k’\rangle dx=\dots
$$
Ex 1.10.4
直接用式(1.10.59)
Shankar_ch.2
Ex 2.1.3
表达出速度:
$$
ds=\sqrt{(dr)^2+(rd\theta)^2+(r sin\theta d\phi)^2}
$$
$$
v=\frac{ds}{dt} \quad,\quad T=\frac{1}{2}mv^2\quad\dots
$$
先看一波section 18.4
关于电动力学,默认函数都有足够光滑性,符合含参变量积分求导条件,因此$\nabla$或$\frac{\partial}{\partial t}$或$\int$都可以互相穿过
Ex 18.4.2
$E$与$B$完全不变
Ex 18.4.3
(1)
$$
\frac{\partial\Lambda}{\partial t}=-c\phi(r,t)\quad\therefore \phi ‘=0
$$
(2)
注意“once more”是在$A’$的基础上:
$$
\begin{align*}
\nabla\cdot A’’&=\nabla \cdot A’-\nabla^2\Lambda’
\newline
\newline
&=\nabla\cdot A’+\frac{1}{4\pi}\int\frac{\nabla^2 (\nabla\cdot A’(r’,t))d^3r’}{\vert r-r’\vert}
\newline
\newline
&=\nabla\cdot A’+\frac{1}{4\pi}\int -4\pi\delta^3(r-r’){(\nabla\cdot A’(r’,t))d^3r’}
\newline
\newline
&=\nabla\cdot A’-(\nabla\cdot A’(r,t))=0\qquad
\end{align*}
$$
(注:Recall的内容参见Cohen P558附录二.4)
(3)
$$
\phi’’=\phi’-\frac{1}{c}\frac{1}{4\pi}\int\frac{(\nabla\cdot \frac{\partial A’(r’,t)}{\partial t})d^3r’}{\vert r-r’\vert}
\newline
\newline
\phi’=0
$$
对$A’$情况考虑$\nabla\cdot E=0$,可知
$$
\nabla\cdot\frac{\partial A’}{\partial t}=0
$$故上式为$0$
注:前文提到的free electromagnetic field是$\nabla\cdot E=0$的依据
(4)
结论就是亥姆霍兹定理
在Coulomb Gauge中变换,看$\phi=0=\phi’$即知$\Lambda$不含时,
看$\nabla\cdot A=0=\nabla\cdot A’$即知$\nabla^2\Lambda=0$(调和函数)
独特性:考虑已经在Coulomb Gauge中的某个$A$变换到$A’$过程,运用的$\Lambda$已知是调和函数,且边界条件由$\nabla\cdot\Lambda=A\vert_{\text{边界}}$给定,调和函数是被边界条件唯一确定的(类比唯一性定理)
回到第二章
2.4 notes
Legendre变换:若有关系$u(x)=\frac{df}{dx}$,可定义如下函数:
$$
g(u):=x(u)u-f(x(u))
$$
$$
\therefore \frac{dg}{du}=x(u)
$$
意义:$f$与$g$的互换,达成$x$与$u$地位互换
多变量情形:
若$f(x_1\dots x_n)$,可以将前$j$个$x$换成$u$,其中$u_i=\frac{\partial f}{\partial x_i}$,可定义:
$$
g(u_1\dots u_j,x_{j+1}\dots x_n):=\sum\limits_{i=1}^{j}u_i x_i-f(x_1\dots x_n)
$$
$$
\therefore \frac{\partial g}{\partial u_i}=x_i
$$
由此启发
$$
H(q,p):=\sum\limits_{i=1}^n p_i \dot{q_i}-L(q,\dot{q})
$$
Ex 2.5.1
本题即齐次函数的Euler定理。
$p_i=\frac{\partial L}{\partial\dot{q_i}}$中$L=T-V$,$V$与$\dot{q_i}$无关,只用对$T=\sum_i\sum_j T_{ij}\dot{q_i}\dot{q_j}$偏导:
$$
\frac{\partial T}{\partial\dot{q_k}}=2T_{kk}\dot{q_k}+\sum_r (T_{rk}+T_{kr})\dot{q_r}
$$
应用到所求式:
$$
\sum_i p_i\dot{q_i}=\sum_i \frac{\partial T}{\partial \dot{q_i}}\dot{q_i}=\sum_k 2T_{kk}\dot{q_k}\dot{q_k}+\sum_k\sum_r (T_{rk}+T_{kr})\dot{q_r}\dot{q_k}=2T
$$
2.7 notes
引入泊松括号,一个函数$\omega$对时间的导数即$\{\omega,H\}$
Ex 2.7.1
仅验证第三个
$$
\begin{align*}
\{ \omega,\lambda\sigma\}&=\sum_i[\frac{\partial\omega}{\partial q_i}\frac{(\sigma\partial\lambda+\lambda\partial\sigma)}{\partial p_i}-\frac{\partial\omega}{\partial p_i}\frac{(\sigma\partial\lambda+\lambda\partial\sigma)}{\partial q_i}]
\newline
\newline
&=\sigma\{\omega,\lambda\}+\lambda\{\omega,\sigma\}
\end{align*}
$$
中间写法不规范,但方便看出各项来源
Ex 2.7.2
(i) 仅验证(2.7.5)
$$
\dot{q_i}=
\sum_k \left( \frac{\partial q_i}{\partial q_k}\frac{\partial H}{\partial p_k}-\frac{\partial H}{\partial q_k}\frac{\partial q_i}{\partial p_k}\right)
$$
第二项恒为0,第一项仅i=k时非零
$$
\dot{p_i}=\sum_k \left( \frac{\partial p_i}{\partial q_k}\frac{\partial H}{\partial p_k}-\frac{\partial H}{\partial q_k}\frac{\partial p_i}{\partial p_k}\right)
$$
第一项恒为0,第二项仅i=k时非零
(ii)
$$
\sum_k \left( \frac{\partial l_z}{\partial q_k}\frac{\partial H}{\partial p_k}-\frac{\partial H}{\partial q_k}\frac{\partial l_z}{\partial p_k}\right)
$$
其中k=1,2,3,即$q_k=x,y,z;p_k=p_x,p_y,p_z$
其中$l_z=
\begin{vmatrix}
x & y \\
p_x & p_y \\
\end{vmatrix}
=xp_y-yp_x$,知$k=1,2$时才有贡献,$\therefore =(2p_yp_x+2axy)+(-2p_xp_y-2byx)=0$
Ex 2.7.3
(2.7.15)代入(2.7.14)推(2.7.16),注意(2.7.16)漏了角标:
$$
\dot{\bar{q_j}}=\sum_i \left(\frac{\partial \bar{q_j}}{\partial q_i} \sum_k(\frac{\partial H}{\partial\bar{q_k}}\frac{\partial\bar{q_k}}{\partial p_i}+\frac{\partial H}{\partial\bar{p_k}}\frac{\partial\bar{p_k}}{\partial p_i})-
\frac{\partial \bar{q_j}}{\partial p_i}\sum_k(\frac{\partial H}{\partial\bar{q_k}}\frac{\partial\bar{q_k}}{\partial q_i}+\frac{\partial H}{\partial\bar{p_k}}\frac{\partial\bar{p_k}}{\partial q_i})\right)
$$
Fubini定理换序(有限项求和放心换$\sum_i$与$\sum_k$)
$$
\begin{align*}
&=\sum_k \left[ \frac{\partial H}{\partial\bar{q_k}}\left( \sum_i(\frac{\partial \bar{q_j}}{\partial q_i}\frac{\partial \bar{q_k}}{\partial p_i}-\frac{\partial \bar{q_j}}{\partial p_i}\frac{\partial\bar{q_k}}{\partial q_i} ) \right)+\frac{\partial H}{\partial \bar{p_k}} \left( \sum_i(\frac{\partial}{\partial}\frac{\partial}{\partial}-\frac{\partial}{\partial}\frac{\partial}{\partial})\right)\right]
\newline
\newline
&=\sum_k \left[\frac{\partial H}{\partial \bar{q_k}}\{\bar{q_j},\bar{q_k}\}+\frac{\partial H}{\partial\bar{p_k}}\{\bar{q_j},\bar{p_k}\} \right]
\end{align*}
$$
(2.7.17)式直接把(2.7.16)中$\bar{q_j}$换$\bar{p_j}$即可,地位等价
Ex 2.7.4
例如:
$$
\begin{align*}
\{ \bar{x},\bar{p_x} \}&=\frac{\partial \bar{x}}{\partial x}\frac{\partial \bar{p_x}}{\partial p_x}-\frac{\partial \bar{p_x}}{\partial x}\frac{\partial \bar{x}}{\partial p_x}+\frac{\partial \bar{x}}{\partial y}\frac{\partial \bar{p_x}}{\partial p_y}-\frac{\partial \bar{p_x}}{\partial y}\frac{\partial \bar{x}}{\partial p_y}
\newline
\newline
&=\cos^2\theta+\sin^2\theta=1
\end{align*}
$$
其余同理易证
Ex 2.7.5
书中$p_\rho$表达式符号有误,径向$y$和$x$应当地位等价,符号同。
$$
\{ \rho,\phi\}=\frac{\partial \rho}{\partial x}\frac{\partial \phi}{\partial p_x}-\frac{\partial \phi}{\partial x}\frac{\partial \rho}{\partial p_x}+\frac{\partial \rho}{\partial y}\frac{\partial \phi}{\partial p_y}-\frac{\partial \phi}{\partial y}\frac{\partial \rho}{\partial p_y}
$$
(所有对$p_x$或$p_y$求导都$=0$)$\therefore=0-0+0-0=0$
$$
\{ \rho,p_\rho \}=\frac{\partial \rho}{\partial x}\frac{\partial p_\rho}{\partial p_x}-\frac{\partial p_\rho}{\partial x}\frac{\partial \rho}{\partial p_x}+\frac{\partial \rho}{\partial y}\frac{\partial p_\rho}{\partial p_y}-\frac{\partial p_\rho}{\partial y}\frac{\partial \rho}{\partial p_y}
=\frac{x}{\sqrt{(x^2+y^2)}}\cdot\frac{x}{\sqrt{(x^2+y^2)}}
+\frac{y}{\sqrt{(x^2+y^2)}}\cdot\frac{y}{\sqrt{(x^2+y^2)}}=1
$$
其余同理略
Ex 2.7.7
$$
\{ \bar{q},\bar{p}\}=\frac{\partial \bar{q}}{\partial q}\frac{\partial \bar{p}}{\partial p}-\frac{\partial \bar{p}}{\partial q}\frac{\partial \bar{q}}{\partial p}=\frac{1}{q^{-1}}\frac{-1}{q^{2}}\cdot\frac{-q}{\sin^2p}-\cot p\cdot\frac{1}{\sin p}\cos p
\newline
\newline
=\frac{1}{\sin^2 p}-\frac{\cos^2 p}{\sin^2 p}=1 ~~~~;\dots
$$
Ex 2.7.8
(1) (2) 简单数学计算
(3)
从左边(含bar的视角来看):
$$
\bar{p_i}=\frac{\partial L(\bar{q},\dot{\bar{q}})}{\partial \dot{\bar{q_i}}}=\sum_k \frac{\partial L}{\partial\bar{q_k}}\frac{\partial\bar{q_k}}{\partial\dot{\bar{q_i}}}+\frac{\partial L}{\partial\dot{\bar{q_k}}}\frac{\partial\dot{\bar{q_k}}}{\partial\dot{\bar{q_i}}}
$$(宗量独立性,第一项为$0$)
$$
=\sum_k \frac{\partial L}{\partial\dot{\bar{q_k}}}\frac{\partial\dot{\bar{q_k}}}{\partial\dot{\bar{q_i}}}
$$
(这并没用因为只是在含bar的视角中搞)
有意义的是从最右边(不含bar的视角来看)出发,建立起含bar与不含的桥梁
$$
\bar{p_i}=\frac{\partial L(q,\dot{q})}{\partial \dot{\bar{q_i}}}=\sum_k \frac{\partial L}{\partial q_k}\frac{\partial q_k}{\partial\dot{\bar{q_i}}}+\frac{\partial L}{\partial\dot{q_k}}\frac{\partial\dot{q_k}}{\partial\dot{\bar{q_i}}}
$$
第一项仍为0,因为$q=q(\bar{q})$,不显含$\dot{\bar{q}}$,第二项代入(2)结论
$$
\therefore \bar{p_i}=\sum_k \frac{\partial L}{\partial \dot{q_k}}\frac{\partial q_k}{\partial \bar{q_i}}=\sum_k p_k\frac{\partial q_k}{\partial \bar{q_i}}
$$ 就是(2.7.9)
(4)
$$
\{ \bar{q_i},\bar{p_j}\}=\sum_k \left( \frac{\partial \bar{q_i}}{\partial q_k}\frac{\partial \bar{p_j}}{\partial p_k}-\frac{\partial \bar{p_j}}{\partial q_k}\frac{\partial \bar{q_i}}{\partial p_k}\right)=\sum_k \left( \frac{\partial \bar{q_i}}{\partial q_k}\frac{\partial q_k}{\partial \bar{q_j}}-0\right)=\delta_{ij}
\newline \dots
$$
Ex 2.7.9
$$
\{ \omega,\sigma\}_{\bar{q},\bar{p}}=\sum_k \left( \frac{\partial \omega}{\partial \bar{q_k}}\frac{\partial \sigma}{\partial \bar{p_k}}-\frac{\partial \sigma}{\partial \bar{q_k}}\frac{\partial \omega}{\partial \bar{p_k}} \right)
$$
$$
=\sum_k \left( \frac{\partial \omega}{\partial \bar{q_k}}\frac{\partial \bar{q_k}}{\partial q_k}\frac{\partial \sigma}{\partial p_k}\frac{\partial \bar{p_k}}{\partial p_k}
-\frac{\partial \sigma}{\partial q_k}\frac{\partial \bar{q_k}}{\partial q_k}\frac{\partial \omega}{\partial p_k}\frac{\partial \bar{p_k}}{\partial p_k} \right)
$$
提出公因子再对该因子用(2.7.11):
$$
\frac{\partial \bar{q_k}}{\partial q_k}\frac{\partial \bar{p_k}}{\partial p_k}=\frac{\partial \bar{q_k}}{\partial q_k}\frac{\partial q_k}{\partial \bar{q_k}}=1;
$$
$$
\therefore=\sum_k \left( \frac{\partial \omega}{\partial q_k}\frac{\partial \sigma}{\partial p_k}-\frac{\partial \sigma}{\partial q_k}\frac{\partial \omega}{\partial p_k} \right)=\{ \omega,\sigma\}_{q,p}
$$
Ex 2.8.2
以一个条件为例:(只用在原来不含bar的基础上分析,添加新增的项,并忽略高阶)
$$
\begin{align*}
\{\bar{q_j},\bar{p_m}\}&=\sum_k \left( \frac{\partial ( q_j+\epsilon\frac{\partial g}{\partial p_j})}{\partial q_k}\frac{\partial (p_m-\epsilon\frac{\partial g}{\partial q_m})}{\partial p_k}-\frac{\partial (p_m+\epsilon\frac{\partial g}{\partial q_m})}{\partial q_k}\frac{\partial ( q_j-\epsilon\frac{\partial g}{\partial p_j})}{\partial p_k} \right)
\newline
\newline
&=\delta_{jm}
+
\frac{\partial \epsilon\frac{\partial g}{\partial p_j}}{\partial q_m}
\frac{\partial \epsilon\frac{\partial g}{\partial q_m}}{\partial p_j} \sum_k \frac{\partial \epsilon\frac{\partial g}{\partial p_j}}{\partial q_k}\frac{\partial\epsilon\frac{\partial g}{\partial q_m}}{\partial p_k}-\sum_k \left(\frac{\partial \epsilon\frac{\partial g}{\partial q_m}}{\partial q_k}\frac{\partial\epsilon\frac{\partial g}{\partial p_j}}{\partial p_k} \right)
\newline
\newline
&=\delta_{jm}+o(\epsilon)\approx\delta_{jm}
\end{align*}
$$
Ex 2.8.3
1.non-canonical:此时的变换为:
$$
\bar{x}=x-y\epsilon
\qquad
\bar{y}=y+x\epsilon
\newline
\bar{p_x}=p_x
\qquad
\bar{p_y}=p_y
\newline
$$
下面这项会出问题:
$$
\{\bar{x},\bar{p_y}\}=\left(\frac{\partial (x-y\epsilon)}{\partial x}\frac{\partial p_y}{\partial p_x}-\frac{\partial p_y}{\partial x}\frac{\partial (x-y\epsilon)}{\partial p_x} \right)
\newline
\newline
+\left(\frac{\partial (x-y\epsilon)}{\partial y}\frac{\partial p_y}{\partial p_y}-\frac{\partial p_y}{\partial y}\frac{\partial (x-y\epsilon)}{\partial p_y} \right)=-\epsilon\neq 0
$$
2.不可能写为${H,g}$形式:
假设有一个生成函数$g$,计算
$$
\{H,g\}=m\omega^2(x\frac{\partial g}{\partial p_x}+y\frac{\partial g}{\partial p_y})-\frac{1}{m}(p_x\frac{\partial g}{\partial x}+p_y\frac{\partial g}{\partial y})
$$
由于$\delta p_i=0$(题目假定)$\therefore g$不显含$x,y$,即第二项为零
若$\delta H=\{H,g\}=0$必有$x\frac{\partial g}{\partial p_x}+y\frac{\partial g}{\partial p_y}=0$
由于$\frac{\partial g}{\partial p_i}$也不含x、y,故只有$\frac{\partial g}{\partial p_i}=0$,即g平凡解为常数。
Ex 2.8.4
类似于x与y的旋转
$$
\bar{x}=x-\epsilon p
\newline
\bar{p}=p+\epsilon x
$$
试图让
$$
\frac{\partial g}{\partial p}=-p
\qquad\frac{\partial g}{\partial x}=-x,
$$
取$g=-\frac{1}{2}(x^2+p^2)$即可