Lagrange方程为保守系(系统中力的做功与粒子运动的路径无关,如重力、静电力)的动力学方程,其表达式为 \frac{\mathscr{d}}{\mathscr{d} t}\left( \frac{\partial \rm{L}}{\partial\dot{q}_\alpha} \right)-\frac{\partial\rm{L}}{\partial q_\alpha}=0\\ 其中 \rm{L}=\it{T}-V 为Lagrange函数, T 为系统动能, V 为系统势能; q_\alpha 为广义坐标, \dot{q}_\alpha 为广义速度。 下面是推导过程:根据Newton运动方程,有 \boldsymbol{F}_i+\boldsymbol{R}_i-m_i\boldsymbol{\ddot{r}_i}=0\tag{1} 其中, \boldsymbol{F}_i 为主动力, \boldsymbol{R}_i 为被动力。因此我们可以得到 \sum_{i=1}^{N}{\left( \boldsymbol{F}_i+\boldsymbol{R}_i-m_i\boldsymbol{\ddot{r}_i} \right)}\cdot\delta\boldsymbol{r}_i=0\tag{2} 其中 \delta\boldsymbol{r}_i 为第 i 个粒子的虚位移。在理想约束条件下,约束力与粒子位移的方向垂直,因此 \boldsymbol{R}_i\cdot\delta\boldsymbol{r}_i=0 ,所以(2)可化为 \sum_{i=1}^{N}{\left( \boldsymbol{F}_i-m_i\boldsymbol{\ddot{r}_i} \right)}\cdot\delta\boldsymbol{r}_i=0\tag{3} 由于约束的存在,各个粒子的虚位移并不彼此独立,因此我们无法推断 \boldsymbol{F}_i-m_i\boldsymbol{\ddot{r}_i}=0 。由于 \delta \boldsymbol{r}_i=\sum_{\alpha=1}^{s}{\frac{\partial \boldsymbol{r}_i}{\partial q_\alpha}}\delta q_\alpha ,代入(3)式可得 \sum_{i=1}^{N}{\left( \boldsymbol{F}_i-m_i\boldsymbol{\ddot{r}_i} \right)}\cdot\sum_{\alpha=1}^{s}{\frac{\partial \boldsymbol{r}_i}{\partial q_\alpha}}\delta q_\alpha=0\tag{4} 由于各个广义坐标的虚变更 \delta q_\alpha 彼此独立,并且 \alpha 与 i 不相关,因此可以交换(4)中的求和顺序 \sum_{\alpha=1}^{s}{\left\{ \sum_{i=1}^{N}{\left[ \left( \boldsymbol{F}_i-m_i\boldsymbol{\ddot{r}_i} \right)\cdot\frac{\partial\boldsymbol{r}_i}{\partial q_\alpha} \right]} \right\}}\delta q_\alpha=0\tag{5} \sum_{\alpha=1}^{s}{\left[ \color{red}{\sum_{i=1}^{N}{\left( \boldsymbol{F}_i\cdot\frac{\partial \boldsymbol{r}_i}{\partial q_\alpha} \right)}}-\color{blue}{\sum_{i=1}^{N}{\left( m_i\boldsymbol{\ddot{r}_i}\cdot\frac{\partial\boldsymbol{r}_i}{\partial q_\alpha} \right)} }\right]}\delta q_\alpha=0\tag{6} 我们可以看到,公式(6)中,红色的部分与动力学量相关,蓝色的部分与运动学量相关。为了简化公式,我们定义 Q_\alpha\equiv\sum_{i=1}^{N}{\left( \boldsymbol{F}_i\cdot\frac{\partial \boldsymbol{r}_i}{\partial q_\alpha} \right)}, \quad \alpha=1,2,...,s \tag{7} P_\alpha\equiv\sum_{i=1}^{N}{\left( m_i\boldsymbol{\ddot{r}}_i\cdot\frac{\partial \boldsymbol{r}_i}{\partial q_\alpha} \right)}, \quad \alpha=1,2,...,s \tag{8} 由于 \delta q_\alpha 彼此独立,因此式(6)中中括号内的部分恒等于0,即 P_\alpha-Q_\alpha=0 \tag{9} 我们首先分析 P_\alpha 。根据求导法则 \left( fg \right)'=f'g+fg' ,我们可得 P_\alpha=\frac{\rm{d}}{\rm{d} \it{t}}\left[ \sum_{i-1}^{N}{\left( m_i\boldsymbol{\dot{r}_i}\cdot\frac{\partial\boldsymbol{r}_i}{\partial q_\alpha} \right)} \right]-\sum_{i=1}^{N}{\left[ m_i\boldsymbol{\dot{r}} \cdot\frac{\rm{d}}{\rm{d}\it{t}}\left( \frac{\partial \boldsymbol{r}_i}{\partial q_\alpha} \right) \right]} \tag{10} 其中,在第二项中 \begin{align} \frac{\rm{d}}{\rm{d}\it{t}}\left( \frac{\partial \boldsymbol{r}_i}{\partial q_\alpha} \right)&=\sum_{\beta=1}^{s}{\left( \frac{\partial^2\boldsymbol{r}_i}{\partial q_\beta\partial q_\alpha}\dot{q}_\beta \right)}+\frac{\partial^2\boldsymbol{r}_i}{\partial t\partial q_\alpha}\\ &=\frac{\partial}{\partial q_\alpha}\left[ \sum_{\beta=1}^{s}{\left( \frac{\partial\boldsymbol{r}_i}{\partial q_\beta}\dot{q}_\beta \right)}+\frac{\partial\boldsymbol{r}_i}{\partial t}\right]\\ &=\frac{\partial \boldsymbol{\dot{r}_i}}{\partial q_\alpha} \end{align}\tag{11} 则(10)式中,第二项可化为 \sum_{i=1}^{N}{\left[ m_i\boldsymbol{\dot{r}} \cdot\frac{\partial \boldsymbol{\dot{r}_i}}{\partial q_\alpha} \right]} ,根据 \frac{\partial \boldsymbol{A}^2}{x}=2\boldsymbol{A}\cdot\frac{\partial\boldsymbol{A}}{\partial x} ,第二项可进一步简化为 \sum_{i=1}^{N}{\left[ m_i\boldsymbol{\dot{r}} \cdot\frac{\partial \boldsymbol{\dot{r}_i}}{\partial q_\alpha} \right]}=\sum_{i=1}^{N}{\left( \frac{1}{2}m_i\frac{\partial\left( \boldsymbol{\dot{r}}_i \right)^2}{\partial q_\alpha} \right)}=\frac{\partial T}{\partial q_\alpha} \tag{12} 其中, T 为系统的动能。 接下来我们简化第一项。由于 \boldsymbol{\dot{r}}_i=\frac{\rm{d}\boldsymbol{r}_i}{\rm{d}\it{t}}=\sum_{\alpha=1}^{s}{\frac{\partial \boldsymbol{r}_i}{\partial q_\alpha}}\dot{q}_\alpha+\frac{\partial \boldsymbol{r}_i}{\partial t}\tag{13} 将等式两边两边同时对 \dot{q}_\alpha 求导可得 \frac{\partial \boldsymbol{\dot{r}}}{\partial \dot{q}_\alpha}=\sum_{\alpha=1}^{s}{\frac{\partial}{\partial \dot{q}_\alpha}\left( \frac{\partial \boldsymbol{r}_i}{\partial q_\alpha} \right)\dot{q}_\alpha}+\frac{\partial \boldsymbol{r}_i}{\partial q_\alpha}+ \frac{\partial}{\partial \dot{q}_\alpha}\left( \frac{\partial \boldsymbol{r}_i}{\partial t} \right)\tag{14} 由于位矢对广义坐标的偏导数与位矢对时间的偏导数均与广义速度无关,因此 \frac{\partial}{\partial \dot{q}_\alpha}\left( \frac{\partial \boldsymbol{r}_i}{\partial q_\alpha} \right)=0 , \frac{\partial}{\partial \dot{q}_\alpha}\left( \frac{\partial \boldsymbol{r}_i}{\partial t} \right)=0 。代入式(14)中可得 \frac{\partial \boldsymbol{\dot{r}}}{\partial \dot{q}_\alpha}=\frac{\partial \boldsymbol{r}_i}{\partial q_\alpha}\tag{15} 因此,式(10)中的第一项可改写为 \begin{align} \frac{\rm{d}}{\rm{d} \it{t}}\left[ \sum_{i-1}^{N}{\left( m_i\boldsymbol{\dot{r}_i}\cdot\frac{\partial\boldsymbol{r}_i}{\partial q_\alpha} \right)} \right]&=\frac{\rm{d}}{\rm{d} \it{t}}\left[ \sum_{i-1}^{N}{\left( m_i\boldsymbol{\dot{r}_i}\cdot\frac{\partial \boldsymbol{\dot{r}}}{\partial \dot{q}_\alpha} \right)} \right]\\ &=\frac{\rm{d}}{\rm{d}\it{t}}\left\{ \sum_{i=1}^{N}{\frac{\partial}{\partial\dot{q}_\alpha}\left[ \frac{1}{2}m_i\left( \boldsymbol{\dot{r}_i} \right)^2 \right]}\right\} \\ &=\frac{\rm{d}}{\rm{d}\it{t}}\left( \frac{\partial T}{\partial\dot{q}_\alpha} \right) \end{align}\tag{16} 因此 P_\alpha=\frac{\rm{d}}{\rm{d}\it{t}}\left( \frac{\partial T}{\partial\dot{q}_\alpha} \right)-\frac{\partial T}{\partial q_\alpha}\tag{17} 我们再来分析 Q_\alpha 。 Q_\alpha 本身难以化简,但是我们令其与 \delta q_\alpha 相乘可得到 Q_\alpha\delta q_\alpha=\sum_{i=1}^{N}{\boldsymbol{F}_i\cdot\frac{\partial \boldsymbol{r}_i}{\partial q_\alpha}\delta q_\alpha}=\sum_{i=1}^{N}{\boldsymbol{F}_i\cdot\delta \boldsymbol{r}_{i\alpha}}\tag{18} 即 Q_\alpha\delta q_\alpha 为主动力 \boldsymbol{F}_i 在体系中,对所有粒子在第 \alpha 个广义坐标方向上的虚位移所做的虚功之和。由于功为作用力与位移的点积,因此我们称 Q_\alpha 为广义力。若广义坐标取直角坐标,则 \begin{cases} Q_x=\boldsymbol{F}\cdot\frac{\partial \boldsymbol{r}}{x}=\boldsymbol{F}\cdot\boldsymbol{e}_x=F_x\\ Q_y=\boldsymbol{F}\cdot\frac{\partial \boldsymbol{r}}{y}=\boldsymbol{F}\cdot\boldsymbol{e}_y=F_y\\ Q_z=\boldsymbol{F}\cdot\frac{\partial \boldsymbol{r}}{z}=\boldsymbol{F}\cdot\boldsymbol{e}_z=F_z \end{cases}\tag{19} 即 Q_x, Q_y, Q_z 分别为主动力\boldsymbol{F}_i 在坐标 x,y,z 三个方向的分量。 此时我们可以得到基本形式的Lagrange方程: \frac{\rm{d}}{\rm{d}\it{t}}\left( \frac{\partial T}{\partial\dot{q}_\alpha} \right)-\frac{\partial T}{\partial q_\alpha}=Q_\alpha\tag{20} 在保守系中,力的大小与方向仅与粒子的位置有关,与势能的梯度互为相反矢量,即 \boldsymbol{F}_i=-\boldsymbol{\nabla}_i V,\quad i=1,2,...N\tag{21} 其中, \nabla_i 为梯度算子,其定义为 \boldsymbol{\nabla}_i=\boldsymbol{e}_x\frac{\partial}{\partial x_i}+\boldsymbol{e}_y\frac{\partial}{\partial y_i}+\boldsymbol{e}_z\frac{\partial}{\partial z_i}, \quad i=1,2,...,N\tag{22} 则 F_{ix}=-\frac{\partial V}{\partial x_i},F_{iy}=-\frac{\partial V}{\partial y_i},F_{iz}=-\frac{\partial V}{\partial z_i},\quad i=1,2,...,N\tag{23} 由于 \frac{\partial \boldsymbol{r}_i}{\partial q_\alpha}=\frac{\partial x_i}{\partial q_\alpha}\boldsymbol{e}_x+\frac{\partial y_i}{\partial q_\alpha}\boldsymbol{e}_y+\frac{\partial z_i}{\partial q_\alpha}\boldsymbol{e}_z,\quad i=1,2,...,N\tag{24} 可推导 \begin{align} Q_\alpha &=\sum_{i=1}^{N}{\boldsymbol{F}_i\cdot\frac{\partial \boldsymbol{r}_i}{\partial q_\alpha}}\\ &=\sum_{i=1}^{N}{\left( F_{ix}\frac{\partial x_i}{\partial q_\alpha}+F_{iy}\frac{\partial y_i}{\partial q_\alpha}+F_{iz}\frac{\partial z_i}{\partial q_\alpha} \right)}\\ &=\sum_{i=1}^{N}{\left( -\frac{\partial V}{\partial x_i}\frac{\partial x_i}{\partial q_\alpha}-\frac{\partial V}{\partial y_i}\frac{\partial y_i}{\partial q_\alpha}-\frac{\partial V}{\partial z_i}\frac{\partial z_i}{\partial q_\alpha} \right)}\\ &=-\frac{\partial V}{\partial q_\alpha} \end{align}\tag{25} 此时基本形式的Lagrange方程可改写为 \frac{\rm{d}}{\rm{d}\it{t}}\left( \frac{\partial T}{\partial\dot{q}_\alpha} \right)-\frac{\partial T-V}{\partial q_\alpha}=0\tag{27} 由于势能 V 与广义速度无关,所以 \frac{\partial T}{\partial \dot{q}_\alpha}=\frac{\partial \left( T-V \right)}{\partial \dot{q}_\alpha}\tag{28} 定义 \begin{align} \rm{L}\equiv\it{T}-V&=\sum_{i=1}^{N}{\frac{1}{2}m_i\left( \boldsymbol{\dot{r}}_i \right)^2}-V\left( \boldsymbol{r}_1,\boldsymbol{r}_2,...,\boldsymbol{r}_N \right)\\ &=\rm{L}\left( \it{q_1},q_2,...,q_s; \dot{q}_1,\dot{q}_2,...,\dot{q}_s; t \right) \end{align}\tag{29} 此时,我们可以得到Lagrange方程的最终形式 \frac{\mathscr{d}}{\mathscr{d} t}\left( \frac{\partial \rm{L}}{\partial\dot{q}_\alpha} \right)-\frac{\partial\rm{L}}{\partial q_\alpha}=0\tag{30} 证毕。 疑惑点(已解决,可参考评论区):在推导公式(14)->(15)的过程需要用到一个条件“位矢对广义坐标的偏导数与位矢对时间的偏导数均与广义速度无关“,即 \frac{\partial}{\partial \dot{q}_\alpha}\left( \frac{\partial \boldsymbol{r}_i}{\partial q_\alpha} \right)=0 , \frac{\partial}{\partial \dot{q}_\alpha}\left( \frac{\partial \boldsymbol{r}_i}{\partial t} \right)=0 。然而对于这个公式不知道如何证明,或者物理意义上应该如何理解,希望能得到知友的解答。 笔者随笔:本文的证明过程参考卢文发教授编写的《量子力学与统计力学》,教授从高等数学的角度出发证明了较为复杂的物理公式,简单易懂。其实在证明过程中,很多公式不仅仅有数学层面的意义,也有物理层面的意义。例如在公式(25)中,我们可以得到在保守系中,广义力在某一广义坐标上的分量等于势能对这个广义坐标偏导的负数,而势能对广义坐标的偏导数可以理解为势能在这个坐标方向单位距离增长的速率。所以还是推荐大家独立推导一遍公式,思考每一步的物理意义,可以对公式有更深的理解。 参考文献 卢文发.量子力学与统计力学. 上海交通大学出版社, 2013. |