Laplace 算子在极坐标下的形式

在二维空间中的函数 \(u:\mathbb{R}^2 \mapsto \mathbb{R}\), 满足 Laplace 方程: \[ -\Delta u = 0 \] 式中 \(\Delta\) 为 Laplace 算子, 在直角坐标下为 \[ \Delta = \frac{\partial^2~}{\partial x^2} + \frac{\partial^2~}{\partial y^2} \tag{D} \] 在极坐标下写作 \[ \Delta = \frac{\partial^2~}{\partial r^2} + \frac{1}{r} \frac{\partial~}{\partial r} + \frac{1}{r^2} \frac{\partial^2~}{\partial \theta^2} \tag{P} \] 这两种坐标形式下的 Laplace 算子表示可以通过坐标变换 \[ x = r\cos \theta, \quad y = r \sin \theta \] 或者 \[ r = \sqrt{x^2 + y^2}, \quad \theta = \arctan \frac{y}{x} \] 得到, 但是我现在不想这样做, 因为我已经这么做了很多次了, 并且每一次推导都是 “全新” 的体验. 我现在想做的是怎么样才能快速记住极坐标变换的形式.

首先可以观察方程的量纲, 可以验证, 式 (P) 与式 (D) 在量纲上是一致的. 但是这也仅仅限于验证, 并不能帮助我得到极坐标表达式的形式.

如果尝试将一阶微分写成矩阵的形式, 那么就有: \[ \begin{pmatrix} \frac{\partial~}{\partial x} \\ \frac{\partial~}{\partial y} \end{pmatrix} = R(\theta) \begin{pmatrix} \frac{\partial~}{\partial r} \\ \frac{1}{r} \frac{\partial~}{\partial \theta} \end{pmatrix} \] 式中, \(R(\theta)\) 为旋转矩阵 \[ R(\theta) \triangleq \begin{pmatrix} \cos\theta & -\sin\theta \\ \sin\theta & \cos\theta \end{pmatrix} \] 但是, 无法再继续得到用矩阵表示的二阶偏导数形式, 这是因为算子的系数是非线性的. 如果是线性的, 那么极坐标表达式 (P) 就不会出现关于 \(r\) 的一次偏导数.

那么还可以继续怎么做下去呢? 最终我在 Math stack Exchange 上找到了这一篇回答: How to remember laplacian in polar and (hyper)spherical coordinates

设直角坐标下的函数形式为 \(u(x,y)\), 考虑点 \(P\)\((x,0)\) 的特殊情况, 此时 \(r \equiv x\), 所以 \(u_{xx} \equiv u_{rr}\). 现在我们要考虑在圆周曲线上的导数与垂直于 \(x\) 轴直线的导数的区别. \[ u_{\theta}(\theta) = u(r\cos\theta, r\sin\theta) \] \(U_{\theta}\)\(\theta\) 的一阶导数为 \[ \frac{\mathrm{d} u_{\theta}}{\mathrm{d} \theta} = u_x (-r\sin\theta)|_{\theta=0} + u_{y}(r\cos\theta)|_{\theta=0} = r u_y \] 这是很合理的, 因为在垂直于 \(x\) 轴方向上有微元关系 \(r~\mathrm{d} \theta = \mathrm{d}y\).

接下来要求二阶导数, 注意到不能直接对上式代入数值之后的关系式 \({\mathrm{d} u_{\theta}}/{\mathrm{d} \theta}= r u_y\) 关于 \(\theta\) 求导得到 \({\mathrm{d}^2 u_{\theta}}/{\mathrm{d} \theta^2}= r^2 u_{yy}\), 因为 \({\mathrm{d} u_{\theta}}/{\mathrm{d} \theta}= 0\) 的项中仍可能对二阶导数项有贡献. 所以 \[ \left.\frac{\mathrm{d}^2 u_{\theta}}{\mathrm{d} \theta^2}\right|_{\theta=0}= -u_x r + u_{yy}r^2 \] 注意到 \(u_x \equiv u_r\), 所以 \(u_{yy}\) 可以表示为 \[ u_{yy} = \frac{1}{r}u_r + \frac{1}{r^2}u_{\theta\theta} \] 这就通过特殊位置的点, 得到极坐标的形式 (P)

还可以继续推广到高维情景