函数矩阵对矩阵求导

函数对矩阵以及函数矩阵对矩阵求导，我理解主要就是一种简化的写法，用矩阵将多个多元函数对每个元求导写成矩阵的形式，看起来比较简洁。

函数对矩阵的导数

设矩阵$\mathbf{X}=({\xi }_{ij}{)}_{m\times n}$，$mn$元函数$f(\mathbf{X})=f({\xi }_{11},{\xi }_{12},{\xi }_{13},\dots ,{\xi }_{m1},\dots ,{\xi }_{mn})$，则$f(\mathbf{X})$对矩阵$\mathbf{X}$的导数为，

$$\frac{df}{d\mathbf{X}}={\left(\frac{\partial f}{\partial {\xi }_{ij}}\right)}_{m\times n}=\left[\begin{array}{cccc}\frac{\partial f}{\partial {\xi }_{11}}& \frac{\partial f}{\partial {\xi }_{12}}& \dots & \frac{\partial f}{\partial {\xi }_{1n}}\\ ⋮& ⋮& & ⋮\\ \frac{\partial f}{\partial {\xi }_{m1}}& \frac{\partial f}{\partial {\xi }_{m2}}& \dots & \frac{\partial f}{\partial {\xi }_{mn}}\end{array}\right]$$

例如，$\mathbf{X}=({\xi }_1,{\xi }_2,\dots ,{\xi }_n{)}^T$，$n$元函数$f(\mathbf{X})=f({\xi }_1,{\xi }_2,\dots ,{\xi }_n)$，
则，
$$\frac{df}{d\mathbf{X}}=(\frac{df}{d{\xi }_1},\frac{df}{d{\xi }_2},\dots ,\frac{df}{{\xi }_n}{)}^T$$
又有，
$$\frac{df}{d{\mathbf{X}}^T}=(\frac{df}{d{\xi }_1},\frac{df}{d{\xi }_2},\dots ,\frac{df}{{\xi }_n})$$

函数矩阵对矩阵的导数

设矩阵$\mathbf{X}=({\xi }_{ij}{)}_{m\times n}$，$mn$元函数$f_{ij}(\mathbf{X})=f_{ij}({\xi }_{11},{\xi }_{12},{\xi }_{13},\dots ,{\xi }_{m1},\dots ,{\xi }_{mn})(i=1,2,3\dots ,r;j=1,2,\dots ,s)$,定义函数矩阵，
$$\mathbf{F}(\mathbf{X})=\left[\begin{array}{ccc}{f}_{11}(\mathbf{X})& \dots & f_{1s}(\mathbf{X})\\ ⋮& ⋮& ⋮\\ f_{r1}(\mathbf{X})& \dots & f_{rs}(\mathbf{X})\end{array}\right]$$
对矩阵$\mathbf{X}$的导数为，
$$\frac{d\mathbf{F}}{d\mathbf{X}}=\left[\begin{array}{cccc}\frac{\partial \mathbf{F}}{\partial {\xi }_{11}}& \frac{\partial \mathbf{F}}{\partial {\xi }_{12}}& \dots & \frac{\partial \mathbf{F}}{\partial {\xi }_{1n}}\\ ⋮& ⋮& & ⋮\\ \frac{\partial \mathbf{F}}{\partial {\xi }_{m1}}& \frac{\partial \mathbf{F}}{\partial {\xi }_{m2}}& \dots & \frac{\partial \mathbf{F}}{\partial {\xi }_{mn}}\end{array}\right]$$
其中，
$$\frac{\partial \mathbf{F}}{\partial {\xi }_{ij}}=\left[\begin{array}{cccc}\frac{\partial f_{11}}{\partial {\xi }_{ij}}& \frac{\partial f_{12}}{\partial {\xi }_{ij}}& \dots & \frac{\partial f_{1s}}{\partial {\xi }_{ij}}\\ ⋮& ⋮& & ⋮\\ \frac{\partial f_{r1}}{\partial {\xi }_{ij}}& \frac{\partial f_{r1}}{\partial {\xi }_{ij}}& \dots & \frac{\partial f_{rs}}{\partial {\xi }_{ij}}\end{array}\right]$$

例如，$\mathbf{X}=({\xi }_1,{\xi }_2,\dots ,{\xi }_n{)}^T$，$n$元函数$f(\mathbf{X})=f({\xi }_1,{\xi }_2,\dots ,{\xi }_n)$，
则，
$$\frac{df}{d\mathbf{X}}=(\frac{df}{d{\xi }_1},\frac{df}{d{\xi }_2},\dots ,\frac{df}{{\xi }_n}{)}^T$$
因此，
$$\frac{d}{d{\mathbf{X}}^T}\left(\frac{df}{d\mathbf{X}}\right)=\left[\begin{array}{cccc}\frac{{\partial }^{2}f}{\partial {\xi }_1^2}& \frac{{\partial }^{2}f}{\partial {\xi }_1\partial {\xi }_2}& \cdots & \frac{{\partial }^{2}f}{\partial {\xi }_1\partial {\xi }_n}\\ \frac{{\partial }^{2}f}{\partial {\xi }_2\partial {\xi }_1}& \frac{{\partial }^{2}f}{{\partial {\xi }_2}^2}& \cdots & \frac{{\partial }^{2}f}{\partial {\xi }_2\partial {\xi }_n}\\ ⋮& ⋮& ⋮& ⋮\\ \frac{{\partial }^{2}f}{\partial {\xi }_n\partial {\xi }_1}& \frac{{\partial }^{2}f}{\partial {\xi }_n\partial {\xi }_2}& \cdots & \frac{{\partial }^{2}f}{\partial {{\xi }_n}^2}\end{array}\right]$$