关于隐式积分方程的一些问题

牛顿运动方程基础就是牛二运动定律
\[ {\bf F} = {\bf M}\ddot{\bf x} \]
将此二阶方程写为两个一阶方程组
\[ \begin{cases} \dot{\bf v} &= {\bf FM}^{-1} \\ \dot{\bf x} &= \bf v \end{cases} \]
其隐式表达式为
\[ \begin{cases} {\bf v}^{t+h} &= {\bf v}^t + {\bf F}^{t+h}{\bf M}^{-1}h \\ {\bf x}^{t+h} &= {\bf x}^t + {\bf v}^{t+h}h \end{cases} \]
利用泰勒级数对\(\bf F(x,v)\)进行一阶展开，
\[ \begin{align} {\bf F}^{t+h} &= {\bf F}^t + \frac{{\bf F}'}{1!}({\bf x}^{t+h}-{\bf x}^t,{\bf v}^{t+h}-{\bf v}^t) \\ &= {\bf F}^t + \frac{\partial \bf F}{\partial \bf x}\Delta {\bf x} + \frac{\partial \bf F}{\partial \bf v}\Delta {\bf v} \end{align} \]
其中\(\Delta {\bf x} = {\bf x}^{t+h} - {\bf x}^{t}\)， \(\Delta {\bf v} = {\bf v}^{t+h} - {\bf v}^t\)
带入上面的方程组，整理，可以得到下面的欧拉隐式方程
\[ ({\bf M} - \frac{\partial{\bf F}}{\partial \bf v}h - \frac{\partial{\bf F}}{\partial \bf x}h^2)\Delta {\bf v} = ({\bf F} + \frac{\partial \bf F}{\partial \bf x}{\bf v}h)h \]

(配图的面的顺序，弄反了，右手系下X2与X3的位置应该换一下。。。)
在布料系统中，很多受力与速度\(\bf v\)无关，所以问题通常集中在求解\(\frac{\partial{\bf F}}{\partial \bf x}\)上。假设有如下4个点\({X_0,X_1,X_2,X_3}\)，其中\(X_1\),\(X_2\),\(X_3\)组成三角形，\(X_0\)在三角形上的投影为\(P = \alpha{_1}X_1+\alpha{_2}X_2 + \alpha{_3}X_3\)。它们的受力为斥力，有
\[ \begin{aligned} F_{X_0}&= f_{\tt stiff} \cdot (X_0 - P) = f \cdot (X_0 - (\alpha_1X_1+\alpha_2X_2+\alpha_3X_3))\\ F_{X_1}&= -\alpha_1 F_{X_0} \\ F_{X_2}&= -\alpha_2 F_{X_0} \\ F_{X_3}&= -\alpha_3 F_{X_0} \end{aligned} \]
，\(f_{\tt stiff}\)为系数。
那么，在求\(\frac{\partial{\bf F}}{\partial \bf x}\)的时候，就要需要对\(F_{X_0}\)，\(F_{X_1}\)，\(F_{X_2}\)，\(F_{X_3}\)，分别对\({X_0,X_1,X_2,X_3}\)进行求解。
对于点\(X_0\)来说，
\[ \begin{aligned} \frac{\partial F}{\partial X_0} &= (\frac{\partial F_{X_0}}{\partial X_0}, \frac{\partial F_{X_1}}{\partial X_0},\frac{\partial F_{X_2}}{\partial X_0},\frac{\partial F_{X_3}}{\partial X_0}) \\ &= (\frac{\partial F_{X_0}}{\partial X_0}, -\alpha_1\frac{\partial F_{X_0}}{\partial X_0},-\alpha_2\frac{\partial F_{X_0}}{\partial X_0}, -\alpha_3\frac{\partial F_{X_0}}{\partial X_0}) \end{aligned} \]
对其他三个点，情形类似。
\[ \begin{aligned} \frac{\partial F}{\partial X_1} &= (\frac{\partial F_{X_0}}{\partial X_1}, \frac{\partial F_{X_1}}{\partial X_1},\frac{\partial F_{X_2}}{\partial X_1},\frac{\partial F_{X_3}}{\partial X_1}) \\ &= (\frac{\partial F_{X_0}}{\partial X_1}, -\alpha_1\frac{\partial F_{X_0}}{\partial X_1},-\alpha_2\frac{\partial F_{X_0}}{\partial X_1}, -\alpha_3\frac{\partial F_{X_0}}{\partial X_1}) \\ \\ \frac{\partial F}{\partial X_2} &= (\frac{\partial F_{X_0}}{\partial X_2},- \alpha_1\frac{\partial F_{X_0}}{\partial X_2},-\alpha_2\frac{\partial F_{X_0}}{\partial X_2}, -\alpha_3\frac{\partial F_{X_0}}{\partial X_2}) \\ \\ \frac{\partial F}{\partial X_3} &= (\frac{\partial F_{X_0}}{\partial X_3},-\alpha_1\frac{\partial F_{X_0}}{\partial X_3},-\alpha_2\frac{\partial F_{X_0}}{\partial X_3}, -\alpha_3\frac{\partial F_{X_0}}{\partial X_3}) \\ \end{aligned} \]

至于求\(\frac{\partial F_{X_0}}{\partial X_i}\)，按照雅可比公式来求就可以了
\[ \frac{\partial{\bf F}}{\partial \bf x} = \begin{bmatrix} \frac{\partial f_x}{\partial \bf x} \\ \frac{\partial f_y}{\partial \bf x} \\ \frac{\partial f_z}{\partial \bf x} \end{bmatrix} = \begin{bmatrix} \frac{\partial f_x}{\partial {\bf x}_x} \ \frac{\partial f_x}{\partial {\bf x}_y} \ \frac{\partial f_x}{\partial {\bf x}_z} \\ \frac{\partial f_y}{\partial {\bf x}_x} \ \frac{\partial f_y}{\partial {\bf x}_y} \ \frac{\partial f_y}{\partial {\bf x}_z} \\ \frac{\partial f_z}{\partial {\bf x}_x} \ \frac{\partial f_z}{\partial {\bf x}_y} \ \frac{\partial f_z}{\partial {\bf x}_z} \end{bmatrix} \]

求解：
\[ \begin{aligned} \frac{\partial F_{X_0}}{\partial X_0} &= \begin{bmatrix} &f_{\tt stiff} \ &0 \ &0 \\ &0 \ &f_{\tt stiff} \ &0 \\ &0 \ &0 \ &f_{\tt stiff} \\ \end{bmatrix} \\ \frac{\partial F_{X_0}}{\partial X_1} &= -\alpha_1 \cdot \frac{\partial F_{X_0}}{\partial X_0} \\ \frac{\partial F_{X_0}}{\partial X_2} &= -\alpha_2 \cdot \frac{\partial F_{X_0}}{\partial X_0} \\ \frac{\partial F_{X_0}}{\partial X_3} &= -\alpha_3 \cdot \frac{\partial F_{X_0}}{\partial X_0} \end{aligned} \]

以上。

转载于:https://www.cnblogs.com/daiday/p/8918283.html