Variational Calculus

1. What is variational calculus?
To find a stationary function of a functional $I[f]$ (function of functions), solve (usually differential) equation for stationary function $f(x)$.
(e.x.) We have two points $A(x_1,y_1),B(x_2,y_2)$, find a path $f(x)$ such that distance between $A$ and $B$ is minimized.
$$\begin{array}{rl}I& ={\int }_A^{B}dS\\ & ={\int }_A^B\sqrt{d{x}^2+d{y}^2}\\ & ={\int }_{x_1}^{x_2}\sqrt{1+(\frac{dy}{dx}{)}^2}dx\end{array}$$

(Brachistochrone Problem) Given $v(x,y)$ of a particle, find a path $y=f(x)$ such that the time $T$ taken by the particle is minimized.
$$T={\int }_{x_1}^{x_2}\frac{\sqrt{1+(dy/dx{)}^2}}{v(x,y)}dx$$

In general, calculus of variations seeks to find $y=f(x)$ such that this integral:
$$I(f)={\int }_{x_1}^{x_2}F(x,y,\frac{dy}{dx})dx$$
is stationary.

2. Euler-Lagrange Equations

Find $y=f(x)$ such that the functional $$I={\int }_{x_1}^{x_2}F(x,y,y^{\prime })dx$$ is stationary. Boundary conditions: $$y(x_1)=y_1,y(x_2)=y_2$$

Derivation: Suppose $y(x)$ (a.k.a. Extremal) makes $I$ stationary and satisfies the above boundary conditions.

• (Implicit: all the functions talked below has $2^{nd}$ derivative)
• Introduce a function $\eta (x)$, which means the small change applied to $y(x)$, the only condition on $\eta (\cdot )$ is $\eta (x_1)=\eta (x_2)=0$ to satisfy the boundary condition. At other locations between $(x_1,x_2)$, $\eta (\cdot )$ is arbitrary function
• Define $\bar{y}(x)=y(x)+\epsilon \eta (x)$
• Then the problems was converted to
  $$I(\epsilon )={\int }_{x_1}^{x_2}F(x,\bar{y},\overline{y^{\prime }})dx$$
  that’s just a regular single variable integration, set $$\frac{dI}{d\epsilon }{|}_{\epsilon =0}=0$$
  Why $\epsilon =0$? Because only when $\epsilon =0\to \bar{y}=y$, which is the extremal, can $I(\epsilon )$ get to stationary point.
• Now we have
  $$\begin{array}{rl}\frac{d}{d\epsilon }{|}_{\epsilon =0}{\int }_{x_1}^{x_2}F(x,\bar{y},{\bar{y}}^{\prime })dx& =0\\ {\int }_{x_1}^{x_2}\frac{\partial }{\partial \epsilon }F(x,\bar{y},{\bar{y}}^{\prime }){|}_{\epsilon =0}dx& =0\\ {\int }_{x_1}^{x_2}[\frac{\partial F}{\partial \bar{y}}\frac{\partial \bar{y}}{\partial \epsilon }+\frac{\partial F}{\partial {\bar{y}}^{\prime }}\frac{\partial {\bar{y}}^{\prime }}{\partial \epsilon }]{|}_{\epsilon =0}dx& =0\end{array}$$

Because
$$\frac{\partial \bar{y}}{\partial \epsilon }=\eta$$ $$\frac{\partial {\bar{y}}^{\prime }}{\partial \epsilon }={\eta }^{\prime }$$

We can have
$$\begin{array}{rl}{\int }_{x_1}^{x_2}[\frac{\partial F}{\partial \bar{y}}\eta +\frac{\partial F}{\partial {\bar{y}}^{\prime }}{\eta }^{\prime }]{|}_{\epsilon =0}dx& =0\end{array}$$

Because
$$\begin{array}{rl}& {\int }_{x_1}^{x_2}\frac{\partial F}{\partial {\bar{y}}^{\prime }}{\eta }^{\prime }dx\\ \text{set:}& \\ & u=\frac{\partial F}{\partial {\bar{y}}^{\prime }},dv={\eta }^{\prime }\\ & du=\frac{d}{dx}(\frac{\partial F}{\partial {\bar{y}}^{\prime }}),v=\eta \\ & \int udv=uv-\int vdu\\ =& \frac{\partial F}{\partial {\bar{y}}^{\prime }}{\int }_{x_1}^{x_2}{\eta }^{\prime }dx-{\int }_{x_1}^{x_2}(\int {\eta }^{\prime })\frac{d}{dx}[\frac{\partial F}{\partial {\bar{y}}^{\prime }}]dx\\ =& \frac{\partial F}{\partial {\bar{y}}^{\prime }}\eta {|}_{x_1}^{x_2}-{\int }_{x_1}^{x_2}\eta \frac{d}{dx}[\frac{\partial F}{\partial {\bar{y}}^{\prime }}]dx\\ =& -{\int }_{x_1}^{x_2}\eta \frac{d}{dx}[\frac{\partial F}{\partial {\bar{y}}^{\prime }}]dx\end{array}$$

Due to $y=\bar{y}$ when $\epsilon =0$, we end up with
$$\begin{array}{rl}\frac{dI}{d\epsilon }& ={\int }_{x_1}^{x_2}[\frac{\partial F}{\partial \bar{y}}-\frac{d}{dx}[\frac{\partial F}{\partial {\bar{y}}^{\prime }}]]\eta dx\\ & ={\int }_{x_1}^{x_2}[\frac{\partial F}{\partial y}-\frac{d}{dx}[\frac{\partial F}{\partial y^{\prime }}]]\eta dx\\ & =0\\ ⟶\frac{\partial F}{\partial y}-\frac{d}{dx}[\frac{\partial F}{\partial y^{\prime }}]& =0\end{array}$$

Remark: Euler-Lagrange Equation is a necessary condition rather than a sufficient one. It doesn’t tell you whether it’s maximum or minimum.