CoV16

From Exampleproblems

Jump to: navigation, search

Minimize J(y) = \int_0^\infty \left(y^2 + y'^2 + (y''+y')^2\right)dx, y(0)=1, y'(0)=2, y(\infty)=0, y'(\infty)=0\,

Let F(x,y,y',y'')\, denote the integrand in the above. The Euler equation (see CoV4) for this problem is

\frac{d^2}{dx^2}\left(\frac{\partial F}{\partial y''}\right) - \frac{d}{dx}\left(\frac{\partial F}{\partial y'}\right) 
  + \frac{\partial F}{\partial y} = 0

\implies\frac{d^2}{dx^2} 2(y''+y') - \frac{d}{dx}[2y'+2(y''+y')] + 2y = 0

\implies y^{(iv)} - 2y'' + y = 0

The characteristic equation for this differential equation is m4 − 2m2 + 1 = 0 which factors to (m − 1)2(m + 1)2 = 0; so the roots are m = 1,1, − 1, − 1. Thus, the general solution of the differential equation is y = (c_1 + c_2 x)e^x + (c_3 + c_4 x)e^{-x}\,

The conditions at infinity require that c_1 = c_2 = 0\,. Thus, y = (c_3 + c_4 x)e^{-x},\quad y' = \left((c_4 - c_3) - c_4 x\right)e^{-x}\,

Applying the remaining conditions,

c_3 = 1,\quad c_4 - c_3 = 2\, so c_3 = 1, c_4 = 3\, and so the desired solution is

y = (1 + 3x)e^{-x}\,

Satisfying the Euler equation is a necessary condition for the given solution to minimize J(y)\,; proving that it actually does minimize J(y)\, is a lot more difficult (and will not be dealt with here).


Calculus of Variations

Main page

Personal tools

Get A Wifi Network Switcher Widget for Android