Jacobi equation (source code)

= Jacobi equation
{c}

The <Jacobi equation> is the linearization of an <Euler-Lagrange equation> along a stationary path. For a smooth scalar integral $I[u]=\int F(x,u,u\prime)dx$, its fixed-endpoint <second variation> has coefficients $P=F_{pp}$, $B=F_{up}$ and $C=F_{uu}$ evaluated along the path:
$$
\delta^2I[u;\eta]=\int(P\eta\prime{}^2+2B\eta\eta\prime+C\eta^2)dx
=\int(P\eta\prime{}^2+Q\eta^2)dx,\qquad Q=C-B\prime.
$$
The boundary term vanishes because $\eta$ vanishes at the endpoints. Linearizing the stationarity equation gives
$$
-(P\eta\prime)\prime+Q\eta=0.
$$
The same equation is the stationarity equation of this quadratic form. If $P>0$ continuously on the closed interval, it is a regular second-order equation.

For $F=(p^2-\lambda u^2)/2$, it becomes $\eta\prime\prime+\lambda\eta=0$. When $\lambda>0$, the solution vanishing at zero is a multiple of $\sin(\sqrt\lambda x)$; its next zero at $\pi/\sqrt\lambda$ gives a <conjugate point>. More generally, solutions describe infinitesimal families of stationary paths, and their zeros enter the <Jacobi positivity criterion>. This is a linearized equation around a specified extremal, not the original nonlinear boundary problem.