TOPICS
Search

Euler-Lagrange Differential Equation


The Euler-Lagrange differential equation is a central equation of the calculus of variations. It characterizes stationary curves of a functional of the form

 J=int_(t_1)^(t_2)f(t,y,y^.)dt,
(1)

where

 y^.=(dy)/(dt).
(2)

For fixed endpoint values and twice continuously differentiable f and y, J is stationary precisely when the Euler-Lagrange differential equation

 (partialf)/(partialy)-d/(dt)((partialf)/(partialy^.))=0
(3)

is satisfied.

If time-derivative notation y^. is replaced instead by space-derivative notation y_x, the equation becomes

 (partialf)/(partialy)-d/(dx)(partialf)/(partialy_x)=0.
(4)

The Euler-Lagrange differential equation will be implemented in a future version of the Wolfram Language as EulerLagrangeEquation[f, u[x], x].

In many physical problems, f_x (the partial derivative of f with respect to x) turns out to be 0, in which case a manipulation of the Euler-Lagrange differential equation reduces to the greatly simplified and partially integrated form known as the Beltrami identity,

 f-y_x(partialf)/(partialy_x)=C.
(5)

For three independent variables (Arfken 1985, pp. 924-944), the equation generalizes to

 (partialf)/(partialu)-partial/(partialx)(partialf)/(partialu_x)-partial/(partialy)(partialf)/(partialu_y)-partial/(partialz)(partialf)/(partialu_z)=0.
(6)

To derive the Euler-Lagrange differential equation, examine

deltaJ=deltaintL(q,q^.,t)dt
(7)
=int((partialL)/(partialq)deltaq+(partialL)/(partialq^.)deltaq^.)dt
(8)
=int[(partialL)/(partialq)deltaq+(partialL)/(partialq^.)(d(deltaq))/(dt)]dt,
(9)

since deltaq^.=d(deltaq)/dt. Applying integration by parts to the second term gives

 int(partialL)/(partialq^.)(d(deltaq))/(dt)dt=int(partialL)/(partialq^.)d(deltaq)=[(partialL)/(partialq^.)deltaq]_(t_1)^(t_2)-int_(t_1)^(t_2)(d/(dt)(partialL)/(partialq^.)dt)deltaq.
(10)

Combining (9) and (10) then gives

 deltaJ=[(partialL)/(partialq^.)deltaq]_(t_1)^(t_2)+int_(t_1)^(t_2)((partialL)/(partialq)-d/(dt)(partialL)/(partialq^.))deltaqdt.
(11)

But we are varying the path only, not the endpoints, so deltaq(t_1)=deltaq(t_2)=0 and (11) becomes

 deltaJ=int_(t_1)^(t_2)((partialL)/(partialq)-d/(dt)(partialL)/(partialq^.))deltaqdt.
(12)

At a stationary curve, the first variation deltaJ vanishes for every admissible perturbation deltaq. The fundamental lemma of calculus of variations then gives

 (partialL)/(partialq)-d/(dt)((partialL)/(partialq^.))=0.
(13)

If the times t_1 and t_2 are fixed and the endpoint values are free, the same equation holds in the interior. The first variation then also requires the natural boundary conditions partialL/partialq^.=0 at each free endpoint. A cost depending on the endpoint values changes these conditions in the Bolza problem (Gelfand and Fomin 1963, pp. 24-26).

For fixed endpoint values and four continuous derivatives of f, the change in J under the perturbation kappav can be expanded to fourth order in kappa,

DeltaJ=int_(t_1)^(t_2)[f(t,y+kappav,y^.+kappav^.)-f(t,y,y^.)]dt
(14)
=kappaI_1+1/2kappa^2I_2+1/6kappa^3I_3+1/(24)kappa^4I_4+o(kappa^4),
(15)

where v(t_1)=v(t_2)=0 and v^.=dv/dt. The first four variations are

I_1=int(vf_y+v^.f_(y^.))dt
(16)
I_2=int(v^2f_(yy)+2vv^.f_(yy^.)+v^.^2f_(y^.y^.))dt
(17)
I_3=int(v^3f_(yyy)+3v^2v^.f_(yyy^.)+3vv^.^2f_(yy^.y^.)+v^.^3f_(y^.y^.y^.))dt
(18)
I_4=int(v^4f_(yyyy)+4v^3v^.f_(yyyy^.)+6v^2v^.^2f_(yyy^.y^.)+4vv^.^3f_(yy^.y^.y^.)+v^.^4f_(y^.y^.y^.y^.))dt.
(19)

The second variation can be re-expressed using

 d/(dt)(v^2lambda)=v^2lambda^.+2vv^.lambda,
(20)

so

 I_2+[v^2lambda]_(t_1)^(t_2)=int_(t_1)^(t_2)[v^2(f_(yy)+lambda^.)+2vv^.(f_(yy^.)+lambda)+v^.^2f_(y^.y^.)]dt.
(21)

But

 [v^2lambda]_(t_1)^(t_2)=0.
(22)

If f_(y^.y^.) is nonzero on [t_1,t_2], the integrand can be completed to a square whenever lambda satisfies

 f_(y^.y^.)(f_(yy)+lambda^.)=(f_(yy^.)+lambda)^2
(23)

on that interval. Substituting

 f_(yy^.)+lambda=-(f_(y^.y^.))/z(dz)/(dt)
(24)

gives the Jacobi differential equation for z,

 f_(y^.y^.)z^..+f^._(y^.y^.)z^.-(f_(yy)-f^._(yy^.))z=0.
(25)

If this equation has a solution z nonvanishing on [t_1,t_2], then

I_2=intf_(y^.y^.)(v^.+(f_(yy^.)+lambda)/(f_(y^.y^.))v)^2dt
(26)
=intf_(y^.y^.)(v^.-v/z(dz)/(dt))^2dt.
(27)

See also

Beltrami Identity, Bolza Problem, Brachistochrone Problem, Calculus of Variations, Euler-Lagrange Derivative, Functional Derivative, Jacobi Differential Equation, Variation

Explore with Wolfram|Alpha

References

Arfken, G. Mathematical Methods for Physicists, 3rd ed. Orlando, FL: Academic Press, 1985.Forsyth, A. R. Calculus of Variations. New York: Dover, pp. 17-20 and 29, 1960.Gelfand, I. M. and Fomin, S. V. Calculus of Variations. Englewood Cliffs, NJ: Prentice-Hall, 1963.Goldstein, H. Classical Mechanics, 2nd ed. Reading, MA: Addison-Wesley, p. 44, 1980.Lanczos, C. The Variational Principles of Mechanics, 4th ed. New York: Dover, pp. 53 and 61, 1986.Morse, P. M. and Feshbach, H. "The Variational Integral and the Euler Equations." §3.1 in Methods of Theoretical Physics, Part I. New York: McGraw-Hill, pp. 276-280, 1953.

Referenced on Wolfram|Alpha

Euler-Lagrange Differential Equation

Cite this as:

Weisstein, Eric W. "Euler-Lagrange Differential Equation." From MathWorld--A Wolfram Resource. https://mathworld.wolfram.com/Euler-LagrangeDifferentialEquation.html

Subject classifications