Minimization using Lagrange multipliers

1 reply · 3K views · last reply
TheCanadian
Messages
361
Reaction score
13
Given the following expressions:

Equations (11) and definitions: time derivative of Omega_j minus curl of (U_j cross Omega_j) equals zero for j=1,2; Omega_1 = B_s, Omega_2 = B_s + curl V; U_1 = V - curl B_s, U_2 = V; helicities and energy integrals.


and that ## \bf{B}_s = \nabla \times \bf{A}_s ##

how does one solve for the following expressions given in (12) and (13)?

Text excerpt stating the coupled Beltrami equations and the variational condition with Lagrange multipliers mu1 and mu2, followed by Euler equations (12) B_s = a(V - curl B_s) and (13) B_s + curl V = bV, with a = 1/mu1, b = 1/mu2.


I've attempted doing so and derive the following expressions (where the hat indicates a unit vector):

## bV = \bf{ \hat{V}} \cdot {\bf{B}_s} + \bf{A}_s \cdot (\nabla \times \bf{\hat{V}}) ##

## {\bf{B}}_s \cdot (\nabla \times {\bf{\hat{A}}_s)} = B_s = ( \frac{1}{a} + \frac {1}{b}){B}_s + \frac{1}{2b}[\bf{V} \cdot (\nabla \times\bf{ \hat {A}}_s) + \bf{\hat {A}}_s \cdot (\nabla \times {\bf{V}})] ##

Although similar in some terms, this is clearly not equivalent to what's stated above after considering further vector identities and there is no curl(B) term present anywhere. It seems so simple yet my calculation is quite a bit off. If anyone could guide me through this with steps they've taken (and possibly ensure the above equations are actually correct), that would be kindly appreciated.
 
Attachments
  • Equations (11) and definitions: time derivative of Omega_j minus curl of (U_j cross Omega_j) equals zero for j=1,2; Omega_1 = B_s, Omega_2 = B_s + curl V; U_1 = V - curl B_s, U_2 = V; helicities and energy integrals.
    Screen Shot 2018-04-21 at 8.13.00 PM.png
    9.9 KB · Views: 1,322
  • Text excerpt stating the coupled Beltrami equations and the variational condition with Lagrange multipliers mu1 and mu2, followed by Euler equations (12) B_s = a(V - curl B_s) and (13) B_s + curl V = bV, with a = 1/mu1, b = 1/mu2.
    Screen Shot 2018-04-21 at 8.13.11 PM.png
    9.1 KB · Views: 1,014
Last edited:
You have an integral of the type
$$E = \int d^3x L(\phi(x), \partial_i \phi(x))$$
which should be minimized. First, I suggest you to re-write the function ##L## inside the integral in tensor notations, for example ##\mathbf B = ∇ \times \mathbf A## is written as ##B_a = \epsilon_{abc}\partial_{b} A_c## (with ##\partial_i## I mean ##\frac {\partial}{\partial x_i}##). Once you have done that, use Euler-Langrange equations for fields https://en.wikipedia.org/wiki/Lagrangian_(field_theory) and Eq. (12) and (13) should follow. I didn't do it, but it seems to me it should work.