How can I find an outer expansion for f'(r) in this ODE?

  • Level: Graduate 
  • Thread starter Thread starter Chewie666
  • Start date Start date
  • Tags Tags
    Expansion Ode
Join the discussion
Registration is free. Start your own thread to ask a follow-up.
1 reply · 3K views
Chewie666
Messages
1
Reaction score
0
Ahoy!

I'm trying to approximate [itex]f'(r)[/itex] for the following equation using matched asymptotic expansions

[itex]-\frac{1}{2}\epsilon ff''=\left[\left(\epsilon+2r\right)f''\right]'[/itex]

where [itex]\epsilon \ll 1[/itex] and with the boundary conditions [itex]f(0)=f'(0)=0, \quad f'(\infty)=1[/itex]

The inner expansion which satisfies [itex]f'(0)=0[/itex] is simple enough by choosing an appropriate inner variable.

My problem is trying to form an outer expansion of the form

[itex]f'=1+\sigma(\epsilon) f_1+ \dots[/itex]

where [itex]\sigma[/itex] is found through matching. In my working I find [itex]f_i≈A_i\ln r[/itex] which obviously doesn't satisfy [itex]f'(\infty)=0[/itex] unless the constants equal zero.

I've tried introducing a stretched variable of the form [itex]\gamma =\epsilon r[/itex] but with no success.

Any suggestions?

Cheers
 
Physics news on Phys.org
It is obvious from inspection that [itex]f(r) = r[/itex] is a solution of the full ODE for any [itex]\epsilon[/itex]. It satisfies [itex]f(0) = 0[/itex] and [itex]f'(\infty) = 1[/itex]. This suggests that higher order corrections to [itex]f(r) = r[/itex] are unnecessary, except in the boundary layer near the origin since [itex]f'(0) \neq 0[/itex]. This is consistent with your conclusion that the outer corrections must vanish.