The equation is $\frac{dA}{d\tau} = \frac{3}{2}iA^2 A^*$, where $A: \mathbb{R}\rightarrow\mathbb{C}$.


The function $A$ arises in a perturbation methods problem, using the method of multiple scales. The original equation is the undamped Duffing equation, $y^{\prime\prime}+y+\varepsilon y^3 = 0, \ y(0)=1, \ y^\prime(0)=0.$ It is determined that $y$ can be approximated as $y(t) = A(\tau)e^{it}+A^*(\tau)e^{-it}$. Under this framework, the original equation arises as a solvability condition for preventing secular terms, and the solution is $A = \frac{1}{2}e^{\frac{3}{8}i\tau}$ after imposing the intial conditions.

My question is, how does one reach the given solution?


My attempt so far has been to deal with this the way I would if A was a real-valued function, which is to separate the equation into $\frac{dA}{A^2A^*} = \frac{3}{2}id\tau$, then integrate. The problem I run into is that the left hand side cannot be integrated like a real-valued function since it involves a conjugate. In $\mathbb{R}$, the $^*$ operator is an identity, so the LHS becomes $\frac{dA}{A^3}$, and solving the equation gives $A$ as a square root function of $\tau$. I thought of using contour integration, but there's no domain to integrate over. When doing the method of separation with a real-valued function, I would use an indefinite integral. However, I don't know how to use an indefinite integral with a complex-valued integrand, if it's even possible.


Note that if $s(\tau) = A(\tau)\overline{A}(\tau)$, then $s'(\tau) = A(\tau) \overline{A'(\tau)}+A'(\tau) \overline{A(\tau)} = 0$, hence we can treat $A \overline{A} = |A|^2$ as a constant and the equation becomes $A' = i{3 \over 2}|A(0)|^2 A$ which has solution $A(\tau) = A(0) e^{i{3 \over 2}|A(0)|^2 \tau}$.

Presumably in the case above, you have $A(0) = {1 \over 2}$.


Note that $A'(\tau) = i{3 \over 2} A(\tau)^2 \overline{A(\tau)}$, so $A(\tau)\overline{A'(\tau)} = -i{3 \over 2}|A(\tau)|^4$, and $A'(\tau) \overline{A(\tau)} = i{3 \over 2}|A(\tau)|^4$. Summing gives zero.

  • $\begingroup$ Why is $s^\prime(\tau)=0$? I'm getting $s^\prime(\tau) = 2uu^\prime + 2vv^\prime$, where $A=u+iv$. $\endgroup$ – Alex Jones Mar 25 '17 at 3:58
  • $\begingroup$ @AlexanderJ93: Let me check, wine & maths. don't mix well... $\endgroup$ – copper.hat Mar 25 '17 at 4:03
  • $\begingroup$ @AlexanderJ93: I think my computation is correct, I presume by $z^*$ you mean the conjugate? Just substitute the formula for $A'(\tau)$ into the expression above for $s'(\tau)$. $\endgroup$ – copper.hat Mar 25 '17 at 4:04
  • $\begingroup$ Yes, we're talking about the same thing. But $|A|^2 = u^2+v^2$ and $[|A|^2]^\prime = 2uu^\prime+2vv^\prime$, as far as I can tell. I get the same expression for $s^\prime$ as you are, but I'm not getting it to be 0. Can you show me what steps you're using to get there? $\endgroup$ – Alex Jones Mar 25 '17 at 4:07
  • 1
    $\begingroup$ @AlexanderJ93: Sometimes a cigar is just a cigar :-). $\endgroup$ – copper.hat Mar 25 '17 at 4:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.