MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I was wondering if anybody could give me some references to already existing literature for the following open ended problem.

Namely, I am interested in studying the equation of "complex harmonic oscillator"


where $z:\mathbb{R}\to\mathbb{C}$.

The case when $t$ is complex is also interesting and might shed some light on to the real case. Assume that the function $q(t)$ is real (for real time) and strictly decreasing. Actually, without lost of generality let $q(t)=1/(1+t)$. I have strong numerical evidence for the following claim. There exist the set of initial conditions such that $|z|$ is a constant or strictly monotone decreasing (increasing) functions on the solutions satisfying those initial conditions.

I am considering the following auxiliary function $\rho(t)=\ln(z\cdot \bar{z})$ the real case but I still can not make a use of the fact that q(t) is positive and decreasing on $[0,\infty)$. For the curious, I am essentially play the same kind a game like in the proof of Sturm-Picone comparison theorem or Poincare-Benedixon theorem.

I started looking into the case of the complex time due to the following simple observation. Suppose that $z_1(t)$ and $z_2(t)$ are two linearly independent holomorphic solutions of the equation. The Schwarzian derivative in notation $S$ of the ratio $z_1/z_2$ satisfies


Thank you.

share|cite|improve this question
Why do you say "without lost of generality" when you choose $q(t)$? Does that choice of $q(t)$ obey a universality property? – S. Carnahan May 26 '11 at 5:43
There is a whole class of quantum mechanics systems and an interesting question about them which essentially boils down to the above question about that particular ODE. The value of the function $q(t)=1/(1+t)$ is a particular value for function $q(t)$ for one of those systems. There are even some in which $q(t)$ is a periodic function (Floquet case). So the phrase without loss of generality is really a bad choice but even with that particular value of $q(t)$ I will be able to answer the question at least about one interesting quantum mechanics problem. – Predrag Punosevac May 26 '11 at 13:51
If you would like to read about the Schroedinger equation with a complex valued <i>polynomial</i> potential, I point you to my licentiate thesis: The property about the Schwarzian derivative is mentioned there. There might be some inspiration there? – Per Alexandersson May 26 '11 at 18:49
up vote 4 down vote accepted

The d.e. $z'' + z/(1+t) = 0$ is not at all general, in fact it has closed-form solutions $z \left( t \right) =c_{{1}}\sqrt {1+t}\ {J_1 \left(2\sqrt {1+t}\right)}+c_{{2}}\sqrt {1+t}\ {Y_1 \left(2\sqrt {1+t}\right)}$ where $J_1$ and $Y_1$ are the Bessel functions of the first kind and order 1. In fact your equation becomes Bessel's equation of order 1 under the transformation $ t = s^2/4 - 1,\ z(t) = s y(s)$.

The asymptotics of these functions are well known: $z(t) = - \frac{t^{1/4}}{\sqrt{\pi}}(c_1 \cos(2 \sqrt{1+t} +\pi/4) + c_2 \sin(2 \sqrt{1+t} + \pi/4)) + O(t^{-1/4})$. So your conjecture is false: all the real solutions (except the constant 0) oscillate with increasing amplitude as $t \to \infty$.

share|cite|improve this answer
I appreciate your answer very much! I will have to think more about it but I have a quick question. My understanding is that Bessel DE is real ODE. Note that the function $z(t)$ is complex valued even when $t$ is real. Actually the above equation easily decouples in the case of real $t$ (real valued function $q(t)$) into two Bessel ODEs. $$\ddot{x}+q(t)x=0$$ $$\ddot{y}+q(t)y=0$$ My claim is about $|z|=\sqrt{x^2+y^2}$. Showing that $|z|$ increasing is very, very GOOD! See above comment why. Could you give me a reference? – Predrag Punosevac May 26 '11 at 14:32
The complex solutions with $c_1= \pm i c_2$ will have $\frac{∣z(t)∣^2}{∣c1|^2} =\frac{\sqrt{t}}{\pi}+\frac{19}{32 \pi \sqrt{t}} +O(t^{−3/2})$ and $\frac{d}{dt} \frac{|z|^2}{∣c1|^2} = \frac{1}{2 \pi \sqrt{t}} −\frac{19}{64 t^{3/2} \pi} +O(t^{−5/2})$ as $t \to + \infty$, and in particular $|z|$ will be increasing for sufficiently large real $t$. In fact, it appears to be increasing for all $t>−1$. I used Maple for all this, but the asymptotics of Bessel functions should be found in standard references such as Abramowitz and Stegun. – Robert Israel May 26 '11 at 16:31

Relevant references:

share|cite|improve this answer
I have both books on my shelf. – Predrag Punosevac May 26 '11 at 13:33
I should be more careful before dismissing answers. For the past two-three days I have carefully read chapter five of the Hille's book at least dozen of times as well as skimming through several other chapters. The book contains exactly the kind of information I was craving for! – Predrag Punosevac Jul 8 '11 at 18:10

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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