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

Consider the delay differential equation:

$ y_x(x) = \sqrt{y(x-\bar{x})} $

where $y$ is the unknown function of $x$, and where $\bar{x}$ is a fixed parameter.

This equation does not seem to have a known closed-form solution.

Would anyone know how to get a series solution for $y(x)$?


share|cite|improve this question
why do you want a series solution ? You could instead use qualitative tools. A reference book on delay differenctial equation is that of Jack Hale. – Denis Serre Oct 11 '12 at 20:02
Thank you, well ideally I'd like to see an approximate formula for $y(x)$, with $\bar{x}$ a parameter (or for the special case $\bar{x}=1$ or for the special case where $\bar{x}$ small), that would help in understanding the solution. My math is very limited, a series solution is merely the first thing I thought about. – PatrickT Oct 13 '12 at 15:42

As mentioned already by Denis Serre, there is a rich literature investigating delay equations.

If you make an experiment, and fix $\bar{x}=1$, then you see that you need as an initial value the complete past on $[-1,0]$. To play a bit, tak as an initial function the constant function $y(s)=1$ for $s\in[-1,0]$. Then you can calculate the solution explicitly for $x\in[0,1]$, then using this you can calculate the solutuion on $[1,2]$, etc. We see that it is far from being analytic. Hence, no chance for a series sepresentation of a solution.

If you are interested in classical stuff, then Bellman and Cooke is an excellent book. An other good reference is the one by Hale and Verduyn Lunel.

ADDED: If it is a delay equation (i.e., $\bar{x}>0$), then the initial condition has to be a function (you have to know the whole past). Then the iteration procedure I described works always. This gives you a possible approximation formula, most numerical methods also work this way.

You are right about analyticity: series representation does it. Smoothness is a consequence. The example I presented to you is only once differentiable at $x=1$, twice at $x=2$, etc. Hence, cannot be analytic.

share|cite|improve this answer
Thanks! I have seen the books you cite and read parts of them, but I didn't see anything there that might be directly applicable to this problem. I had simulated the solution following the scheme you suggest, for various choices of initial conditions, but I'm not entirely sure how to write the initial conditions; I was hoping an approximate formula might help me understand how to set the initial conditions. Based on related problems solved without reduction to a DDE, I expect the solution is continuous and differentiable for strictly positive values of $x$. The case $\bar{x}<0$ may be easier. – PatrickT Oct 13 '12 at 15:54
Thanks for pointing out the analyticity problem. I'm not sure what "analytic" means, to be quite honest. How can you tell when a function is or isn't? I seem to remember something like: non-analytic functions don't admit a (any?) series representation, but I'm quite ignorant beyond that. The derivative of $y(x)$ exists, doesn't it? since it's the right-hand side of the DDE, provided $y(x)$ itself exists... or is that a fallacy? How about other kinds of series? Thanks again! – PatrickT Oct 13 '12 at 16:01
And by the way, I'm such a newbie that I can't upvote or do anything useful like that, sorry... – PatrickT Oct 13 '12 at 16:03
Thanks András for the additional information. – PatrickT Oct 16 '12 at 14:05

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.