# A question on existence of solutions of a linear ODE system

I am working on a problem of harmonic functions on surfaces, and in one step I got the following system of ODEs with prescribed asymptotes. I was wondering what methods could give us the existence or non-existence, thank you very much.

For $r$ large enough, we want a solution to \begin{equation*} \begin{cases} -\frac{u''}{u}+w''\geq-C,\\ -\frac{u''}{u}+\frac{u'}{u}w'\geq-C,\\ \\ u(r)= O(e^{-r^{2+\epsilon}}),\\ %|w(r)|= O(r^{1+\delta}),\\ \limsup |r^{-1-\delta} w(r)|>0,\\ \end{cases} \end{equation*} where $\epsilon>0,\delta>0$ are small constants.

• Where is the "following system of ODE" ? – Alexandre Eremenko Jun 28 '14 at 8:47
• OD inequalities...sorry. – littlelittlelittle Jun 28 '14 at 21:57

Subtracting the second inequality from the first and also keeping the second one is an equivalent set of inequalities: $-u''/u + (u'/u) w' \ge -C$, $w'' - (u'/u) w' \ge 0$. A little bit of manipulation then shows that this system of inequalities can be rewritten as $$\begin{cases} -\frac{e^w}{u}\left(e^{-w}u'\right)' \ge -C , \\ u (w'/u)' \ge 0 . \end{cases}$$
If you make assumptions about signs of the multiplicative factors, you can move those to the other side of the inequalities and then integrate. Since (definite) integrals preserve inequalities, you get new inequalities that no longer involve derivatives of $u$ and $w$. Perhaps some version of those can help you make a conclusion about the existence or non-existence of a desired solution.
Update: Actually, I feel rather silly about part of what I wrote. One cannot subtract inequalities as I did above ($a \ge b$ and $c \ge d$ does not imply $a-c \ge b-d$). So the second inequality that I wrote is not actually valid. Apologies for any confusion! I'll leave what I wrote as is, though, in case the algebraic manipulations do prove useful.