I have a system of ODEs: $$f_1'(t) + \sum_{j=1}^n f_i(t)a_{j1}(t)=d_1(t)$$ $$...$$ $$f_n'(t) + \sum_{j=1}^n f_i(t)a_{jn}(t)=d_n(t)$$

How can I show that there exists a unique solution $(f_1$, ..., $f_n)$?

According to wiki, if I put this equation into the form $y'(t)= F(y(t),t)$ I need $F$ to be Lipschitz wrt. $y$, which in my case works if I look at each individual equation but as a system I don'tknow. What assumptions do I need on the functions $a$, $c$ and $d$?

I don't necessarily have continuity of $a$ or $d$. They may lie in some Lebesgue or Sobolev space.

Thanks for any help


It's a linear system of odes. Let $A_{ij}:=-a_{ji}$, then you have that


Suppose that $A$ and $d$ are defined on some interval $\mathcal{I}\subseteq\mathbb{R}$. Then sufficient conditions for the existence and uniqueness of are that $A$ and $d$ are continuous functions. That is, if $A:J\rightarrow\mathbb{R}^{n\times n}$ and $d:J\rightarrow\mathbb{R}^n$ are continuous, then for every $x\in\mathbb{R}^n$ and $t_0\in J$ there exists a unique continuous function $f:J\rightarrow\mathbb{R}^n$ such that $f(t_0)=x$ and


holds for all $t\in J$. A proof for this is can be found page 54 of this book (and, I suspect, also just googling "existence and uniqueness of solutions, linear systems" for a bit).

EDIT: Ok, so apparently $d$ and $A$ being measurable and locally integrable is sufficient to guarantee the conclusion above. By locally integrable I mean that for all $a,b\in J$,

$$\int_a^b||A(t)||_\infty dt<\infty,\quad\quad, \int_a^b||d(t)||dt<\infty.$$

For details see Proposition C.3.8. of the appendix of this book.

  • $\begingroup$ Thanks, but yeah as you wrote we don't have continuity.. $\endgroup$ – matt.w Apr 25 '13 at 15:30
  • $\begingroup$ See the edit to the answer. $\endgroup$ – jkn Apr 25 '13 at 16:59

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.