Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

I am trying to find some general properties of the zeros of

$P(z) = \sum_{i=1}^n \frac{\alpha_i}{z+z_i}$,

with $\sum_{i} \alpha_i = 0$, $z_i \in [-M\; 0], i=1,\ldots,n$ and all $\alpha_i$ and $z_i$ are real, and $M<\infty$.

Actually, what I really hope to find is that the real part of the zeros lie in the same interval as the poles, i.e., in the interval $[-M \; 0]$.

This problem arises in a particular type of dynamic systems over graphs. The function $P(z)$ is the transfer-function of a linear system running a consensus protocol, so the poles of the system are the eigenvalues of the combinatorial Laplacian.

Thanks!

share|improve this question
    
The poles are $-z_i\in[0,M]$, not in $[-M,0]$. Or did you want to write $\sum_{i=1}^n\frac{\alpha_i}{z-z_i}$? –  Emil Jeřábek Feb 25 '11 at 11:23
    
Hi Emil, Yes, you are correct. $z_i \in [0, \; M]$ and the poles of the system are therefore in the interval $[-M, \; 0]$. –  dan Feb 25 '11 at 11:58
    
I received a couple comments (that appear deleted now) that suggest the Gauss-Lucas theorem might be applicable. I guess this is not the case, as the constants $\alpha_i$ are not integer (and may not even be rational). Are there any generalizations of the Gauss-Lucas theorem that can handle this case? –  dan Feb 25 '11 at 12:34
    
Sorry that was my comment earlier, and I deleted it because as you say it doesn't answer the question. I somehow read the condition as $\alpha_i\geq 0$. I think that by approximating $\alpha_i$ by rationals and rescaling makes it safe to assume that they are integers, bu this doesn't help as the only case I know how to put any sort of bound on the roots of $P$ is when $\sum \alpha_i \neq 0$. –  Gjergji Zaimi Feb 25 '11 at 13:09
    
Dan, are you still here? Does my answer of 27 February meet the specifications, or did I miss the point? –  Gerry Myerson Mar 5 '11 at 3:16

1 Answer 1

up vote 2 down vote accepted

$${1\over z+1}+{-2\over z+10}+{1\over z+20}={170-z\over(z+1)(z+10)(z+20)}$$ seems to satisfy your conditions with $n=3$ and $M=20$ but the zero at 170 is not in the same interval as the poles.

share|improve this answer

Your Answer

 
discard

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.