Let for $n > 1$ and $m = n-1$ $$ p = x^n + a_1 x^{n-1} + \cdots + a_m x $$ be a polynomial with $a_i \in \CC$. Call $p^{(i)}(x) = \frac{d^ip}{dx^i}(x)$.

The following question would have an impact on the problem, if sequences of resultants $\mathrm{res}_x(p(x), p^{(n-i)}(x))$, $i = 1,\ldots,s$ are regular sequences:

Take $s < m$ and consider the space $X$ of monic polynomials $p$ as above such that $\alpha_1,\ldots,\alpha_s \in \CC$ (not necessarily distinct, could be even all equal) exist so that

\begin{equation} p^{(n-1)}(\alpha_1) = p^{(n-2)}(\alpha_2) = \cdots = p^{(n-s)}(\alpha_s) = 0 \end{equation}

and \begin{equation} p(\alpha_1) = \cdots = p(\alpha_s) = 0 \end{equation}


Can one show that for a certain solution $p(x) \in X$ with $a_{s+1} = 0$ there exists in every neighbourhood (identifying $X \cong \CC^m$) a solution in $X$ with $a_{s+1} \neq 0$?

This is obviously the case for $2 s + 1 \leqslant m$ where one can think of the $\alpha_i$ as fixed and consider the problem as a statement of linear algebra. If $2 s + 1 > m$ it may become necessary to "move the $\alpha_i$ a bit", to get $a_{s+1} \neq 0$.

I did some computer calculations with Maple and it always seemed to be possible to do such a small move, leading to $a_{s+1} \neq 0$, but I could not find a general proof.

Is the answer to the above question maybe already known to specialists in (Birkhoff-)interpolation? What could be helpful to read for making progress on this problem?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.