5

Suppose we have two polynomials $p(x)$ and $q(x)$ with degrees $m$ and $n$, respectively, and we consider $S^T$: the transpose of the $(m+n)\times(m+n)$ Sylvester matrix associated with those polynomials. The determinant of $S^T$ is the resultant of $p$ and $q$, and we know that if $\text{det}(S^T)=0$, then the polynomials share a root (essentially Bézout's identity).

I'm wondering how one would visualize this result geometrically. My current interpretation is to imagine the columns of $S^T$ embedding $p$ as a vector in $(m+n)$-dimensional space along with $n-1$ rotations about the origin (or $n-1$ multiplications/division by $x$) and then doing the same with $q$ along with $m-1$ rotations. The resultant is then the volume of the parallelepiped formed by these $m+n$ vectors. Having a resultant of $0$ would correspond to these $m+n$ vectors all being in the same plane, but where does the common root fit in this picture? Along a similar vein, how would one go about interpreting the discriminant of a polynomial in this manner (where the discriminant can be defined as the resultant of a polynomial and its derivative)?

  • 1
    Let ${\cal P}n$ denote the space of polynomials of degree $\le n$ and $L_p$ denote left-multiplication $L_p:f(x)\mapsto p(x)f(x)$. We consider the (transpose of) the matrix representation of $L_p\oplus L_q$ as a map ${\cal P}_m\oplus {\cal P}_n\to{\cal P}{m+n}$ (using the usual power bases). By Bezout's, the kernel is the space of polynomials divisible by $\gcd(p,q)$. Thus it's instructive to find out if this map has full rank. I don't know if identifying ${\cal P}m\times{\cal P}_n$ with ${\cal P}{m+n}$ is especially "geometric." – coiso Nov 10 '24 at 03:29

0 Answers0