3
$\begingroup$

Consider a function $$ f:\mathbb{R}^n\rightarrow\mathbb{R}^m $$ given by $m$ functions $f_i:\mathbb{R}^n\rightarrow \mathbb{R}$ that we can assume to be polynomials in $x_1,\dots,x_n$.

Does there exist any formula expressing $f$ as a linear combination of $\frac{\partial f}{\partial x_1}(x_1,\dots,x_n),\dots,\frac{\partial f}{\partial x_n}(x_1,\dots,x_n)$, $f(x_1,0,\dots,0), \frac{\partial f}{\partial x_1}(x_1,0,\dots,0),\dots,\frac{\partial f}{\partial x_n}(x_1,0,\dots,0)$, where the coefficients of the linear combination are functions of $x_1,\dots,x_n$? Here $\frac{\partial f}{\partial x_i} = \left(\frac{\partial f_1}{\partial x_i},\dots,\frac{\partial f_m}{\partial x_i}\right)$.

Basically I am asking for an analogue of Euler's formula for homogeneous polynomials which says that if $f:\mathbb{R}^n\rightarrow\mathbb{R}$ is a homogeneous polynomial of degree $d$ then $d\cdot f = x_1\frac{\partial f}{\partial x_1}(x_1,\dots,x_n) + \dots + x_n\frac{\partial f}{\partial x_n}(x_1,\dots,x_n)$.

Thank you.

$\endgroup$
4
  • 1
    $\begingroup$ Consider the constant functions. $\endgroup$ May 15 at 16:57
  • $\begingroup$ are the $f_i$'s homogeneous polynomials? $\endgroup$ May 15 at 16:59
  • $\begingroup$ I think the constant function is ok since we can use also $f(x_1,0,\dots,0)$. No, in general the polynomials are not homogeneous. $\endgroup$
    – R_O
    May 15 at 17:19
  • $\begingroup$ Missed that inclusion given the title. Sorry. $\endgroup$ May 15 at 18:12

1 Answer 1

2
$\begingroup$

$\newcommand{\pa}{\partial}\newcommand{\R}{\mathbb R}$The answer is no. Indeed, suppose the contrary: that for each polynomial $f$ there are functions $a_j,b,c_j$ such that \begin{equation} f(x_1,\dots,x_n)=\sum_{j\in[n]}a_j(x_1,\dots,x_n)(\pa_j f)(x_1,\dots,x_n) \\ +b(x_1,\dots,x_n)f(x_1,0\dots,0) \\ +\sum_{j\in[n]}c_j(x_1,\dots,x_n)(\pa_j f)(x_1,0\dots,0) \tag{1}\label{1} \end{equation} for all $(x_1,\dots,x_n)\in\R^n$, where $[n]:=\{1,\dots,n\}$ and $\pa_j f$ denotes the partial derivative of $f$ with respect to its $j$th argument. Let now \begin{equation} f(x_1,\dots,x_n):=3x_2^2-2x_2^3. \end{equation} Then $(\pa_j f)(1,1,0,\dots,0)=f(1,0\dots,0)=(\pa_j f)(1,0\dots,0)=0$ for all $j\in[n]$, so that the right-hand side of \eqref{1} for $(x_1,\dots,x_n)=(1,1,0,\dots,0)$ is $0$, whereas the left-hand side of \eqref{1} for $(x_1,\dots,x_n)=(1,1,0,\dots,0)$ is $f(1,1,0,\dots,0)=1\ne0$, which contradicts \eqref{1}. $\quad\Box$

$\endgroup$

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.