Q: Suppose $p_0, \dots , p_m$ are polynomeals in $P_m(F)$ s.t. $p_j(2)=0$ for all $j$. Prove that $p_0, \dots , p_m$ is not linearly independant in $P_m(F)$.

Edit: $P_m(F)$ is the vector space of all polynomials of degree at most m in a Field

My Proof

Suppose that $p_0, \dots , p_m$ is linearly independant. Now, consider the set $z, p_0, \dots , p_m$ where this family spans $P_m(F)$.

Note: $1,x, \dots , x^m$ spans $P_m(F)$ and has $m+1$ elements. So, no linearly independant set in $P_m(F)$ has more than $m+1$ elements

So, $z, p_0, \dots , p_m$ is linearly dependent (has $m+2$ elements) So, for $\beta, \alpha_j \in F$ (not all $=0$), $\beta z + \alpha_0 p_0 + \dots +\alpha_m p_m = 0$

if $\beta = 0$, then $\alpha_0 p_0 + \dots +\alpha_m p_m = 0$ and since it is assumed that $p_0, \dots , p_m$ is linearly independant, then this forces all $\alpha_j = 0$, which contradicts the fact that $z, p_0, \dots , p_m$ is linearly dependant. Hence, $\beta \neq 0$.

Then, $z = (-1/\beta)(\alpha_0 p_0 + \dots +\alpha_m p_m)$ which is in the $span\{p_0 , \dots , p_m\}$, so $z \in span\{p_0, \dots , p_m\}$. So, $\exists \alpha_j \in F$ such that $z=\alpha_0 p_0 (z) + \dots + \alpha_m p_m (z)$. But, $p_j(2)=0$ for $\forall j$, and $2 \neq 0 + \dots + 0 = 0$, so contradiction. SO, $p_0, \dots , p_m$ is not linearly independent.

Is this proof sufficient? Are there any holes or inconsistencies that I should be worried about?

  • $\begingroup$ What is $P_m(F)$? $\endgroup$ – Lord Shark the Unknown May 10 '17 at 15:41
  • $\begingroup$ The vector space of polynomials of degree m in a Field. Thank you, I should clarify $\endgroup$ – Pascal May 10 '17 at 15:42
  • 1
    $\begingroup$ Some remarks: a) I suppose that you are in characteristic $0$. In characteristic $2$, you have $2=0$. b) If you suppose that $p_0,\cdots,p_m$ are linearly independants, as they are $m+1$ such polynomials this is a basis for $P_m(F)$. So you can write directly that $z$ (the variable) is a linear combination of the $p_j$ c) You can also write $p_j(z)=(z-2)q_j(z)$. Then the degree of the polynomials $q_j$ are $\leq m-1$, they are in $P_{m-1}(F)$. This family of $m+1$ elements of this vector space,( that is of dimension $m$) are hence linearly dependant, and this imply the same for the $p_j$. $\endgroup$ – Kelenner May 10 '17 at 16:43

The map $\varepsilon:P_m(F)\to F$ given $\varepsilon(p)=p(2)$ is a surjective linear transformation. Therefore, its kernel has dimension $(m+1)-1=m$ and so any $m+1$ elements in the kernel must be linearly dependent.

Here is another way to express the same idea:

If $p_0, \dots , p_m$ were linearly independent, then they'd form a basis for $P_m(F)$ and every element $p \in P_m(F)$ would be a linear combination of them. But then $p(2)=0$, which is certainly not the case for the constant polynomial $p=1$.


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.