4

Let $n \geq 4$ and $\mathbb{R_n}$[x] a vector space of all polynomials of degree n at most. There is given a set: $ U = \{p \in \mathbb{R_n}[x]; p(1) = p(-1), p''(0) = 2p(1)\} $

Find some basis of a vector subspace U and deterime $dimU$

My attempted solution:

$p(x) = a_nx^n + \dots + a_1x + a_0$

$p'(x) = na_nx^{n-1} + \dots + 2a_2x +a_1$

$p''(x) = n(n-1)a_nx^{n-2} + (n-1)(n-2)a_{n-1}x^{n-3} + \dots + 6a_3x +2a_2$

$p(1) = a_n + \dots + a_0$, if $n = 2k, k \in \mathbb{N}$: $p(-1) = a_n - a_{n-1} + \dots -a_1 +a_0$ $\rightarrow 2(a_{n-1} + a_{n-3} + \dots a_1) = 0$

$p''(0) = 2a_2 = 2p(1) = 2(a_n + \dots a_0) \rightarrow a_n + \dots + a_3 + a_1 + a_0 =0$

$$ a_1= -a_{n-1}-a_{n-3}- \dots a_3 \\ a_0 = -a_n - \dots - a_3 -a_1 $$

$p(x) = a_nx^n + \dots + a_2x^2 - x(a_{n-1} + a_{n-3} + \dots a_3) - a_n - \dots - a_3 - a_1 = \\=a_n(x^n-1) + a_{n-1}(x^{n-1}-x-1) + a_{n-2}(x^{n-2}-1)+ \dots + a_3(x^3-x-1) + a_2x^2 - a_1$

Not sure if the process is correct, because the polynomials by the coefficients, which should form a basis, seem different from the solution.

m259
  • 139
  • The whole question boils down to finding a basis of $\mathbb{R}^{n+1}$ such that the above equations you got for the coefficients hold true. Notice that these equations are linear. – YoungMath Jun 12 '22 at 07:36
  • I'm still not sure if my solution is correct, since the given one when $n$ is even is: a basis is ${x^n-1, n^{n-1} -x, \dots, x^4-1, x^3-x, x^2 }$ – m259 Jun 12 '22 at 08:03
  • I don't even know what your solution is supposed to be here. You haven't provided any basis... – YoungMath Jun 12 '22 at 08:13
  • I suppose the polynomials by the coefficients $a_n, \dots a_1$. Meaning the basis would be: ${x^n-1, x^{n-1}-x-1, x^{n-2}-1, \dots, x^4-1, x^3-x-1, x^2}$. – m259 Jun 12 '22 at 08:31

1 Answers1

2

You have not computed any basis. You have tried to eliminate the coefficients using the equations but only have done so for $a_{0}$ and not $a_{1}$. Note that you can actually eliminate both $a_{1}$ and $a_{0}$ and write your $p(x)$ in terms of $a_{n},...a_{2}$ .

You should proceed like this:-

Let $n$ be even . Let $n=2m$ , $m\geq 2 $

Then $p(1)=p(-1)\implies a_{0}+a_{1}+...+a_{n}=a_{0}-a_{1}+...-a_{n-1}+a_{n}$

Which gives $a_{1}+a_{3}+...+a_{2m-1}=0$

$p''(0)=2\cdot(\text{coefficient of}\, x^{2})=2a_{2}$ .

So $2a_{2}=2(a_{0}+a_{1}+...+a_{2m})\implies a_{2}=a_{0}-(a_{3}+a_{5}+...+a_{2m-1})+a_{2}+...+a_{2m}\\=a_{0}+a_{2}+...+a_{2m}$ .

So $a_{0}+a_{4}+a_{6}+...+a_{2m}=0$ . So $a_{0}=-(a_{4}+a_{6}+...+a_{2m})$

This means $p(x)=-(a_{4}+...+a_{2m})-(a_{3}+a_{5}+...+a_{2m-1})x+a_{2}x^{2}+a_{3}x^{3}+...+x^{2m}a_{2m}$

So $p(x)=a_{2}x^{2}+a_{3}(x^{3}-1)+a_{4}(x^{4}-1)+...+(x^{2m}-1)a_{2m}$.

So $U=\{ a_{2}x^{2}+a_{3}(x^{3}-x)+a_{4}(x^{4}-1)+a_{5}(x^{5}-x)+...+(x^{2m}-1)a_{2m}\,,a_{i}\in\Bbb{R}\}$

That is $U=\text{span}\{x^{2},x^{3}-x,x^{4}-1,...,x^{2m-1}-x,x^{2m}-1\}$

Now you need to extract a linearly independent set out of the above. Matrices and Row Reduced Echelon Form is the best way to go.

These above vectors represented as row vectors of an $(2m-1)\times (2m+1)$ matrix gives us the following :-

$$\begin{bmatrix} 1 & 0 & 0 &\cdots & 0 & 0 & -1 \\ 0 & 1 & 0 &\cdots & 0 & -1 &0 \\ \vdots &\vdots&\vdots &\cdots&\vdots&\vdots&\vdots\\ 0 & 0 & 0 &\cdots & 0 & -1 & 0 \\ 0 & 0 & 0&\cdots & 1 & 0 & 0 \end{bmatrix}$$ .

Now you can use induction to prove that this matrix has full rank of $2m-1$ (as is evident from the Row reduced form of the matrix which it is already in ) and this would imply that $\{x^{2m}-1,x^{2m-1}-x,...,x^{3}-x,x^{2}\}$ is a basis for $U$.

For the case $m=1$ we have that $p(1)=p(-1)\implies a_{0}+a_{1}+a_{2}=a_{0}-a_{1}+a_{2}\implies a_{1}=0 $ and $2a_{2}=2(a_{0}+a_{1}+a_{2})\implies a_{0}=0 $ and $U=\{a_{2}x^{2}\,,a_{2}\in\Bbb{R}\}=\text{span}\{x^{2}\}$ .

Now use the same method to conclude when $n$ is an odd number . First solve the case for $n=1$ and then for $n=2m-1\,m\geq 2 $.

  • Thank you, for $n=2m-1, m \geq 2$ i've gotten a basis ${x^n-x, x^{n-1}-1, \dots, x^3-x, x^2}.$ I'm also having some trouble on the next part of the exercise, which asks to fill a basis of a given subspace $U$ to the basis of whole $\mathbb{R_n}[x]$. I know the basis of $\mathbb{R_n}[x]$ is ${x^n, x^{n-1}, \dots , x, 1}$, but I'm not sure where to go from here. Would maybe adding ${x,1}$ to the basis of $U$ be enough? – m259 Jun 12 '22 at 10:53
  • 1
    Yeah it should be enough. Just take a look at the matrix, if you add ${x,1}$ to the basis, you actually get $2m+1 $ non-zero rows in the Row Reduced form which is the dimension of the space . (I am still talking about when n is even). You are on the right track !! – Mr.Gandalf Sauron Jun 12 '22 at 11:03
  • Thank you really! I'm just not sure I really understand why it is enough. So basically if $U$ is a subspace of a space $\mathbb{R_n}$ [x] and they have the same dimension, it means that $U = \mathbb{R_n}[x]$? – m259 Jun 12 '22 at 11:49
  • 1
    for finite dimensional vector spaces ,any vector subspace $U$ of a space $V$ satisfies $\dim(U)\leq \dim(V)$ . So if $\dim(U)=\dim(V)$ then it has to be that $U=V$ . – Mr.Gandalf Sauron Jun 12 '22 at 12:10
  • 1
    It is enough precisely because, the non-zero rows of a matrix in Row Reduced form gives you linearly independent vectors. Now if you add ${1,x}$ to the basis for $U$ , then you end up with $2m+1$ non-zero vectors in the Row reduced form of the matrix (Alternatively compute the determinant and show it is non-zero) and hence you have a linearly independent set of $2m+1$ vectors. Now use the fact that if you have a linearly independent set of $|\dim(V)|$ vectors , then the set automatically is a basis for $V$. – Mr.Gandalf Sauron Jun 12 '22 at 12:15
  • 1
    Also a bit of extra knowledge, If you are given a matrix $M$ , and you reduce it to the Row(Column) Reduced form , then the non-zero rows(Columns) give you a basis for the Row(Column) space of the matrix $M$ . It is advisable to learn about elementary row operations because these will let you conclude linear dependence or independence and also a basis for the row(column) spacemore easily , as you have an algorithmic method to calculate the row reduced form. – Mr.Gandalf Sauron Jun 12 '22 at 12:19