Let's call $\Bbb{R[x]}_4$ a linear space of all polynomials of 4-th degree with real coefficients. We are given functionals $\phi_j \in (\Bbb{R[x]}_4)^*$:

  • $\phi_0(p)=p(-1)$
  • $\phi_1(p)=p(0)+p(1)$
  • $\phi_2(p)=p(0)-p(1)$
  • $\phi_3(p)=p(2)$
  • $\phi_4(p)=p(-2)$

The problems asks us to prove that given functionals are basis for $(\Bbb{R[x]}_4)^*$. Then we are to find coefficients $a_0,a_1,a_2,a_3,a_4 \in \Bbb{R}$ such that if given functionals are dual basis to polynomials $p_0,p_1,p_2,p_3,p_4$ and $p(t)=t$ then $p=\sum_{i=0}^4a_ip_i$.

When it comes to the 1st part of this problem I thought of picking functionals $\beta_0=p(0)$, $\beta_1=p(1)$, $\beta_2=p(-1)$, $\beta_3=p(2)$, $\beta_4=p(-2)$, and these functionals must be dual basis for Lagrange basis constructed from them. Then you can show that every $\beta$ function can be written as a linear combination of $\phi$ functions, so their span is equal, and as we know that $\beta$ functions must be basis of $(\Bbb{R[x]}_4)^*$ so the same must be true for $\phi$ functions.

When it comes to the 2nd part of the problem I thought of computing the Lagrange basis for $\beta$ functions and then do some manipulations to it to get the Lagrange basis of $\phi$ functions and from that point I would compute the coordinates of $p$ in that basis which would give the answer. But this solution seems too complex and too hard to compute. Is there a simpler one? Also, are my earlier arguements correct?

  • $\begingroup$ do you already experimented on ${\Bbb{R}}[x]_2$? $\endgroup$ – janmarqz Jan 6 '15 at 2:39
  • $\begingroup$ what do you mean? $\endgroup$ – qiubit Jan 6 '15 at 2:57
  • $\begingroup$ that you are trying to solve for ${\Bbb{R}}[x]_4$, try for ${\Bbb{R}}[x]_2$ or/and ${\Bbb{R}}[x]_3$ $\endgroup$ – janmarqz Jan 6 '15 at 3:03

Firstly, you want to show that the functionals are linearly independent. Suppose $ \phi_c = \sum_i c_i \phi_i = 0 $, then $ \phi_c( x (x-1) (x-2) (x+2) ) = 0 $ but also $ \phi_c( x (x-1)(x_2)(x+2) ) = c_0 $, by directly checking explicitly what it evaluates for the other basis functionals. So $ c_0 = 0 $. Similarly conclude that the other $ c_i = 0 $.

Secondly, you want to show that they span the vector space. Can you appeal to the theorem that the dual space has the same dimension?

Finally, if $ t = \sum_i a_i p_i(t) $, then $ \phi_j(t) = a_j $ by the fact that $ \phi_j (p_i) = \delta_{ij} $. So just evaluate the functionals on $ t $.

You might also try to calculate explicitly the dual basis for practice. For example $ p_0 = \frac{-1}{6} x(x-1)(x-2)(x+2) $ evaluates to zero at the right places, and is scaled so that $ \phi_0(p_0) = 1 $. Then you can check explicitly if the $ a_i $ are correct.


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.