1

Sorry if the title sounds convoluted, I couldn't find any better.

In $R^d$, let $(e_1,\ldots, e_d)$ be a basis. Show there exists $(a_1,\ldots, a_d)$ d vectors of $R^d$ such that $$\forall x \in R^d, x=\sum_{i=1}^{d} \langle x, e_i \rangle a_i $$

I tried to mimic the proof of Gram-Schmidt orthogonalization, but failed. After that, I introduced a linear mapping of which it suffices to prove the injectivity but it yields awful computations...

Thanks for your help.

Gabriel Romon
  • 35,428
  • 5
  • 65
  • 157

1 Answers1

0

Solution 1:

Consider the $d$ following linear forms $\begin{array}{ccccc} f_i & : & E & \to & \mathbb R \\ & & x & \mapsto & \langle x|e_i\rangle \\ \end{array}$

We first prove that $(f_1,...,f_d)$ is a linearly independent family (hence a basis) of the dual space $E^*$: it suffices to prove that $\operatorname{rank}(f_1,...,f_d)=d $.

But $$\operatorname{rank}(f_1,...,f_d)=d-\dim(\cap_{i=1}^d \ker f_i)$$ $$\operatorname{rank}(f_1,...,f_d)=d-\dim(\cap_{i=1}^d \{e_i\}^\bot)$$

Since $(e_1,...,e_d)$ is a basis, $$\operatorname{rank}(f_1,...,f_d)=d-0=d$$

Therefore $(f_1,...,f_d)$ is basis of $E^*$, which has a unique "prebasis" $(a_1,...,a_d)$ where $(a_1,...,a_d)$ is a basis of $E$ and $a_i^*=f_i$.

Now, $$\forall x \in E, x=\sum_{i=1}^{d} a_i^*(x) a_i=\sum_{i=1}^{d} f_i(x) a_i=\sum_{i=1}^{d} \langle x|e_i\rangle a_i$$

This achieves the proof.

Solution 2

Here : Proving that $\exists A, \forall X, X^TGA=0 \Rightarrow A=0 $

Gabriel Romon
  • 35,428
  • 5
  • 65
  • 157