We work in an infinite Hilbert space $H$.

Let $(x_k)$ an arbitrary linearly independent sequence, then we define $$e_1=\frac{x_1}{\lVert x_1\Vert}\quad\text{and}\quad e_k=\frac{x_k-\sum_{j<k}\langle x_k, e_j\rangle e_j}{\lVert x_k-\sum_{j<k}\langle x_k, e_j\rangle e_j \Vert}\quad (k\ge 2).$$ For construction $(e_k)$ is an ortonormal sequence.

I must prove that $$\text{span}(e_1, \dots, e_k)=\text{span}(x_1,\dots, x_k)\quad\forall k\ge 1\tag 1.$$

First, for indution on $k$ we prove that $$(e_1,\dots, e_k)\subseteq \text{span}(x_1,\dots, x_k)\tag 2.$$ For $k=1$ we have that $(e_1)\subseteq\text{span}(x_1)$ for the definition of $e_1$. Suppose that $(e_1,\dots, e_k)\subseteq \text{span}(x_1,\dots, x_k)$ is true for same $k$ and we prove that $$(e_1,\dots, e_{k+1})\subseteq \text{span}(x_1,\dots, x_{k+1}).$$

At this point I have problems, because I don't know how to do it after doing the following:

$$\{e_1,\dots, e_{k+1}\}=\{e_1, \dots, e_k\}\cup \{e_{k+1}\}\subseteq\text{span}(x_1,\dots, x_k)\cup \{e_{k+1}\}.$$

From $(2)$ follow that $$\text{span}(e_1,\dots, e_k)\subseteq \text{span}(x_1,\dots, x_k).$$

On the other hand the vectors $e_1, \dots, e_k$ are linearly independent since they are ortogonal. Therefore $$\dim\text{span}(e_1,\dots, e_k)=k=\dim\text{span}(x_1,\dots, x_k),$$ from which follows $(1)$.

Could you help me complete the proof of $(2)$?


1 Answer 1


For your inductive step, you have that $$ e_{k+1}=\beta_{k+1}x_{k+1}-\sum_{j=1}^k\beta_je_j $$ for certain coefficients $\beta_1,\ldots,\beta_{k+1}$. The inductive hypothesis gives you that $$ \sum_{j=1}^k\beta_je_j=\sum_{j=1}^k\gamma_jx_j $$ for certain coefficients $\gamma_1,\ldots,\gamma_k$. Then $$ e_{k+1}=\beta_{k+1}x_{k+1}-\sum_{j=1}^k\gamma_jx_j\in\operatorname{span}\{x_1,\ldots,x_{k+1}\}. $$ That completes the induction.


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.