I need to find an inner product such that given a set $S$ of linearly independent vectors in a Hilbert space $H$, $S$ will be orthogonal with these product.

I thought Gram -Schmidt Process would help but it's not, because for the process you already have the inner product.


Start with a set $\{ x_1,\cdots,x_n\}$. Define $$ U : \mathbb{C}^{n} \rightarrow H $$ by $$ U(\alpha_1,\cdots,\alpha_n) = \alpha_1 x_1+\cdots+\alpha_n x_n $$ Let $P$ the be orthogonal projection of $H$ onto the closed subspace spanned by $\{x_1,\cdots,x_n\}$. Define $$ (x,y)_{\mbox{new}} = (U^{-1}Px,U^{-1}Py)_{\mathbb{C}^{n}}+((I-P)x,(I-P)y)_{H}. $$ Because $(I-P)x_k=0$ for $1 \le k \le n$ and $U^{-1}x_k$ is the $k$-th standard basis element in $\mathbb{C}^{n}$, then $(x_j,x_k)_{\mbox{new}}=\delta_{j,k}$ and $(x_j,(I-P)y)_{\mbox{new}}=0$ for all $1 \le j \le n$ and $y\in H$.

  • $\begingroup$ Excelent! Totally Works! $\endgroup$ – Tom Builder Jun 5 '15 at 0:03
  • $\begingroup$ I just have a question, why does the second term in the definition of the inner product have to be? I think it still works even without that! $\endgroup$ – Tom Builder Jun 5 '15 at 0:18
  • 1
    $\begingroup$ @TomBuilder : The second part extends the inner product to all of $H$ in such a way that the vectors $x_1,\cdots,x_n$ are orthogonal to the original orthogonal completment of the subspace generate by the $x_k$. Without the second term, $(\cdot,\cdot)_{\mbox{new}}$ would not be positive definite. By the way, the original inner product and this one generate the same topology on $H$. $\endgroup$ – DisintegratingByParts Jun 5 '15 at 0:58

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.