Let $(\mathcal{V},\langle\cdot,\cdot\rangle_1)$ be a Hilbert space. If we change the inner product, can we then say anything about if that is a Hilbert space as well, i.e. when is $(\mathcal{V},\langle\cdot,\cdot\rangle_2)$ also a Hilbert space? Is it enough that $\langle\cdot,\cdot\rangle_2$ defines an inner product on $\mathcal{V}$?

  • $\begingroup$ In finite dimension, yes. $\langle \cdot, \cdot \rangle_1$ and $\langle \cdot,\cdot \rangle_2$ induces norms $\|\cdot \|_1$ and $\|\cdot\|_2$, which will be equivalent. For infinite dimensional spaces I do not know an answer right now. $\endgroup$ – Ivo Terek Feb 20 '15 at 20:23

No, it is not automatic that the vector space endowed with another inner product will be complete.

For example, let $H$ be any infinite dimensional Hilbert space,let $B$ be a Hamel basis of $H$ and consider the inner product on $H$ for which $B$ is an orthonormal basis. Then with that inner product $H$ is not complete.

  • $\begingroup$ (+1) Thanks a lot for the answer. Do you know of any sufficient conditions that would do it? For example, what if $\langle\cdot,\cdot\rangle_2$ was coercive with respect to the $\|\cdot\|_1$-norm? That it is to say that for all $v\in\mathcal{V}$ it holds for some $\alpha>0$ that $\langle v,v\rangle_2 \geq \alpha \|v\|_1^2$. $\endgroup$ – Eff Feb 20 '15 at 20:40
  • $\begingroup$ Under that condition you in fact that that $\Vert v\Vert_2\geq a\Vert v\Vert_1$, and then a Cauchy sequence wrt the new norm is also a Cauchy sequence wrt the old one, so yes that works. $\endgroup$ – Mariano Suárez-Álvarez Feb 20 '15 at 20:55
  • $\begingroup$ Notice that the correct condition is «there is an $a>0$ such that for all $v\in V$ we have $\langle v,v\rangle_2\geq a\Vert v\Vert_1^2$» and not what you wrote: you have to be careful about where you place quantifiers in what you write! $\endgroup$ – Mariano Suárez-Álvarez Feb 20 '15 at 20:56
  • $\begingroup$ Thanks a lot Mariano. You are completely correct of course, I was not careful writing the comment. I do understand the difference :-). $\endgroup$ – Eff Feb 20 '15 at 21:32

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.