I know that not all norms are induced by any inner product.

But if we have an inner product, $\langle\cdot,\cdot\rangle$ we can define a norm $||v||=\sqrt{\langle v,v\rangle}$.

My question is, can we somehow reconstruct the inner product from this norm? I.e. can we define an inner product in terms of a norm?

If not, can we construct it if we assume some additional structure on the normed space induced by this inner product?


Indeed we can. Note that we have $\langle v, v \rangle = \Vert v \Vert^2$. This is the quadratic form associated to the bilinear form $\langle \cdot, \cdot \rangle$. However, by the polarisation formula (see https://en.wikipedia.org/wiki/Polarization_identity), we can recover the bilinear form from its associated quadratic form.

As AGF remarked, we have a criterion to decide when a norm is induced by an inner product. Namely, it is induced by an inner product iff the following identity holds true for all $x,y$ in our vector space $$ \Vert x + y \Vert^2 + \Vert x-y \Vert^2 = 2(\Vert x \Vert^2 + \Vert y \Vert^2) $$ This is the so-called parallelogram identity with you can find here https://en.wikipedia.org/wiki/Parallelogram_law

  • $\begingroup$ Might want to add that a norm has an associated inner product exactly when it satisfies the parallelogram law. $\endgroup$ – AGF Feb 24 '19 at 11:02
  • $\begingroup$ @AGF That is a great remark. I assumed the OP was aware of that (from his first sentence), but surely it might be helpful for others. $\endgroup$ – Severin Schraven Feb 24 '19 at 11:03
  • 1
    $\begingroup$ One can also use the polarisation identity to show that a linear isometry (i.e. a linear map such that $\|Tx\| = \|x\|$) must preserve inner products. So, if a linear map preserves distances, it also preserves angles. $\endgroup$ – Theo Bendit Feb 24 '19 at 11:09
  • $\begingroup$ @TheoBendit That is a nice fact. I never thought about this, but using the polarisation identity is pretty elegant (I would have tried to use functional analysis tools to show it). $\endgroup$ – Severin Schraven Feb 24 '19 at 11:18

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.