One of my homework questions is: Let $v_1, v_2,\dots ,v_n$ be a spanning set (in particular, a basis) in an inner product space $V$ . Prove that
a) If$(x,v)=0$ for all $v\in V$, then $x=0$;
b) If$(x,v_k)=0$ $\forall k$, then $x=0$;
c) If $(x,v_k) = (y,v_k)$ $\forall k$, then $x=y$.

I initially thought for part a) to just say do a proof where $v$ is set to $x$, which would ultimately get $(x,x)=0$, so $x$ would also have to be equal to $0$. But upon thinking about it, would that proof work since it is not specified that $x$ is in $V$, or can you still go about it that way? and if not, what step should I take first?

I am very lost on how to start part b, but was thinking if I am set on part a then b should follow more naturally. Does anyone have a recommendation for a starting point?

And for c, what I thought to do was just go by the definitions that $(x,vk)$ is the same as $||x||||v_k||$, and so we then have $||x||||v_k||=||y||||v_k||$. Can I then cancel out the $||v_k||$ to get $||x||=||y||$? Even this would not be super clean because it ends with $\pm x= \pm y$. Any suggestions how to clear this up/ get on the right track?

Sorry this is so long and detailed. Just trying to understand as thoroughly as possible.

  • $\begingroup$ Not every spanning set is a basis... $\endgroup$
    – DonAntonio
    Sep 29 '16 at 21:57


B: Let x = x_i*v_i and compute ||x|| using the triangle inequality. What does this tell you about x?

C: (x,v_k)=(y,v_k) if and only if (x-y,v_k)=0. Then use B.

  • $\begingroup$ How would you use x=x_iv_i with the triangle inequality? I guess, since the triangle inequality is ||x+y|| <= to ||x|| + ||y|| where would the x_iv_i fit in? $\endgroup$ Sep 29 '16 at 22:38

For the a)

The inner product is defined as a function $\varphi: V \times V \to k$ where $k$ is the field that makes $V$ a vector space. (In your case, probably $k=\Bbb{R}$ or $k=\Bbb{C}$ ). Therefore $(x,v)$ with $x \notin V$ doesn´t make any sense in this case. So $x \in V$ and your reasoning is perfectly well done.

For b)

Hint: Since $v_1,...,v_n$ is a spanning set, in particular $x=\sum \alpha_i v_i$, therefore $(x,x)=(x,\sum \alpha_i v_i)=\sum \alpha_i^*(x,v_i)$

(here, $\alpha_i^*$ denotes the complex conjugate of $\alpha_i$). Try finishing the proof.


$(x,v_k)=(y,v_k)$ for all $k$ if and only if $(x,v_k)-(y,v_k)=0$ for all $k$ if and only if $(x-y,v_k)=0$ for all $k$ (using linearity of the inner product.) Can you finish the proof using a) and b)?

  • $\begingroup$ I understand the reasoning for part c and see how it follows from a and b- clever. With the explanation of part b, I am still sort of unsure. Is there another way you could explain this same concept? $\endgroup$ Sep 29 '16 at 22:32
  • $\begingroup$ Can you be a little bit more precise in what makes you unsure about part b)? $\endgroup$ Sep 29 '16 at 22:39
  • $\begingroup$ I see that it stats similarly to the proof of a, but I am not following why we are using the complex conjugate of a exactly. $\endgroup$ Sep 29 '16 at 22:42
  • $\begingroup$ Because the inner product is not exactly linear in the second coordinate: the coefficient multiplying the vector in the second coordinate, comes out as it´s complex conjugate. If you are working on an $\Bbb{R}$-vector space then $\alpha = \alpha^*$ for any $\alpha \in \Bbb{R}$. But if you are working on a $\Bbb{C}$-vector space then $\alpha$ and $\alpha^*$ are not neccesairily the same. $\endgroup$ Sep 29 '16 at 22:48
  • $\begingroup$ Ah ok. So then would you prove this the same way as the first except noting that there is a chance of complex. So basically saying x= av_i and then using that v_i point so then it sort of cancels the same way as the first one? If that makes sense. $\endgroup$ Sep 29 '16 at 23:01

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.