Let $B = \{v_1, v_2, ..., v_k\} \in \mathbb{R}^n$ and $B$ is linearly independent. Prove that if $\overrightarrow{v} \not \in \text{span } B$ then $\{v_1, v_2, ..., v_k, v\}$ is linearly Independent.


Assume $\{v_1, v_2, ..., v_k, v \}$ is linearly dependent. Then because $\{v_1, v_2, ..., v_k\}$ is linearly independent from hypothesis, it must follow that $v \in \text{span } B$, contradiction.

Is this correct? If not, can you tell me why?


We have $c_1v_1 + c_2v_2 + ... + c_kv_k = 0$ for $c_1 = c_2 = ... = c_k= 0 $.Since $B + v$ is linearly dependent, we must have

$c_1v_1 + c_2v_2 + ... + c_kv_k + xv = 0$ where $x \ne 0$, thus it follows that

$-c_1v_1/x - c_2v_2/x - ... - c_kv_k/x = v$ thus $v \in \text{span } B$

  • $\begingroup$ It's not wrong, but I think you're a bit brief. I would personally prefer if you expand a bit on why this means $v\in \operatorname{Span} B$. Something with an explicit linear combination. $\endgroup$ – Arthur Jan 24 '17 at 6:59
  • 1
    $\begingroup$ @Arthur, how is it with the Edit? $\endgroup$ – Aditya Kalra Jan 24 '17 at 7:05

If $\{v_1, v_2, ..., v_k, v\}$ is linearly dependent there exist scalars $\lambda_i$ not all null such that $$\lambda_1v_1+\lambda_2 v_2+\cdots \lambda_k v_k+ \lambda_{k+1}v=0$$ As $\{v_1, v_2, ..., v_k\}$ is linearly independent, necessarily $\lambda_{k+1}\ne0.$ So, $$v=\left(-\frac{\lambda_1}{\lambda_{k+1}}\right)v_1+\left(-\frac{\lambda_2}{\lambda_{k+1}}\right)v_2+\cdots +\left(-\frac{\lambda_k}{\lambda_{k+1}}\right)v_k\in \text{Span }B\text{ (contradiction).}$$

  • $\begingroup$ I'm probably missing something here, but what exactly is being contradicted and how does it prove anything about the original problem statement? $\endgroup$ – infinitylord Jan 24 '17 at 7:42
  • $\begingroup$ $$v \not \in \text{Span } B\Rightarrow\{v_1, v_2, ..., v_k, v\}\text{ is linearly independent}$$ is equivalent to $$\{v_1, v_2, ..., v_k, v\}\text{ is linearly dependent}\Rightarrow v \in \text{Span } B.$$ $\endgroup$ – Fernando Revilla Jan 24 '17 at 7:50

It's much better with the edit. While all the elements are there, I would perhaps phrase that proof a bit differently: start with $c_1v_1 + c_2v_2 + \ldots + c_kv_k + xv = 0$ where not all coefficients are $0$. Argue that $x\neq0$ (since otherwise we would have $c_1v_1 + c_2v_2 + \ldots + c_kv_k = 0$, which implies that all coefficients are $0$), which means that $-c_1v_1/x - c_2v_2/x - \ldots - c_kv_k/x = v$ makes sense.


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.