I need to proof the following theorem and I would appreciate if you could check if my proof is sound. I'm generally new to proofs, so any recommendations on proof strategy/style are more than welcome. Thank you.


Let $S=\{\textbf{v}_1, \textbf{v}_2, ..., \textbf{v}_k\}$ be a linearly independent subset of some vector space $V$. Let $\textbf{u} \in V$. If $\textbf{u} \notin \operatorname{span}(S)$, then $S \cup \{\textbf{u}\}$ is linearly independent.

$\textbf{Proof:}$ I use a contrapositive proof, i.e. $\lnot Q \Rightarrow \lnot P $. Suppose $S \cup \{ \textbf{u} \}$ is linearly $\textbf{dependent}.$ Then, by definition of linear dependence, there exist $\alpha_1, \alpha_2, ..., \alpha_{k+1} \in \mathbb{R}$ not all zero s.t.

$$\sum_{i=1}^{k+1}\alpha_i v_i = 0$$ $$= \sum_{i=1}^{k}\alpha_i v_i + \alpha_j u = 0$$ where, without loss of generality, $\alpha_j u= \alpha_{k+1}v_{k+1}$.

Then it follows, $$\alpha_j u = -\sum_{i=1}^{k}\alpha_i v_i$$ $$= u =\sum_{i=1}^{k}\alpha_i^{'} v_i$$ where $\alpha_i^{'} = \frac{-\alpha_i}{\alpha_j}$, for $\alpha_j \neq 0$, implying that $u \in \operatorname{span}(S) \cup \{u\}$.

Thus, we can conclude that if $u \notin \operatorname{span}(S)$, $S \cup \{u\}$ is linearly independent. This completes the proof.

  • $\begingroup$ A clear way to know that your proof is incomplete is that you haven’t used the linear independence of $S$ supposistion in your proof. The good news is, other than some weird notational quirks, your proof is almost complete. You have divided by $\alpha_j$ stating that $\alpha_j$ is non-zero, but younhavent justifies why it is non-zero. Also, you say at the end that $u$ is in Span$(S) \cup \{u\}$, but this is true for every $S$, so I’m unsure what you mean here? $\endgroup$ – Adam Higgins Dec 27 '18 at 12:15

Your proof is essentially correct, but I would use more expressive notations, along the following lines:

Suppose$S\cup\{\mathbf{u}\}$ is linearly dependent. Then there exist coefficients $\alpha_1,\dots,\alpha_k\:$ and $\:\beta$, not all $0$, such that $$\sum_{i=1}^k\alpha_i\,\mathbf{v}_i+\beta\, \mathbf{u}=0.$$ Observe that $\beta\ne 0$, for otherwise, $S$ would be linearly dependent, contrary to the hypotheses. So we can deduce that $$\mathbf u=\sum_{i=1}^k\Bigl(-\tfrac{\alpha_i}\beta\Bigr)\mathbf{v}_i,$$ which means$\;\mathbf u\in\operatorname{span} S$.


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.