The Statement

Show that if a set of vectors is linearly independent, then any vector in the span of that set has a unique representation as a linear combination of these vectors.

My Proof

I'm going for a proof by contrapositive.

Let $S=(v_{1},v_{2},\dots,v_{n})$ be our set of vectors.

Let $x_{0}$ be a vector in the span of S. If it does not have a unique representation, it can be written in atleast two distinct ways:

$x_{0}= a_1v_{1}+a_{2}v_{2}+\dots a_{n}v_{n}$

$x_{0}= b_{1}v_1+b_2v_2+\dots+b_nv_n$

$1)$For atleast one $b_j$ for $1\leq j\leq n$ in those linear combinations it will be the case that $b_j \neq a_j$

$0 = x_0-x_0=(a_1-b_1)v_1+(a_2-b_2)v_2+\cdots+(a_n-b_n)v_n$

By fact 1 that means atleast one of these coefficients is not zero, which means S is linearly dependent.

My Question

I'm pretty happy with my solution(which I think is correct). I was wondering if for learning sake anyone could provide a direct proof(or a proof via another method if a direct is impossible)?


This proof works fine. Here's a "direct" one.

Suppose that $$ \alpha_1 v_1+\dotsb+\alpha_n v_n=\beta_1 v_1+\dotsb+\beta_n v_n $$ Then $$ (\alpha_1-\beta_1)v_1+\dotsb+(\alpha_n-\beta_n) v_n=\vec 0 $$ But $\{v_1,\dotsc,v_n\}$ is linearly independent so $\alpha_j-\beta_j=0$ for each $1\leq j\leq n$. Hence $\alpha_j=\beta_j$ for $1\leq j\leq n$.


A direct method is of course possible, and it is pretty straightforward.

Let $x \in L(\{v_1, \ldots, v_n\})$. Then, there are $a_1, \ldots, a_n \in \mathbb K$, such that $x = \sum_i a_i v_i$. Let $x = \sum_i b_i v_i$ be another representation of $x$ in those vectors. We have:

$$\sum_i (a_i - b_i)v_i = 0$$

As the $v_i$'s are linearly independent, it follows: $a_i - b_i = 0$, $\forall \ i$. Which shows that the $b_i$'s are exactly the $a_i$'s.

This demonstrates the uniqueness of the $a_i$'s. Essentially, this is the same as your argument, but working by contradiction wasn't necessary.


Your proof is great. A direct proof is very similar. Suppose we have $S=(v_1,\dots,v_n)$, a linearly independent set of vectors. Then suppose we have two ways to write $x_o\in \text{sp}(S)$,


$$\implies (a_1-b_1)v_1+\dots+(a_n-b_n)v_n=0$$

because the set is linearly independent, this means

$$a_i-b_i=0\implies a_i=b_i\text{ ($0\leq i\leq n$)}$$

So our way of writing $x_o$ is unique.


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.