I was working on a solution to this question (linked below), but wanted to understand it from a more abstract perspective (not considering an explicit counterexample) because I got it wrong initially.

Prove or give Counterexample: Is this a Basis.

Problem Statement

Prove or give a counterexample: If $𝑣_1,𝑣_2,𝑣_3,𝑣_4$ is a basis of 𝑉 and 𝑈 is a subspace of 𝑉 such that $𝑣_1,𝑣_2∈𝑈$ and $𝑣_3,𝑣_4∉𝑈$, then $𝑣_1,𝑣_2$ is a basis of 𝑈.


Suppose that $v_1,v_2$ is a basis of $U$. Then $span(v_1,v_2) = U$ and $v_1,v_2$ are linearly independent, by the definition of basis.

We know $v_1,v_2$ is linearly independent because removing vectors from a linearly independent list results in a shorter linearly independent list of vectors. In this case, start with $v_1,v_2,v_3,v_4$ and remove vectors to get that $v_1,v_2$ is linearly independent.

So, we just have to check if $v_1,v_2$ spans $U$.

If $v_1,v_2$ spans $U$, then no vector $v_n∉span(v_1,v_2)$ exist such that $v_n \in U$.

I reasoned here that the only vectors in the larger space $V$ that were not in $U$ were $v_3$ and $v_4$, meaning that $v_n \in span(v_3,v_4)$.

Here is my mistake: I reasoned that because $v_3 \notin U$ and $v_4 \notin U$, then $v_n \notin U$.

From this, I incorrectly concluded $span(v_1,v_2) = U$ and that $v_1,v_2$ was a basis of $U$

From the counterexample, I realized it was possible for $v_3+v_4 \in U$ despite $v_3 \notin U$ and $v_4 \notin U$.

I want to more clearly understand how two linearly independent vectors ($v_3,v_4$ in this example) can not be in some subspace, while their sum can exist in that subspace.


  • $\begingroup$ Note: $(1,1)$ and $(1,-1)$ are not in the $1$-dimensional space spanned by $(1,0)$, but their sum is $\endgroup$ – J. W. Tanner Jul 25 at 16:21
  • $\begingroup$ can you already think of a counter example in say, $\Bbb{R}^2$? also could you be more specific about what you're looking for, because I'm not sure what you mean by "I want to more clearly understand... " Because to me a counterexample already shows you that this is a misconception which you should just erase from memory $\endgroup$ – peek-a-boo Jul 25 at 16:23
  • $\begingroup$ Got it, I think I just got hung up on the misconception that $v_3, v_4$ not existing in $U$ did not imply that all their linear combinations also did not exist in $U$ $\endgroup$ – Richard K Yu Jul 25 at 16:25

Is a concrete example enough? Consider $V=\mathbb{R}^2$ with the usual vector space structure. The vectors $(0,1)$ and $(1,0)$ are linearly independent. Neither belongs to the subspace $U = \{(x,x) \mid x \in \mathbb{R}\}$ determined by the line $y=x$, but the sum $(0,1)+(1,0)=(1,1)$ certainly belongs to $U$.

  • $\begingroup$ I think this helps clear it up. The idea that I started to see with the counterexamples is that the dimension of individual vectors is not necessarily the same as the dimension of their sum. $\endgroup$ – Richard K Yu Jul 25 at 16:32

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.