I am given the following

$V = \mathbb R^4$

$W = \{(w,x,y,z)\in \mathbb R^4|w+2x-4y+2 = 0\}$

I have to prove or disprove that $W$ is a subspace of $V$.

Now, my linear algebra is fairly weak as I haven't taken it in almost 4 years but for a subspace to exist I believe that:

1) The $0$ vector must exist under $W$

2) Scalar addition must be closed under $W$

3) Scalar multiplication must be closed under $W$

I don't think the first condition is true because if I were to take the vector, there is no way I can get the zero vector back. Is that correct or am I doing something very wrong?

  • $\begingroup$ Why the downvote? $\endgroup$ – mrnovice Mar 5 '17 at 9:13
  • 1
    $\begingroup$ Possible duplicate of How to prove that a subspace is in a vector space? $\endgroup$ – Dietrich Burde Mar 5 '17 at 9:15
  • $\begingroup$ No. I did not know that question existed. Sorry. Also, that question uses transposes. Mine doses not. $\endgroup$ – Future Math person Mar 5 '17 at 9:17
  • $\begingroup$ Yeah. $0$ vector is not in $W$, so its not even a vector space. $\endgroup$ – rookie Mar 5 '17 at 9:18
  • $\begingroup$ It is a bit different question, you can figure out infinite number of simple examples of a set which is not a vector space :-) For this one, you are right, the first condition fails. $\endgroup$ – Mirek Olšák Mar 5 '17 at 9:21

You're right in what you think, but as how you argue that I am not sure what you mean by "getting the zero vector back"? Back...from where?

Anyway, if we take $\;(w,x,y,z)=(0,0,0,0)\;$ , this vector belongs to $\;W\;$ iff

$$0+2\cdot0-4\cdot0+2=0\iff 2=0$$

and since the last equality is false we get that $\;(0,0,0,0)\notin W\implies W\;$ is not a subspace and you were right.

  • $\begingroup$ I mean like if I plug in 0 into the vector, I don't get 0. I get 2 if w x and y are all 0. $\endgroup$ – Future Math person Mar 5 '17 at 9:20
  • $\begingroup$ @SubhashisChakraborty you don't plug in 0 into a vector, you plug in the zero vector into the defining equation of $W$. (An important notational difference.) $\endgroup$ – Paŭlo Ebermann Mar 5 '17 at 17:21
  • $\begingroup$ Ahh thanks for clearing that up. $\endgroup$ – Future Math person Mar 5 '17 at 23:17

No, it is not a subspace.

It is because you have to verify your first point: "$0\in W$".

But $(0,0,0,0)\notin W$ because

$$0+2\times 0-4\times 0+2\ne 0.$$

  • $\begingroup$ Thanks. My linear algebra is fairly rusty so I just needed to be sure. $\endgroup$ – Future Math person Mar 5 '17 at 9:27

Let me give you a more general result: Consider a linear system of equations $$\begin{cases} a_{11}x_1 + a_{12}x_2 + \ldots + a_{1n}x_n &= b_1\\ a_{21}x_1 + a_{22}x_2 + \ldots + a_{2n}x_n &= b_2\\ &\vdots&\\ a_{m1}x_1 + a_{m2}x_2 + \ldots + a_{mn}x_n &= b_m \end{cases},$$ with $m \leq n$. Suppose one of the $b_i \neq 0$, then clearly this system of equations does not form a subspace, since in the $i$th equation, we would find that (filling in the zero-vector $(0, 0, \ldots, 0)$ that $0 = b_i \neq 0$, which is not possible.

However, if all of the $b_i$ are zero, then this forms a subspace: clearly the zerovector is a solution to all equations and hence to the system. If both $(y_1, \ldots, y_n), (z_1, \ldots, z_n)$ are solutions, then so is $(y_1 + z_1, \ldots y_n + z_n)$ (in order to see this: we fill in this vector and use distributivity of the product over the summation). Analogously, $(\lambda y_1, \ldots, \lambda y_n)$ for some $\lambda \in \mathbb{R}$ will form a solution (note that $(\lambda y_1, \ldots, \lambda y_n) = \lambda (y_1, \ldots, y_n)$).

  • $\begingroup$ This makes a lot of sense. Thanks Why is it required to assume m is less than or equal to n though? $\endgroup$ – Future Math person Mar 5 '17 at 9:56
  • $\begingroup$ Note that there are only $n$ variables which we want to find. So if we take $m > n$, we have an overdetermined system of linear equations. If there would be some $b_i \neq 0$, it is possible that there is no solution (however, then we have the empty set, which is not a vectorspace either, so in this case it does not really matter). In case all the $b_i = 0$, then there will be equations which are linearly dependent, so we have some redundant equations. Therefore, we can assume $m \leq n$. $\endgroup$ – Student Mar 5 '17 at 10:29
  • $\begingroup$ However, you might need some result from linear algebra to show this. The dimension of $\mathbb{R}^n$ is $n$, so if we would write this system of linear equations in a matrix, then we could rowreduce this matrix and because of the dimension, we will always have that the rank of the matrix would be $n$ or less. This means that if we would have $m$ equations (which translates to $m$ rows in the matrix), we would find zerorows (and this corresponds to redundant equations in the system). $\endgroup$ – Student Mar 5 '17 at 10:31
  • $\begingroup$ The keypoint is that we can assume, without loss of generality, that $m \leq n$ (however, without this assumption, the things I wrote are still valid). $\endgroup$ – Student Mar 5 '17 at 10:32
  • $\begingroup$ That makes sense. Thanks a lot :) . $\endgroup$ – Future Math person Mar 5 '17 at 10:40

I feel I should indicate that condition 2 and condition 3 fails as well. The vector $v = (2, 0 , 0 , 0)$ belongs in the subset, but neither $2v$ nor $v+v$ belongs in the subset.

It is thus not a subspace.

  • $\begingroup$ Yes I realize that as well. Thank you. $\endgroup$ – Future Math person Mar 5 '17 at 23:18

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.