So I had an exam two days ago and admittedly, I couldn't do this last question very well. I was wondering if someone could give me a guide on how to do this or how to think about it? I'll post what I did in the exam.


Let $V$ be a vector space over a field $\mathbb{F}$ and $U$ a subspace of $V$. For any $\textbf{v}$ in $V$ we define the set
$$\textbf{v} + U = \{\textbf{v} + \textbf{u}|\textbf{u}\in U\}.$$
We write $V/U$ for the set of all of these sets, $$V/U = \{\textbf{v} + U|\textbf{v} \in V\},$$ and we define addition and scalar multiplication on $V/U$ by
$$(\textbf{v}_1 + U) + \textbf{v}_2 + U) = (\textbf{v}_1 + \textbf{v}_2) + U \quad \text{and} \quad \alpha(\textbf{v}+U) = (\alpha \textbf{v})+U.$$ You may assume these operations are well defined.
i) Find the identity element for addition in $V/U$. Prove that your answer is correct.
ii) Prove that the vector distributive law holds in $V/U$. (That is, the one involving a scalar and two vectors, not two scalars and a vector.)
iii) Let $\textbf{v}_1,\textbf{v}_2$ be in $V$. Show that $\textbf{v}_1 + U = \textbf{v}_2 + U$ if and only if $\textbf{v}_1 - \textbf{v}_2$ is in $U$.
iv) $\textbf{You may now assume}$ that $V/U$ is a vector space over $\mathbb{F}$, with operations defined as above. Prove that if $V$ is finite-dimensional, then $V/U$ is also finite-dimensional and
$$\dim V/U = \dim V - \dim U.$$

My attempts

Firstly, I'm not sure what the "standard" name for these type of subspaces are, as I've never encountered them before, so I was lost on what it even meant to "add" an element to another subspace (referring to $V/U$).
i) I claimed that the identity element is $U$ since for any $\textbf{w} = \textbf{v}_1 + U$ in $V/U$, the additive inverse is $-\textbf{w} = -(\textbf{v}_1 + U) = -\textbf{v}_1 + U$ (by definitions of the addition/scalar multiplication).
$\textbf{w} + (-\textbf{w}) = U +(-1)U = (1-1)U = 0U = U$.
ii) I used the same reasoning (and the definitions to show that $\lambda (\textbf{w}_1 + \textbf{w}_2) = \lambda \textbf{w}_1 + \lambda \textbf{w}_2$.
iii) & iv) I had no clue how to start or think about.

Thinking about it again, I'm confused how $U$ can be an identity element of a set... when $U$ is itself a set (with elements in it)..

  • $\begingroup$ In an algebraic structure like a group, ring, field, vector space, etc., the members can be any objects at all. A vector sub-space U of the vector space V can also be the 0 vector of some other vector space. $\endgroup$ – DanielWainfleet Jun 26 '17 at 5:26

The sets $\mathbf{v} +U$ are not subspaces technically (they are not closed under addition), but the standard name for them is "cosets" (of $U$). Your intuition for $(i)$ and $(ii)$ are correct.

For $(iii)$, suppose first that $\mathbf{v_0} = \mathbf{v}_1 - \mathbf{v}_2 \in U$. Then for any $\mathbf{u} \in U$, we have $\mathbf{v}_2+\mathbf{u} = \mathbf{v}_1 + (\mathbf{u}-\mathbf{v_0})$, which shows that $\mathbf{v}_2 + U \subseteq \mathbf{v}_1 + U$, and also $\mathbf{v}_1 + \mathbf{u} = \mathbf{v}_2 + (\mathbf{u}+\mathbf{v}_0)$, which shows that $\mathbf{v}_1+U \subseteq \mathbf{v}_2+U$. Conversely, suppose that $\mathbf{v}_1+U=\mathbf{v}_2+U$. Take any $\mathbf{u} \in U$. Then $\mathbf{v}_1+\mathbf{u} \in \mathbf{v}_1+U$, and so also by assumption, it is in $\mathbf{v}_2+U$. So there is some other $\mathbf{u}' \in U$ so that $\mathbf{v}_1+\mathbf{u} = \mathbf{v}_2+\mathbf{u}'$. Then $\mathbf{v}_1-\mathbf{v}_2=\mathbf{u}'-\mathbf{u} \in U$.

Part $(iv)$ is a well-known result that is known by the name "Rank-Nullity Theorem". A more abstract version is also known by the name "First Isomorphism Theorem". You should be able to find a proof by just searching on Google.


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.