Subspace of Euclidean Space is Closed
Theorem
Let $\R^n$ be a Euclidean space.
Every subspace of $\R^n$ is closed in $\R^n$.
Proof
First we note that from Euclidean Space is Normed Vector Space, $\R^n$ has a norm $\norm {\, \cdot \,}$.
Let $V$ be a subspace of $\R^n$.
Let $\set {\mathbf v_1, \ldots, \mathbf v_k}$ be a basis for $V$.
Extend this to basis $\set {\mathbf v_1, \ldots, \mathbf v_k, \mathbf v_{k + 1}, \ldots, \mathbf v_n}$ for $\R^n$.
By using Gram-Schmidt procedure, there exists a set of orthonormal vectors $\set {\mathbf u_1, \ldots, \mathbf u_n}$ such that:
- $\forall k \in \N_{> 0} : k \le n : \map \span {\mathbf v_1, \ldots, \mathbf v_k} = \map \span {\mathbf u_1, \ldots, \mathbf u_k}$
where $\span$ stands for the linear span.
Let $A \in \R^{\paren {n - k} \times n}$ be a matrix as follows:
- $A = \begin {bmatrix} {\mathbf u_{k + 1} }^\intercal \\ \vdots \\ {\mathbf u_n}^\intercal \end {bmatrix}$
where $\mathbf u^\intercal$ denotes the transpose of $\mathbf u$ when expressed as a column matrix.
$V \subseteq \ker A$
By definition of orthonormality of $\set {\mathbf u_i}$ it follows that:
- $\forall i \in \N_{> 0} : i \le k : A \mathbf u_i = 0$.
Hence, any linear combination of $\mathbf u_1, \ldots, \mathbf u_k$ lies in the kernel of $A$.
That is, $V \subseteq \ker A$.
$\Box$
$\ker A \subseteq V$
Suppose $\mathbf x \in \ker A$.
Let $\ds \mathbf x = \sum_{i \mathop = 1}^n \alpha_i \mathbf u_i$, where $\set {\alpha_i}$ are scalars.
By definition of kernel:
- $A \mathbf x = 0$.
Then:
\(\ds \mathbf 0\) | \(=\) | \(\ds A \mathbf x\) | ||||||||||||
\(\ds \) | \(=\) | \(\ds \begin {bmatrix} \ds \mathbf u_{k + 1}^\intercal \sum_{i \mathop = 1}^n \alpha_i \mathbf u_i \\ \vdots \\ \ds \mathbf u_n^\intercal \sum_{i \mathop = 1}^n \alpha_i \mathbf u_i \end {bmatrix}\) | ||||||||||||
\(\ds \) | \(=\) | \(\ds \begin {bmatrix} \alpha_{k + 1} \\ \vdots \\ \alpha_n \end {bmatrix}\) | Definition of Orthonormal Basis of Vector Space |
Hence:
- $\forall i \in \N_{\ge k + 1} : i \le n : \alpha_i = 0$
So:
- $\ds \mathbf x = \sum_{i \mathop = 1}^k$
That is:
- $\mathbf x \in V$
Therefore by definition of subset:
- $\ker A \subseteq V$
$\Box$
By definition of set equality:
- $V = \ker A$
By Kernel of Linear Transformation between Finite-Dimensional Normed Vector Spaces is Closed, $V$ is closed.
$\blacksquare$
Sources
- 2017: Amol Sasane: A Friendly Approach to Functional Analysis ... (previous) ... (next): Chapter $\S 2.3$: The normed space $\map {CL} {X,Y}$