Vector Space and Geometry

Vector Space

Let \(\mathcal{F}\) be a scalar field (such as \(\mathbb{R}\) or \(\mathbb{C}\)).

Attention

  • Field refers to the algebraic definition with properly defined addition and multiplication operators on them.

  • Not to be confused with scalar fields which represents functionals that maps vectors into scalers.

[Definition] Vector Space

\(V_\mathcal{F}\) is a vector space over \(\mathcal{F}\) with \(0\in \mathcal{F}\) iff:

  • \(\forall a\in \mathcal{F},\mathbf{u}\in V_\mathcal{F}\implies a\cdot\mathbf{u}\in V_\mathcal{F}\)

  • \(\mathbf{u},\mathbf{v}\in V_\mathcal{F}\implies \mathbf{u}+\mathbf{v}\in V_\mathcal{F}\)

with the following properties:

  • [Commutative Addition]: \(\mathbf{u}+\mathbf{v}=\mathbf{v}+\mathbf{u}\)

  • [Identity Element]: \(\exists\mathbf{0}\in V_\mathcal{F}\) such that

    • \(0\cdot\mathbf{u}=\mathbf{0}\)

    • \(\mathbf{u}+\mathbf{0}=\mathbf{0}+\mathbf{u}=\mathbf{u}\)

  • [Inverse Element]: \(\forall\mathbf{u}\in V_\mathcal{F},\exists\mathbf{v}\in V_\mathcal{F}\) (represented as \(-\mathbf{u}\)) such that

    \(\mathbf{u}+\mathbf{v}=\mathbf{0}\)

Attention

Vector spaces are Abelian groups w.r.t \(+\) but the addition of scalar multiplication provides an even richer structure.

Tip

  • Elements of vector space are called vectors.

  • We often omit the underlying scalar field \(\mathcal{F}\) and write the vector space as \(V\).

  • Example of finite dimensional vectors: Euclidean vectors \(\mathbb{R}^n\) where the scalar field is \(\mathbb{R}\) or complex vectors \(\mathbb{C}^n\) over the scalar field \(\mathbb{C}\).

Euclidean Vector Space

These defintions and theorems are based on Graybill.

Vector Space

[Definition] n-component Vector

Let \(n\) be a positive integer and let \(a_1,\cdots,a_n\) be elements from \(\mathcal{F}\). The ordered \(n\)-tuple \(\mathbf{a}=(a_1,\cdots,a_n)^T\) is defined as n-component (or \(n\times 1\) vector)

See also

Vector spaces with \(n\times 1\) vectors are denoted here by \(V_n\).

Theorem

Let \(R_n\) be the set of all \(n\times 1\) vectors for a fixed positive integer \(n\). Then \(R_n\) is a vector space.

Vector Subspace

[Definition] Vector Subspace

Let \(S_n\) be the subset of vectors in the vector space \(V_n\). If the set \(S_n\) itself is a vector space, then \(S_n\) is called a vector subspace of \(V_n\).

Theorem

If \(S_n\) is a subset of the vector space \(V_n\) such that, for every two vectors, \(\mathbf{s}_1\) and \(\mathbf{s}_2\) in \(S_n\), \(a_1\mathbf{s}_1+a_2\mathbf{s}_2\) is in \(S_n\) for all real numbers \(a_1\) and \(a_2\), then \(S_n\) is a vector subspace of \(V_n\).

Theorem

The set \(\{\mathbf{0}\}\) where \(\mathbf{0}\) is the \(n\times 1\) null-vector, is a subspace of every vector space \(V_n\). Every vector space \(V_n\) is a subspace of itself.

Linear Dependence and Independence

[Definition] Linear Dependence and Independence

Let \(\{\mathbf{v}_1,\cdots,\mathbf{v}_m\}\) be a set of \(m\) vectors each with \(n\) components, so that \(\mathbf{v}_i\in R_n;i=1,\cdots,m\). This set is defined to be linearly dependent if and only if there exists a set of scalars \(\{c_1,\cdots,c_m\}\), at least one of which is not equal to zero, such that

\[\sum_{i=1}^m c_i\mathbf{v_i}=\mathbf{0}\]

If the only set of scalars that satisfies the above is \(\{0,\cdots,0\}\), then the set of vectors is called linearly independent.

Theorem

If the vector \(\mathbf{0}\) is included in a set of vectors, the set is linearly dependent.

Theorem

If \(m > 1\) vectors are linearly dependent, it’s always possible to express at least one of them as a linear combination of the others.

Theorem

In the set of \(m\) vectors \(\{\mathbf{v}_1,\cdots,\mathbf{v}_m\}\), if there are \(s\) vectors, \(s\le m\), that are linearly dependent, then the entire set of vectors is linearly dependent.

Theorem

If the set of \(m\) vectors \(\{\mathbf{v}_1,\cdots,\mathbf{v}_m\}\) is a linearly independent set, while the set of \(m+1\) vectors \(\{\mathbf{v}_1,\cdots,\mathbf{v}_{m+1}\}\) is a linearly dependent set, then \(\mathbf{v}_{m+1}\) can be expressed as a linear combination of \(\mathbf{v}_1,\cdots,\mathbf{v}_m\).

Theorem

A necessary and sufficient condition for thet set of \(n\times 1\) vectors \(\{\mathbf{v}_1,\cdots,\mathbf{v}_m\}\) to be linearly dependent set is that the rank of the matrix formed by the vectors (as columns) is less than the number of vectors \(m\); that is \(r < m\).

Theorem

If the rank of the matrix of a set of \(n\times 1\) vectors \(\{\mathbf{v}_1,\cdots,\mathbf{v}_m\}\) is \(r\), then \(r\) must be less than or equal to \(m\), and if \(r > 0\), theree exists exactly \(r\) of those vectors that are linearly independent, while each of the other \(m-r\) (if \(m-r > 0\)) vectors expressible as a linear combination of these \(r\) vectors.

Theorem

The set of \(n\times 1\) vectors \(\{\mathbf{v}_1,\cdots,\mathbf{v}_m\}\) is always linearly independent if \(m > n\).

Basis of a Vector Space

Theorem

Let \(\{\mathbf{v}_1,\cdots,\mathbf{v}_m\}\) be a set of vectors in \(V_n\) and let \(V\) be the set of vectors defined by

\[V=\left\{\mathbf{v}:\mathbf{v}=\sum_{i=1}^m c_i\mathbf{v}_i; c_i\in\mathbb{R} \right\};\]

then \(V\) is a subspace of \(V_n\).

[Definition] Generating (or Spanning) Vectors

Let \(V_n\) be a vector space. If each vector in \(V_n\) can be obtained from a linear combination of the vectors in the set \(\{\mathbf{v}_1,\cdots,\mathbf{v}_m\}\), then the set of vectors \(\{\mathbf{v}_1,\cdots,\mathbf{v}_m\}\) is said to generate (or span) \(V_n\).

[Definition] Basis

Let \(\{\mathbf{v}_1,\cdots,\mathbf{v}_m\}\) be a set of linearly independent set of vectors that spans \(V_n\). Then this set is called a basis of \(V_n\). For the special vector space \(\{\mathbf{0}\}\), we shall say that \(\mathbf{0}\) is the basis (even though it’s not linearly independent).

Theorem

If \(\{\mathbf{v}_1,\cdots,\mathbf{v}_m\}\), \(\{\mathbf{u}_1,\cdots,\mathbf{u}_q\}\) are two bases for \(V_n\), then \(m=q\).

[Definition] Dimensions

Let \(V_n\) be any vector space except \(\{\mathbf{0}\}\). Let the number of vectors in the basis of \(V_n\) be \(m\). Then \(m\) is defined to be the dimension of \(V_n\). The dimension of \(\{\mathbf{0}\}\) is defined to be 0.

Theorem

Let \(\{\mathbf{v}_1,\cdots,\mathbf{v}_m\}\) be a basis for the vector space \(V_n\) (neq \(\{\mathbf{0}\}\)). Let \(\mathbf{v}\) be any vector in \(V_n\). There is oen and only one ordered set of scalars \(\{c_1,\cdots,c_m\}\) such that

\[\mathbf{v}=\sum_{i=1}^m c_i\mathbf{v}_i\]

In other words, \(\mathbf{v}\) is a unique linear combination of a given basis.

Theorem

If \(r\) is the rank of the matrix of vectors \(\mathbf{v}_1,\cdots,\mathbf{v}_m\) that span the vector space \(V_n\), then there are exactly \(r\) independent vectors in that set and every vector in \(V_n\) can be expressed uniquely as a linear combination of these \(r\) vectors.

Theorem

If the vector space \(V_n\) is spanned by a set of \(m\) vectors, and if the matrix of those vectors has a rank \(r\), then any set of \(r+1\) vectors in \(V_n\) is linearly dependent.

Theorem

Let \(\mathbf{V}=\begin{bmatrix}\mathbf{v}_1&\cdots&\mathbf{v}_m\end{bmatrix}\) be a matrix containing a set of vectors that is a basis for \(V_n\), and let \(\mathbf{U}=\begin{bmatrix}\mathbf{u}_1&\cdots&\mathbf{u}_q\end{bmatrix}\) be a matrix that is any set of vectors in \(V_n\). The set of vectors in \(\mathbf{U}\) is a basis set for \(V_n\) if and only if \(m=q\) and there exists a non-singular \(m\times m\) matrix \(\mathbf{A}\) such that \(\mathbf{U}=\mathbf{V}\mathbf{A}\).

Theorem

Let \(\{\mathbf{v}_1,\cdots,\mathbf{v}_m\}\), \(m > 1\), be a basis for the vector space \(V_n\) and let \(\mathbf{v}\) be any vector in \(V_n\) such that \(\mathbf{v}=\sum_{i=1}^m c_i\mathbf{v}_i\). If \(c_t\neq 0\) for some \(t\), then the set \(\{\mathbf{v}_1,\cdots\mathbf{v}_{t-1},\mathbf{v},\mathbf{v}_{t+1},\cdots\mathbf{v}_m\}\) is a basis for \(V_n\). However, if \(c_t=0\), then the set \(\{\mathbf{v}_1,\cdots\mathbf{v}_{t-1},\mathbf{v},\mathbf{v}_{t+1},\cdots\mathbf{v}_m\}\) is a linearly dependent set and hence not a basis for \(V_n\).

Theorem

Let \(\{\mathbf{v}_1,\cdots,\mathbf{v}_q\}\) be a linearly independent vectors of \(V_n\). Then this set is a subset of a basis for \(V_n\).

Inner Product and Orthogonality of Vectors

[Definition] Inner Product

Let \(\mathbf{x}\) and \(\mathbf{y}\) be two vectors in \(V_n\). The inner product of \(\mathbf{x}\) and \(\mathbf{y}\), \(\mathbf{x}\cdot\mathbf{y}\) is defined to be the scalar \(\sum_{i=1}^nx_iy_i\). It is the scalar that is the element in the \(1\times 1\) matrix \(\mathbf{x}^T\mathbf{y}\).

[Definition] Orthogonal Vectors

Let \(\mathbf{x}\) and \(\mathbf{y}\) be two vectors in \(V_n\). \(\mathbf{x}\) and \(\mathbf{y}\) are defined to be orthogonal if and only if the inner product is 0.

Tip

\(\mathbf{0}\) vector is orthogonal to every vector in \(V_n\).

[Definition] Normal Vectors

A vector \(\mathbf{x}\) in \(V_n\) is defined to be a normal vector if and only if the inner product of \(\mathbf{x}\) with itself is equal to \(+1\).

[Definition] Orthogonal and Orthonormal Basis

If \(\{\mathbf{v}_1,\cdots,\mathbf{v}_m\}\) is a basis for \(V_n\) such that \(\mathbf{v}_i^T\mathbf{v}_j\) for all \(i\neq j=1,\cdots m\), then the basis is defined to be orthogonal basis for \(V_n\). If in addition, \(\mathbf{v}_i^T\mathbf{v}_i=1\) for all \(i=1,\cdots m\), the basis is defined to be orthonormal basis.

Theorem

Every vector space has an orthogonal basis.

Theorem

Every vector space except \(\{\mathbf{0}\}\) has an orthonormal basis.

Theorem

Let \(\{\mathbf{v}_1,\cdots,\mathbf{v}_m\}\) be a set of vectors in \(V_n\) such that each and every distinct pair of vectors is orthogonal; that is \(\mathbf{v}_i^T\mathbf{v}_j=0\) for all \(i\neq j\). If none of the vectors is the zero vector, then the set of vectors is a linearly independent set.

Theorem

Any set of \(q\) nonzero pairwise orthogonal vectors in \(V_n\) is a subset of a basis for \(V_n\).

Theorem

Let \(\{\mathbf{v}_1,\cdots,\mathbf{v}_q\}\) be the set of basis for the vector space \(V_n (\neq \{\mathbf{0}\})\). Then the set of \(q\) vectors \(\{\mathbf{z}_1,\cdots,\mathbf{z}_q\}\) is also a basis vector for \(V_n\) and they are an orthonormal set where they are defined as

\[\begin{split}\begin{matrix}\mathbf{y}_1=\mathbf{v}_1;&\mathbf{z}_1=\frac{\mathbf{y}_1}{\sqrt{\mathbf{y}_1^T\mathbf{y}_1}} \\ \mathbf{y}_2=\mathbf{v}_2-\frac{\mathbf{y}_1^T\mathbf{v}_2}{\mathbf{y}_1^T\mathbf{y}_1}\mathbf{y}_1;&\mathbf{z}_2=\frac{\mathbf{y}_2}{\sqrt{\mathbf{y}_2^T\mathbf{y}_2}} \\ \vdots&\vdots\\ \mathbf{y}_q=\mathbf{v}_q-\frac{\mathbf{y}_1^T\mathbf{v}_q}{\mathbf{y}_1^T\mathbf{y}_1}\mathbf{y}_1-\frac{\mathbf{y}_2^T\mathbf{v}_q}{\mathbf{y}_2^T\mathbf{y}_2}\mathbf{y}_2-\cdots-\frac{\mathbf{y}_{q-1}^T\mathbf{v}_q}{\mathbf{y}_{q-1}^T\mathbf{y}_{q-1}}\mathbf{y}_{q-1};&\mathbf{z}_q=\frac{\mathbf{y}_q}{\sqrt{\mathbf{y}_q^T\mathbf{y}_q}} \end{matrix}\end{split}\]

Affine Sets in Euclidean Vector Space

Line

Plane

Orthogonal Projections