Fields

A field is a triple \((K, +, \boldsymbol{\cdot})\), where \(K\) is a set and \(+, \boldsymbol{\cdot}\) are maps (or in detail, binary operations) \(K \times K \longrightarrow K\), satisfying the following axioms [1], [2]:

  • For the addition operation \(+\):

    • (A1) (Commutativity of \(+\)) \(\forall a,b \in K: \; a+b=b+a\)
    • (A2) (Associativity of \(+\)) \(\forall a,b,c \in K: \; (a+b)+c=a+(b+c)\)
    • (A3) (Neutral element of \(+\)) \(\exists 0 \in K: \; \forall a \in K: a+ 0 = a\)
    • (A4) (Inverse element of \(+\)) \(\forall a \in K: \; (\exists -a) \in K: a + (-a) = 0\)


  • For the multiplication operation \(\boldsymbol{\cdot}\):

    • (A5) (Commutativity of \(\boldsymbol{\cdot}\)) \(\forall a,b \in K: \; a\boldsymbol{\cdot} b=b\boldsymbol{\cdot} a\)
    • (A6) (Associativity of \(\boldsymbol{\cdot}\)) \(\forall a,b,c \in K: \; (a\boldsymbol{\cdot} b)\boldsymbol{\cdot} c=a \boldsymbol{\cdot} (b \boldsymbol{\cdot} c)\)
    • (A7) (Neutral Element of \(\boldsymbol{\cdot}\)) \(\exists 1 \in K: \; \forall a \in K: a\boldsymbol{\cdot} 1 = a\)
    • (A8) (Inverse element of \(\boldsymbol{\cdot}\)) \(\forall a \in K\setminus \{0\}:\; \exists a^{-1} \in K: a\boldsymbol{\cdot} a^{-1}=1\)


  • Distributivity of multiplication over addition

    • (A9) \(\forall a,b,c \in K: \; (a+b)\boldsymbol{\cdot} c=a\boldsymbol{\cdot} c + b\boldsymbol{\cdot} c\)


Some remarks regarding to fields

  • A mnemonic for these axioms is CANI [2] — the collection of the first letters — for each operation. The last axiom (A9) connects the addition and multiplication operation.
  • Examples of fields are rational numbers \(\mathbb{Q}\), real numbers \(\mathbb{R}\) and complex numbers \(\mathbb{C}\) with the commonly-used addition and multiplication operations. However, integer \(\mathbb{Z}\) is not a field since \(3\boldsymbol{\cdot} 1/3=1\) but \(1/3\notin \mathbb{Z}\), hence Axiom A8 is not satisfied. The only integers which have inverse elements are \(1\) and \(-1\).
  • In other words, \((K, +)\) is an Abelian group, which corresponds to (A1)—(A4), and \((K\setminus\{0\}, \boldsymbol{\cdot})\) is an Abelian group, which corresponds to (A5)—(A8). Recall that Abelian group (also called commutative group) is a group where the group operation is commutative.
  • The neutral element of addition (multiplication) is unique, while the inverse element of addition (multiplication) is unique for each element of the field.
  • A weaker notion than a field is a ring, which is also a triple \((R, +, \boldsymbol{\cdot})\) that does not satisfy (A5), (A7) and (A8). If (A5) is additionally satisfied, the triple is called a commutative ring. If (A7) is additionally satisfied, the triple is called an unitary ring. An example for a ring is \((\mathbf{M}_{n,n},+,\boldsymbol{\cdot})\), where \(\mathbf{M}_{n,n}\) is a \(n\times n\) square matrix and the operations are the commonly-used matrix operations. It is clear that inverse element does not always exists and matrix operation under \(\cdot\) is not commutative.



Vector Spaces

Let \((K,+,\boldsymbol{\cdot})\) be a field. a \(K\)-vector space — a vector space over field \(K\) — is a triple \((V, \oplus, \odot)\), where V is a set and \(\oplus, \odot\) are binary operations called addition and s-multiplication, respectively [1]–[3]:

\[ \begin{align*} \oplus &: V \times V \longrightarrow V \\ \odot &: K \times V \longrightarrow V \end{align*} \]

satisfying the following axioms:

  • (A1) (Commutativity of \(\oplus\)) \(\forall a,b \in V: \; a\oplus b=b \oplus a\)
  • (A2) (Associativity of \(\oplus\)) \(\forall a,b,c \in V: \; (a\oplus b)\oplus c=a\oplus (b\oplus c)\)
  • (A3) (Neutral element of \(\oplus\)) \(\exists 0 \in V: \; \forall a \in V: a \oplus 0 = a\)
  • (A4) (Inverse element of \(\oplus\)) \(\forall a \in V: \; \exists (-a) \in V: a \oplus (-a) = 0\)
  • (A5) (Associativity of \(\odot\)) \(\forall \mu,\nu \in K, \; \forall a \in V: (\mu\cdot \nu)\odot a = \mu \odot ( \nu \odot a)\)
  • (A6) (Distributivity w.r.t. \(\oplus\)) \(\forall \mu \in K, \; \forall a, b \in V: \mu \odot ( a \oplus b) = (\mu \odot a) \oplus (\mu \odot b)\)
  • (A7) (Distributivity w.r.t. \(+\)) \(\forall \mu, \nu \in K, \; \forall a \in V: (\mu + \nu) \odot a = (\mu \odot a) \oplus ( \nu \odot a)\)
  • (A8) (Unit Element) \(\forall a\in V: 1 \odot a = a\)

Again, some remarks regarding to K-vector spaces

  • One should be careful of not mixing the notations. In particular, \(\boldsymbol{\cdot}\), \(\odot\) and \(+\), \(\oplus\) are totally different operations. Moreover, it is also obvious that the \(0\) notation in vector space is different with the notation in field, so we must know where \(0\) is actually from. For instance, if \(\;0\oplus a = a\), then \(\;0\in V\;\) and \(\;0 \notin K\).
  • A mnemonic for these axioms is . Thankfully, this makes sense in english — can I add (yo)u? [2].
  • Vector spaces are also called linear spaces, and their elements are called vectors. This definition is different with the vectors often defined in physics where vectors are just arrows from some starting point.
  • If field \(K\) is a real (complex) number \(\mathbb{R}\) (\(\mathbb{C}\)), we call \(V\) a real (complex) vector space. Hence, even though the vector elements are complex, if the field is a real number \(\mathbb{R}\), we call it a real vector space (as mentioned in [3]).
  • The definition of field is important when we define span and linear independence of vectors. Details are in this post.



Examples of Vector Spaces


Polynomials with Fixed-order

We define a set \(P_n(\mathbb{R})\), where \(P_n(\mathbb{R})\) is a set of \(n\)-th order polynomials with real coefficients: \[ P_n(\mathbb{R}) = \bigg\{ p: \mathbb{R} \rightarrow \mathbb{R} \; | \; p(x)=\sum_{n\in N}a_nx^n, \;\; a_n \in \mathbb{R}, \;\; N = \{ 0, 1, 2, \cdots n \} \bigg\} \] and define addition \(\oplus: P_n(\mathbb{R}) \times P_n(\mathbb{R}) \longrightarrow P_n(\mathbb{R})\) via point-wise operation:

\[ \forall x \in \mathbb{R}, \forall p_1, p_2 \in P_n(\mathbb{R}): \;\; (p_1 \oplus p_2) (x) = p_1(x) + p_2(x) \] and s-multiplication \(\odot:\mathbb{R} \times P_n(\mathbb{R}) \longrightarrow P_n(\mathbb{R})\) over field \(\mathbb{R}\). \[ \forall x, \lambda \in R, \forall p \in P_n(\mathbb{R}): (\lambda \odot p)(x) = \lambda p(x) \] Then it is straightforward to show that \((P_n(\mathbb{R}), \oplus, \odot)\) is a vector space.



Coordinate space

For any positive integer \(n\), the set of all \(n\)-tuples of elements of field \((K,+,\cdot)\) forms a vector space over \(K\). We often call this vector space as coordinate space and denote it as \(K^{n}\). The formal definition is:

\[ K^{n} = \big\{ (k_1, k_2, \cdots, k_n) \; | \; k_1, k_2, \cdots, k_n \in K \big\} \] and define addition \(\oplus: K^n \times K^n \longrightarrow K^n\) via point-wise operation:

\[ \forall a, b \in K^n: \;\; a \oplus b = (a_1, a_2, \cdots a_n) \oplus (b_1, b_2,\dots, b_n) = (a_1+b_1, a_2+b_2, \cdots, a_n+ b_n) \] and s-multiplication \(\odot:\mathbb{R} \times P_n(\mathbb{R}) \longrightarrow P_n(\mathbb{R})\) over field \(\mathbb{R}\). \[ \lambda \in K, \forall a \in K^n: \lambda \odot a = \lambda \odot (a_1, a_2, \cdots, a_n) = (\lambda \cdot a_1, \lambda \cdot a_2, \cdots, \lambda \cdot a_n ) \]

Again, it is straightforward (but labor-intensive) to show that \(K^n, \oplus, \odot)\) is a vector space. Just for an example, we show when field \(K\) is \((\mathbb{R}, +, \cdot)\) and \(n=3\).



Example — \(\mathbb{R}^3\)

This example is directly from this video. Warning! The proof is labor-intensive.

Let \(V=\mathbb{R}^3\) be a set of all real triples. We equip the set \(V\) with addition \(\oplus: V \times V \longrightarrow V\) and s-multiplication \(\odot: \mathbb{R} \times V \longrightarrow V\), defined by: \[ (a,b,c)\oplus (d,e,f) \triangleq (a+d, b+e, c+f) \] and \(\forall \lambda \in \mathbb{R}\): \[ \lambda \odot (a,b,c) \triangleq (\lambda a, \lambda b, \lambda c) \] Our goal is to check \((V,\oplus, \odot)\) is a vector space.



(A1) Axiom 1

\[ (a,b,c) \oplus (d,e,f) = (a+d, b+e, c+f)=(d+a, e+b, f+c) =(d,e,f) \oplus (a,b,c) \]

(A2) Axiom 2

\[ \begin{align*} \big((a,b,c) \oplus (d,e,f)\big) \oplus (g,h,i) &= (a+d, b+e, c+f) \oplus (g,h,i) = (a+d+g, b+e+h, c+f+i) \\ &= (a,b,c) \oplus (d+g, e+h, f+i) = (a,b,c) \oplus \big((d,e,f) \oplus (g,h,i)\big) \end{align*} \]

(A3) Axiom 3

\[ \exists n =(0,0,0) \in V: (0,0,0)+(a,b,c) = (a,b,c) \]

(A4) Axiom 4

\[ \forall v = (a,b,c) \in V: \exists (-v)\triangleq(-a,-b,-c): \;\; v\oplus (-v) = n \]

(A5) Axiom 5

\[ \forall \lambda, \mu \in \mathbb{R}: \;\; (\lambda \cdot \mu) \odot (a,b,c) = (\lambda\mu a, \lambda\mu b, \lambda\mu c) = \lambda \odot (\mu a, \mu b, \mu c) = \lambda \odot ( \mu \odot (a,b,c)) \]

(A6) Axiom 6

\[ \begin{align*} \forall \mu \in \mathbb{R}: \;\; \mu \odot \big((a,b,c)\oplus (d,e,f)\big) &= \mu \odot (a+d, b+e, c+f) = (\mu a + \mu d, \mu b + \mu e, \mu c + \mu f) \\ &= (\mu a, \mu b, \mu c) \oplus (\mu d, \mu e, \mu f) = (\mu \odot (a,b,c)) \oplus (\mu \odot (d,e,f)) \end{align*} \]

(A7) Axiom 7

\[ \begin{align*} \forall \mu, \lambda \in \mathbb{R}: \;\; (\mu + \lambda) \odot (a,b,c) &= \big( (\mu+\lambda)a, (\mu+\lambda)b, (\mu+\lambda)c \big) = (\mu a + \lambda a, \mu b + \lambda b, \mu c + \lambda c)\\ &= (\mu a, \mu b, \mu c ) \oplus (\lambda a, \lambda b, \lambda c) = ( \mu \odot (a,b,c) ) \oplus ( \lambda \odot (a,b,c) ) \end{align*} \]

(A8) Axiom 8

\[ \begin{align*} 1 \odot (a,b,c) = (a,b,c) \end{align*} \]

Since \((V,\oplus, \odot)\) satisfies all 8 axioms, \((V,\oplus, \odot)\) is a vector space.



References

[1]
P. R. Halmos, Finite dimensional vector spaces. 1958, pp. 1–4.
[2]
F. P. Schuller, “Tensor space theory i: Over a field - lec 08 - frederic p schuller,” 2015. https://www.youtube.com/watch?v=4l-qzZOZt50
[3]
M. Vidyasagar, Nonlinear systems analysis. SIAM, 2002, p. 7.