What the Axioms Require
In familiar coordinate spaces, addition and scalar multiplication follow predictable rules: addition can be regrouped, multiplying a vector by \(1\) leaves it unchanged, and scalar multiplication distributes over vector addition. The vector space axioms isolate the rules that make these operations work together. They also let us decide whether a set is a vector space even when its elements are not coordinate columns or when its operations are defined in an unusual way.
Throughout this tutorial, scalars are real numbers. Let \(V\) be a set with an addition operation \(+\) and a scalar multiplication operation. For the operations to be suitable, addition must assign an element of \(V\) to each pair of elements of \(V\), and scalar multiplication must assign an element of \(V\) to each real number and each element of \(V\). These requirements are called closure: \(u+v\in V\) and \(a v\in V\) for every \(u,v\in V\) and \(a\in\mathbb{R}\). Closure is required in addition to the eight axioms below.
- Commutativity of addition: \(u+v=v+u\).
- Associativity of addition: \((u+v)+w=u+(v+w)\).
- Additive identity: There is an element \(0_V\in V\) such that \(v+0_V=v\) for every \(v\in V\).
- Additive inverses: For every \(v\in V\), there is an element \(-v\in V\) such that \(v+(-v)=0_V\).
- Compatibility of scalar multiplication: \(a(bv)=(ab)v\).
- Scalar identity: \(1v=v\).
- Distributivity over vector addition: \(a(u+v)=au+av\).
- Distributivity over scalar addition: \((a+b)v=av+bv\).
The symbols \(0_V\), \(-v\), and products such as \(av\) refer to the specified operations on \(V\); they need not be ordinary numerical zero, subtraction, or multiplication. In particular, when testing a set with modified operations, do not assume its additive identity is the number or object conventionally called zero. The identity must be determined using the proposed addition.
The quantifiers matter. Each axiom must hold for every choice of the listed vectors and scalars. One calculation can disprove an axiom by providing a counterexample, but verifying an axiom generally requires an argument that covers arbitrary choices. Closure should also be checked: an expression outside \(V\) means the proposed operation does not even define an operation on \(V\).
The Usual Operations on Coordinate Vectors
The standard coordinate space \(\mathbb{R}^n\) consists of all real column vectors with \(n\) entries. Addition is entrywise, and scalar multiplication multiplies each entry by the scalar. The axioms hold because each coordinate is a real number and real-number addition and multiplication obey the corresponding rules.
Proof. Write \(u=(u_1,\dots,u_n)^{\top}\), \(v=(v_1,\dots,v_n)^{\top}\), and \(w=(w_1,\dots,w_n)^{\top}\). Entrywise addition and scalar multiplication produce \(n\)-tuples of real numbers, so both operations are closed. In each coordinate \(i\), commutativity follows from \(u_i+v_i=v_i+u_i\), and associativity follows from \((u_i+v_i)+w_i=u_i+(v_i+w_i)\). The vector \(\mathbf{0}=(0,\dots,0)^{\top}\) satisfies \(v+\mathbf{0}=v\), and the vector \(-v=(-v_1,\dots,-v_n)^{\top}\) satisfies \(v+(-v)=\mathbf{0}\). For scalars \(a,b\), the \(i\)th coordinates of the two sides of the remaining four axioms agree by the real-number identities \(a(bv_i)=(ab)v_i\), \(1v_i=v_i\), \(a(u_i+v_i)=au_i+av_i\), and \((a+b)v_i=av_i+bv_i\). Thus all eight axioms and closure hold. \(\square\)
Worked Example: Checking the Usual Operations in \(\mathbb{R}^2\)
Take \(u=(1,-2)^{\top}\), \(v=(3,4)^{\top}\), and scalars \(a=2\), \(b=-1\). The theorem applies to every pair of vectors and every scalar, while these computations illustrate how the rules look for particular inputs.
Addition gives \(u+v=(4,2)^{\top}\) and \(v+u=(4,2)^{\top}\). The additive identity is \(\mathbf{0}=(0,0)^{\top}\), since \(u+\mathbf{0}=(1,-2)^{\top}\); the inverse of \(u\) is \(-u=(-1,2)^{\top}\), since \(u+(-u)=(0,0)^{\top}\).
For scalar compatibility, \(bv=(-3,-4)^{\top}\), so \(a(bv)=(-6,-8)^{\top}\); also \(ab=-2\), and \((ab)v=(-2)(3,4)^{\top}=(-6,-8)^{\top}\). For the two distributive laws,
and
Also \(1u=u\). The computations check the stated instances; the coordinate proof, rather than these particular choices, establishes the axioms for all vectors and scalars.
Operations Can Be Different from the Usual Ones
A set does not have vector space status simply because its elements are real numbers. The operations are part of the structure being tested. An altered addition may have a different identity and different inverses; the correct question is whether that addition and its accompanying scalar multiplication satisfy the definition.
Worked Example: A Shifted Pair of Operations on \(\mathbb{R}\)
Let \(V=\mathbb{R}\), but define operations by \(x\oplus y=x+y-1\) and \(a\odot x=ax+1-a\). Determine whether \(V\) is a real vector space under these operations.
Both operations are closed: if \(x,y,a\) are real, then \(x+y-1\) and \(ax+1-a\) are real. For arbitrary real \(x,y,z\),
and \(x\oplus y=x+y-1=y+x-1=y\oplus x\). The additive identity is \(1\), because \(x\oplus1=x+1-1=x\). The additive inverse of \(x\) is \(2-x\), because \(x\oplus(2-x)=x+2-x-1=1\).
It remains to check the four scalar axioms. For real \(a,b,x\),
The scalar identity is \(1\odot x=1x+1-1=x\). For distributivity over vector addition,
while
For distributivity over scalar addition,
and
Thus closure and all eight axioms hold. The set \(\mathbb{R}\) with these specified operations is a real vector space.
This example emphasizes why the operations must be written down before testing the axioms. Its additive identity is \(1\), not the usual real number \(0\), and the additive inverse of \(x\) under \(\oplus\) is \(2-x\), not the usual \(-x\). These values are determined by the equations required in the definition.
A Closure Failure Rules Out a Vector Space
Worked Example: Positive Reals with the Usual Operations
Consider \(V=(0,\infty)\), the set of positive real numbers, with the usual real addition and scalar multiplication. This is not a real vector space. For example, \(2\in V\), but multiplying by the real scalar \(-1\) gives \((-1)2=-2\notin V\). Therefore scalar multiplication is not closed on \(V\), so one of the required conditions already fails.
There is another closure failure: the scalar \(0\) and the element \(2\in V\) give \(0\cdot2=0\notin V\). Also, the usual additive identity \(0\) is not in \(V\). Any one of these facts is enough to reject the proposed structure; no further axiom checks can make it a vector space with these operations.
Common Pitfall: Assuming the Usual Zero
A common error is to test a modified addition using the ordinary zero instead of finding the identity from the defining equation. For the shifted operation \(x\oplus y=x+y-1\), the claim that \(0\) is the additive identity is wrong: \(x\oplus0=x-1\), which is not \(x\) for a general real \(x\). In particular, at \(x=3\), this gives \(3\oplus0=2\ne3\). The correction is to solve \(x\oplus e=x\): \(x+e-1=x\) implies \(e=1\), and indeed \(x\oplus1=x\) for every \(x\).
Another useful discipline is to separate the two stages of a decision. First check that each operation actually returns an element of the proposed set. Then check each axiom for arbitrary allowed inputs. A single counterexample settles a failed condition; a successful test on a few chosen inputs does not establish an axiom that must hold universally.
The axioms matter because they specify exactly which algebraic rules are available when manipulating vectors. Once a set and its operations meet the definition, vector-space arguments can use those rules without depending on whether the elements are coordinate vectors, numbers with modified operations, or other objects. For a proposed structure, record the operations, verify closure, identify the additive identity and inverses using those operations, and then test the remaining equalities with their full quantifiers.
Check Your Understanding
For each question, refer to the specified operations rather than assuming the usual ones.
- State the eight vector space axioms and the two closure requirements.
- Under \(x\oplus y=x+y-1\), find the additive identity and the additive inverse of \(x\), and verify both defining equations.
- For the usual operations, why does \((0,\infty)\) fail to be a real vector space? Give one specific scalar-multiplication counterexample.
- In the shifted-operations example, calculate \(2\odot(3\oplus4)\) and compare it with \((2\odot3)\oplus(2\odot4)\).
- Why do a few successful numerical checks not prove an axiom that is required for every vector and scalar?
