A Subspace Is a Subset with Inherited Operations
A vector space is not just a collection of objects. It is a collection together with addition and scalar multiplication that satisfy the vector space axioms. This distinction matters when we ask whether a subset of a vector space is itself a vector space: the question is not whether we can invent some operations on the subset, but whether the operations it inherits from the larger space make it a vector space.
Throughout this tutorial, spaces and scalars are over \(\mathbb{R}\). Let \(V\) be a real vector space, and let \(W\subseteq V\). The addition and scalar multiplication on \(V\) can be restricted to elements of \(W\), provided the results remain in \(W\). In that case, the operations on \(W\) are exactly the ambient operations applied to elements of \(W\). They are not newly chosen operations.
There are three related but distinct things here. First, \(W\) is a subset: it specifies which elements are included. Second, \(W\) uses the addition and scalar multiplication of \(V\), restricted to those elements. Third, \(V\) is the ambient vector space that supplies the original operations and axioms. A subset does not become a subspace merely because its elements happen to be vectors in \(V\); its inherited operations must make it a vector space.
The word “inherited” is essential. For example, if \(V=\mathbb{R}^2\) has coordinatewise addition and the usual scalar multiplication, then a subset \(W\) is checked using those exact operations. Giving the same collection of ordered pairs some different addition or scalar multiplication would be a different question about a different vector space structure.
What the Subspace Inherits
The vector space axioms for a subspace do not need to be established from scratch. Once the inherited operations stay within the subset, the algebraic identities in \(V\) continue to hold for elements of \(W\). The zero vector and additive inverses, however, deserve particular care: they are elements of \(W\), and their agreement with the ambient zero and inverses must be justified.
Proof. Choose \(w\in W\), which is possible because \(W\ne\varnothing\). Closure under scalar multiplication gives \(0w=\mathbf{0}_V\in W\), so the ambient zero is in \(W\). For any \(u,v\in W\), closure under addition ensures that \(u+v\in W\); closure under scalar multiplication ensures that \(cv\in W\) for every \(c\in\mathbb{R}\). Thus the restricted operations are defined within \(W\). Associativity and commutativity of addition, both distributive laws, and the scalar multiplication identities hold for elements of \(W\) because they hold for those same elements in \(V\). For each \(w\in W\), the element \((-1)w\) lies in \(W\), and \(w+(-1)w=\mathbf{0}_V\), so every element has an additive inverse in \(W\). The scalar identity \(1w=w\) also holds because it holds in \(V\). These facts verify the vector space axioms on \(W\).
It remains to identify the zero and inverses precisely. Let \(\mathbf{0}_W\) denote the additive identity in \(W\), whose existence follows from the vector space axioms just verified. Since \(\mathbf{0}_W\in W\), inherited addition gives \(\mathbf{0}_W+\mathbf{0}_W=\mathbf{0}_W\). The ambient identity gives \(\mathbf{0}_W+\mathbf{0}_V=\mathbf{0}_W\). Both equations are in \(V\); cancellation in \(V\) therefore implies \(\mathbf{0}_W=\mathbf{0}_V\). If \(z\in W\) is an additive inverse of \(w\) in \(W\), then \(w+z=\mathbf{0}_W=\mathbf{0}_V\). The additive inverse of \(w\) in \(V\) is unique, so \(z\) is that same ambient inverse. \(\square\)
This theorem gives a practical way to read the definition: closure keeps the inherited operations inside the subset, and the axioms then come from the ambient space. It also explains why the phrase “zero vector of \(W\)” does not refer to a new object. Whenever \(W\) is a subspace, its zero is the ambient zero.
Worked Examples in Different Ambient Spaces
Worked Example: A Line Through the Origin in \(\mathbb{R}^3\)
Consider \(W=\{(t,2t,0):t\in\mathbb{R}\}\subseteq\mathbb{R}^3\), with the usual operations inherited from \(\mathbb{R}^3\). Check that \(W\) is a subspace and calculate with two of its elements.
Take arbitrary \(u=(a,2a,0)\in W\), \(v=(b,2b,0)\in W\), and \(c\in\mathbb{R}\). Addition and scalar multiplication in the ambient space give
and
The set is nonempty because \((0,0,0)\in W\), obtained by setting \(t=0\). Thus the inherited operations stay in \(W\), and the theorem shows that \(W\) is a subspace. For a numerical check, choose \(u=(2,4,0)\) and \(v=(-1,-2,0)\). Then
Check: \((1,2,0)\) has the required form with parameter \(t=1\), and \((6,12,0)\) has that form with \(t=6\). The calculations use the usual operations of \(\mathbb{R}^3\), not separate operations chosen for \(W\).
Worked Example: Linear Polynomials Inside Quadratic Polynomials
Let \(P_2\) be the real vector space of polynomials of degree at most \(2\), including the zero polynomial, and let \(W=P_1=\{a+bx:a,b\in\mathbb{R}\}\subseteq P_2\). Determine how the ambient space affects the statement that \(W\) is a subspace.
For arbitrary \(p=a+bx\) and \(q=c+dx\) in \(W\), their sum, computed using polynomial addition in \(P_2\), is
For any real scalar \(r\), the inherited scalar multiplication gives
The zero polynomial is in \(W\), and the negative \(-p=-a-bx\) is also in \(W\). Therefore \(W\) is a subspace of \(P_2\). For a concrete calculation, take \(p=2-3x\) and \(q=1+4x\). Then
Check: \(3+x\) has coefficients \(3\) and \(1\), while \(-4+6x\) has coefficients \(-4\) and \(6\); both are in \(P_1\), hence in \(W\). The same collection \(P_1\) can also be considered as the ambient space itself. In that setting it is not a proper subset of the ambient space, but it is still a subspace: equality \(W=V\) is allowed by the definition.
This example shows why “subspace” is a relation between a subset and an ambient space, rather than a label attached to a set in isolation. The set \(P_1\) sits inside \(P_2\) with the operations inherited from \(P_2\); when considered on its own, \(P_1\) is a vector space in its own right. In both settings the polynomial addition and scalar multiplication being used are the usual ones.
Worked Example: A Subset That Does Not Inherit a Vector Space Structure
Let \(S=\{(t,1):t\in\mathbb{R}\}\subseteq\mathbb{R}^2\), with the usual coordinate operations inherited from \(\mathbb{R}^2\). Decide whether \(S\) is a subspace.
Take \((0,1)\in S\). Adding it to itself using the ambient operation gives
The result is not in \(S\), because every element of \(S\) has second coordinate \(1\). Scalar multiplication also leaves the set: \(0\cdot(0,1)=(0,0)\notin S\). Either calculation shows that an inherited operation does not always return an element of \(S\), so the restricted operations do not make \(S\) a vector space.
Check: the ambient zero \((0,0)\) is not an element of \(S\), as its second coordinate is \(0\), not \(1\). This is consistent with the failure of closure under scalar multiplication by \(0\).
Nested Ambient Spaces
Ambient spaces can be nested. If \(U\) is a subspace of \(W\) and \(W\) is a subspace of \(V\), then the operations used on \(U\) are still the operations from \(V\): each restriction is simply applied to a smaller set. This is not a change of operations when moving through the inclusions.
Proof. Since \(U\) is a subspace of \(W\), it is nonempty, is closed under the addition on \(W\), and is closed under scalar multiplication on \(W\). Since \(W\) is a subspace of \(V\), the addition and scalar multiplication on \(W\) are restrictions of those on \(V\). Thus for \(u_1,u_2\in U\), their sum computed in \(V\) is the same element as their sum computed in \(W\), and it lies in \(U\). For \(c\in\mathbb{R}\) and \(u\in U\), the product \(cu\) computed in \(V\) is the same as the product computed in \(W\), and it lies in \(U\). Therefore \(U\) is nonempty and closed under the operations from \(V\). The inherited vector space structure theorem shows that \(U\) is a subspace of \(V\). \(\square\)
For instance, the constant polynomials \(U=\{a:a\in\mathbb{R}\}\) sit inside \(P_1\), and \(P_1\) sits inside \(P_2\). Addition and scalar multiplication at both levels are ordinary polynomial operations. The theorem says that the constants are consequently a subspace of \(P_2\) as well as a subspace of \(P_1\).
Common Mistake: Treating Membership as Enough
A common incorrect argument is: “Every element of \(S=\{(t,1):t\in\mathbb{R}\}\) is a vector in \(\mathbb{R}^2\), so \(S\) is a subspace.” The conclusion does not follow from membership in the ambient space. A subspace must itself be a vector space under the inherited operations. In this example, the concrete calculation \((0,1)+(0,1)=(0,2)\notin S\) disproves closure under the inherited addition. Also, \(0(0,1)=(0,0)\notin S\), so the inherited scalar multiplication fails to stay within the set.
Another error is to repair a failure by quietly changing the operations on the subset. If the question is whether \(S\subseteq V\) is a subspace of \(V\), the operations are fixed: they are the restrictions of the operations in \(V\). A different choice of operations might define a vector space on a collection of objects, but it would not establish that the collection is a subspace of the given ambient space.
Keep the three pieces separate when reading or giving a subspace claim: identify the ambient vector space \(V\), identify the subset \(W\), and state that the operations on \(W\) are inherited from \(V\). Once those pieces are clear, closure ensures that the operations stay within \(W\), while the ambient axioms supply the required algebraic identities.
Check Your Understanding
In each question, pay attention to which operations are being used and which space is ambient.
- In the theorem on inherited structure, why must \(W\) be nonempty before using closure to obtain \(\mathbf{0}_V\in W\)?
- For \(W=\{(t,-t):t\in\mathbb{R}\}\subseteq\mathbb{R}^2\), calculate the sum of \((2,-2)\) and \((-3,3)\), and calculate \(4(2,-2)\). Do the results lie in \(W\)?
- Explain why a subset can be a subspace of itself even though it is not a proper subset of its ambient space.
- For \(S=\{(t,1):t\in\mathbb{R}\}\), give one inherited operation that produces an element outside \(S\), and identify the ambient space used.
- If \(U\) is a subspace of \(W\) and \(W\) is a subspace of \(V\), explain why the operations used to test \(U\) as a subspace of \(V\) are not new operations.
