Why Put Two Vector Spaces Together?
Sometimes a mathematical object naturally carries two kinds of information at once. A pair might contain a vector and a matrix, or a function and a polynomial. If each component can be added and scaled within its own space, we can define addition and scalar multiplication on the pairs by doing the operation in each component. The resulting construction is called a product space.
Let \(V\) and \(W\) be real vector spaces. Their elements may be different kinds of objects: for example, \(V\) could be a space of vectors while \(W\) is a space of matrices. The Cartesian product \(V\times W\) is the set of all ordered pairs \((\mathbf{v},\mathbf{w})\) with \(\mathbf{v}\in V\) and \(\mathbf{w}\in W\). The order matters: \((\mathbf{v},\mathbf{w})\) and \((\mathbf{w},\mathbf{v})\) belong to different products in general, and the two coordinates need not even be of the same type.
These definitions make sense because \(V\) and \(W\) are vector spaces: \(\mathbf{v}+\mathbf{u}\) belongs to \(V\), \(\mathbf{w}+\mathbf{z}\) belongs to \(W\), and scalar multiples stay in their respective spaces. Thus the resulting pair remains in \(V\times W\). To establish that the product is itself a vector space, we must also verify all the vector space axioms, not merely that the operations produce pairs.
The Product-Space Theorem
Proof. Let \((\mathbf{v},\mathbf{w}),(\mathbf{u},\mathbf{z}),(\mathbf{x},\mathbf{y})\in V\times W\) and let \(a,b\in\mathbb{R}\). Closure of addition and scalar multiplication follows from closure in \(V\) and \(W\). For addition, commutativity in both component spaces gives \((\mathbf{v},\mathbf{w})+(\mathbf{u},\mathbf{z})=(\mathbf{u},\mathbf{z})+(\mathbf{v},\mathbf{w})\). Associativity follows by calculating \[ ((\mathbf{v},\mathbf{w})+(\mathbf{u},\mathbf{z}))+(\mathbf{x},\mathbf{y}) =((\mathbf{v}+\mathbf{u})+\mathbf{x},(\mathbf{w}+\mathbf{z})+\mathbf{y}) =(\mathbf{v}+(\mathbf{u}+\mathbf{x}),\mathbf{w}+(\mathbf{z}+\mathbf{y})), \] which equals \((\mathbf{v},\mathbf{w})+((\mathbf{u},\mathbf{z})+(\mathbf{x},\mathbf{y}))\). The pair \((\mathbf{0}_V,\mathbf{0}_W)\) is an additive identity because adding it leaves each coordinate unchanged. The pair \((-\mathbf{v},-\mathbf{w})\) is an additive inverse because their componentwise sum is \((\mathbf{0}_V,\mathbf{0}_W)\).
For scalar multiplication, distributivity over vector addition follows from the corresponding axiom in each component: \(a((\mathbf{v},\mathbf{w})+(\mathbf{u},\mathbf{z}))=(a\mathbf{v}+a\mathbf{u},a\mathbf{w}+a\mathbf{z})=a(\mathbf{v},\mathbf{w})+a(\mathbf{u},\mathbf{z})\). Distributivity over scalar addition is \((a+b)(\mathbf{v},\mathbf{w})=((a+b)\mathbf{v},(a+b)\mathbf{w})=(a\mathbf{v}+b\mathbf{v},a\mathbf{w}+b\mathbf{w})=a(\mathbf{v},\mathbf{w})+b(\mathbf{v},\mathbf{w})\). Finally, \(a(b(\mathbf{v},\mathbf{w}))=(a(b\mathbf{v}),a(b\mathbf{w}))=((ab)\mathbf{v},(ab)\mathbf{w})=(ab)(\mathbf{v},\mathbf{w})\), and \(1(\mathbf{v},\mathbf{w})=(\mathbf{v},\mathbf{w})\). These are all the real vector space axioms, so the product is a real vector space with the stated zero and additive inverse. \(\square\)
The proof works because every axiom can be checked one coordinate at a time. The two component spaces do not need to have the same elements, the same notation, or the same dimension; they only need to be real vector spaces so that the required operations and axioms are available in each one.
Computing with Different Kinds of Coordinates
Worked Example: A Vector Paired with a Matrix
Take \(V=\mathbb{R}^2\) and \(W=\mathbb{R}^{2\times 2}\), the real \(2\times2\) matrices. Let \(\mathbf{x}=((1,-2),\begin{bmatrix}1&2\\0&-1\end{bmatrix})\) and \(\mathbf{y}=((3,1),\begin{bmatrix}-2&0\\4&5\end{bmatrix})\). Find \(\mathbf{x}+\mathbf{y}\) and \(-2\mathbf{x}\).
Add the vector coordinates and matrix coordinates separately: \[ \mathbf{x}+\mathbf{y} =\left((1+3,-2+1), \begin{bmatrix}1+(-2)&2+0\\0+4&-1+5\end{bmatrix}\right) =\left((4,-1),\begin{bmatrix}-1&2\\4&4\end{bmatrix}\right). \] For the scalar multiple, multiply every entry of both components by \(-2\): \[ -2\mathbf{x} =\left((-2,-2(-2)), \begin{bmatrix}-2&-4\\0&2\end{bmatrix}\right) =\left((-2,4),\begin{bmatrix}-2&-4\\0&2\end{bmatrix}\right). \]
Check: Subtracting \(\mathbf{x}\) from the proposed sum gives vector coordinate \((4-1,-1-(-2))=(3,1)\) and matrix coordinate \(\begin{bmatrix}-1-1&2-2\\4-0&4-(-1)\end{bmatrix}=\begin{bmatrix}-2&0\\4&5\end{bmatrix}\), recovering \(\mathbf{y}\). Both coordinates of \(-2\mathbf{x}\) are the corresponding coordinates of \(\mathbf{x}\) multiplied by \(-2\).
The pair operation does not combine a vector with a matrix. It preserves their roles: the first coordinate is operated on using the vector-space operations in \(V\), and the second using those in \(W\). This is why the same definition applies even when the component spaces look entirely different.
Worked Example: A Function Paired with a Polynomial
Let \(V\) be the real-valued functions on \(\mathbb{R}\), and let \(W=P_2\), the real polynomials of degree at most \(2\). For \(f(t)=t^2-1\), \(g(t)=2t+3\), \(p(x)=x^2+2x\), and \(q(x)=-x^2+1\), compute \(2(f,p)-(g,q)\).
First apply scalar multiplication to both components, then subtract componentwise: \[ 2(f,p)-(g,q)=(2f-g,2p-q). \] For the function coordinate, \[ (2f-g)(t)=2(t^2-1)-(2t+3)=2t^2-2-2t-3=2t^2-2t-5. \] For the polynomial coordinate, \[ (2p-q)(x)=2(x^2+2x)-(-x^2+1) =2x^2+4x+x^2-1=3x^2+4x-1. \] Thus the result is \((2t^2-2t-5,\;3x^2+4x-1)\).
Check: Adding \((g,q)\) back to the result gives, in the function coordinate, \(2t^2-2t-5+2t+3=2t^2-2=2f(t)\). In the polynomial coordinate it gives \(3x^2+4x-1-x^2+1=2x^2+4x=2p(x)\). Hence adding \((g,q)\) recovers \(2(f,p)\), as required.
Linear Combinations in a Product
A linear combination of pairs also separates into two component calculations. This gives a useful way to organize computations: determine the first coordinate using the vectors in \(V\), and determine the second coordinate using the vectors in \(W\).
Proof. For \(p=1\), the statement is exactly the definition \(c_1(\mathbf{v}_1,\mathbf{w}_1)=(c_1\mathbf{v}_1,c_1\mathbf{w}_1)\). Suppose the formula holds for \(p\) terms. For \(p+1\) terms, use componentwise addition and the formula for the first \(p\) terms: \[ \sum_{i=1}^{p+1}c_i(\mathbf{v}_i,\mathbf{w}_i) =\left(\sum_{i=1}^{p}c_i\mathbf{v}_i,\sum_{i=1}^{p}c_i\mathbf{w}_i\right) +(c_{p+1}\mathbf{v}_{p+1},c_{p+1}\mathbf{w}_{p+1}). \] By the definition of addition, this equals \(\left(\sum_{i=1}^{p+1}c_i\mathbf{v}_i,\sum_{i=1}^{p+1}c_i\mathbf{w}_i\right)\). Induction proves the formula for every \(p\geq1\). \(\square\)
Worked Example: A Vector Paired with a Function
In \(\mathbb{R}^3\times F(\mathbb{R},\mathbb{R})\), take \(\mathbf{a}=((1,0,-1),f)\), where \(f(t)=t\), and \(\mathbf{b}=((2,1,3),g)\), where \(g(t)=t^2\). Compute \(3\mathbf{a}+2\mathbf{b}\).
Use the two coordinates separately: \[ 3\mathbf{a}+2\mathbf{b} =\bigl(3(1,0,-1)+2(2,1,3),\,3f+2g\bigr). \] The vector coordinate is \((3,0,-3)+(4,2,6)=(7,2,3)\). At each input \(t\), the function coordinate is \(3f(t)+2g(t)=3t+2t^2\). Therefore \(3\mathbf{a}+2\mathbf{b}=((7,2,3),\,t\mapsto 3t+2t^2)\).
Check: Subtracting \(2\mathbf{b}\) from this result gives vector \((7,2,3)-(4,2,6)=(3,0,-3)\) and function \((3t+2t^2)-2t^2=3t\). These are exactly the coordinates of \(3\mathbf{a}\).
Common Pitfall: Scaling Only One Coordinate
Scalar multiplication in a product must act on both coordinates. For example, a tempting but incorrect rule on \(\mathbb{R}\times\mathbb{R}\) is \(a(v,w)=(av,w)\), which leaves the second coordinate unchanged. With this rule, \(0(1,1)=(0,1)\), not the zero pair \((0,0)\). But the scalar identity \(0\mathbf{x}=\mathbf{0}\) is required in every real vector space; indeed, under the correct componentwise rule, \(0(1,1)=(0\cdot1,0\cdot1)=(0,0)\).
For a product of spaces with different kinds of elements, the same correction applies: multiply the vector and the matrix, or the function and the polynomial, by the scalar in their own spaces. The product-space theorem guarantees the vector space axioms only for the componentwise operations defined above.
How to Use the Product Construction
Product spaces provide a way to keep several pieces of linear information in one object without losing the structure of either piece. For example, a vector-matrix pair can be added and scaled while retaining one vector coordinate and one matrix coordinate. The linear-combination formula confirms that computations in the product can be checked coordinate by coordinate.
When working with a product, first identify what belongs in each coordinate and which vector-space operations apply there. Then use the componentwise definitions; do not mix entries across coordinates. The zero pair and additive inverse also follow coordinatewise, so they can be written down directly from the zero and inverse in each component space.
Check Your Understanding
Use the componentwise definitions and justify each result in both coordinates.
- In \(\mathbb{R}^2\times\mathbb{R}^{2\times2}\), find the zero pair and the additive inverse of \(((2,-1),\begin{bmatrix}0&3\\1&-2\end{bmatrix})\).
- For \((f,p)\in F(\mathbb{R},\mathbb{R})\times P_2\), write the componentwise expression for \(a(f,p)+(g,q)\).
- Compute \(2((1,2),t\mapsto t)-((3,0),t\mapsto t^2)\) in \(\mathbb{R}^2\times F(\mathbb{R},\mathbb{R})\).
- State which vector space axiom fails for the proposed scalar rule \(a(v,w)=(av,w)\) on \(\mathbb{R}\times\mathbb{R}\), and give a concrete failure.
- Explain why the product-space theorem does not require \(V\) and \(W\) to contain the same kind of elements.
