Finding a Basis of a Subspace: A Practical Guide for Students
When you encounter a subspace in linear algebra—whether it’s the set of all solutions to a homogeneous system, the column space of a matrix, or the null space of a transformation—you often need to determine a basis of a subspace. A basis provides the smallest set of linearly independent vectors that still span the entire subspace, giving you a clear picture of its dimension and structure. This article walks you through several reliable methods for finding a basis of a subspace, explains the underlying theory, and answers common questions to help you master the concept and apply it confidently in exams or real‑world problems.
Understanding Subspaces and Their Bases
A subspace of a vector space is a subset that is closed under vector addition and scalar multiplication. In practice, subspaces appear in three typical forms:
- Span of a set of vectors – the collection of all linear combinations of those vectors.
- Solution set of a homogeneous linear system – all vectors x that satisfy Ax = 0.
- Row space, column space, or null space of a matrix – subspaces derived directly from matrix operations.
Regardless of the origin, a basis of a subspace is a set of vectors that (a) lies entirely within the subspace, (b) is linearly independent, and (c) spans the subspace. The number of vectors in the basis equals the dimension of the subspace Took long enough..
Step‑by‑Step Methods for Finding a Basis
Below are four widely used approaches, each illustrated with a short example. Choose the method that matches how the subspace is presented in your problem Not complicated — just consistent..
1. From a Span of Vectors
When a subspace is described as the span of given vectors, the goal is to extract a linearly independent subset.
- Write the vectors as columns of a matrix.
- Perform Gaussian elimination to identify pivot columns.
- Select the original vectors that correspond to the pivot columns – these form a basis.
Example: Let (S = \text{span}{v_1, v_2, v_3}) where
(v_1 = \begin{bmatrix}1\2\3\end{bmatrix}, v_2 = \begin{bmatrix}2\4\6\end{bmatrix}, v_3 = \begin{bmatrix}0\1\-1\end{bmatrix}).
Form (A = [v_1;v_2;v_3]). Row‑reducing (A) shows that column 1 and column 3 are pivots, while column 2 is a linear combination of column 1. Hence ({v_1, v_3}) is a basis for (S) And that's really what it comes down to..
Key point: The pivot columns in the reduced row‑echelon form (RREF) guarantee linear independence, while the non‑pivot columns reveal redundancy Most people skip this — try not to..
2. From the Solution Set of a Homogeneous System
If a subspace is defined by equations Ax = 0, the basis is built from the free variables after row reduction.
- Write the augmented matrix ([A | \mathbf{0}]) and reduce to RREF.
- Identify pivot variables (corresponding to leading 1’s) and free variables.
- Express each pivot variable in terms of the free variables and assign parameters.
- For each free variable, set it to 1 and the others to 0 to obtain a basis vector.
Example: Solve
[ \begin{cases} x_1 + 2x_2 - x_3 = 0\ 2x_1 + 4x_2 - 2x_3 = 0 \end{cases} ]
The RREF yields one pivot in column 1 and two free variables (x_2, x_3). Setting ((x_2, x_3) = (1,0)) gives (\mathbf{b}_1 = \begin{bmatrix}-2\1\0\end{bmatrix}); setting ((x_2, x_3) = (0,1)) gives (\mathbf{b}2 = \begin{bmatrix}1\0\1\end{bmatrix}). Thus ({\mathbf{b}'hôtel, \mathbf{b}_2}) is a basis for the solution subspace.
Not obvious, but once you see it — you'll see it everywhere.
Why it works: Each free variable generates a independent direction in the null space, and the resulting vectors automatically satisfy the original equations Simple as that..
3. Using Row Reduction on a Matrix (Column Space)
The column space of a matrix (M) is a subspace of (\mathbb{R}^m). Its basis consists of the pivot columns of the original matrix.
- Form the matrix (M) and compute its RREF.
- Mark the columns that contain leading 1’s in the RREF.
- Copy those same columns from the original matrix (M) – they form a basis for the column space.
Example: For
[ M = \begin{bmatrix} 1 & 2 & 3\ 2 & 4 & 6\ 1 & 0 & -1 \end{bmatrix}, ]
the RREF shows pivots in columns 1 and 3. Hence ({,\begin{bmatrix}1\2\1\end{bmatrix},\ \begin{bmatrix}3\6\-1\end{bmatrix},}) is a basis for (\text{Col}(M)) That's the whole idea..
Note: The column space’s dimension equals the number of pivots, also called the rank of the matrix Small thing, real impact. Nothing fancy..
4. Determining the Null Space Basis Directly
The null space (kernel) of a matrix (A) is the set of all vectors x such that (A\mathbf{x}= \mathbf{0}). The procedure is identical to Method 2 but emphasizes the null space terminology.
- Row‑reduce (A) to RREF.
- Identify free variables (non‑pivot columns).
- Create basis vectors by assigning each free variable a value of 1 while setting the others to 0, then solving for the pivot variables.
The resulting vectors constitute a basis for the null space and have dimension equal to the number of free variables (the nullity).
Scientific Explanation: Why These Methods Work
The reliability of each technique rests on fundamental theorems in linear algebra:
- Rank–Nullity Theorem: For an (m \times n) matrix (A),
