Why Basis and Dimension Arguments Go Wrong
Basis arguments often fail not because a calculation is difficult, but because a conclusion is drawn from only part of the definition. A list may span a space and still contain redundant vectors; it may be linearly independent and still fail to span. Likewise, counting vectors is meaningful only when the count is connected to the dimension of the particular space under discussion. These distinctions are especially important when a subspace sits inside a larger ambient space.
Let \(V\) be a vector space over \(\mathbb{R}\). A list \(\mathbf{v}_1,\dots,\mathbf{v}_p\) spans \(V\) if every vector in \(V\) can be written as a linear combination of the list. It is linearly independent if the equation \(c_1\mathbf{v}_1+\cdots+c_p\mathbf{v}_p=\mathbf{0}\) has only the solution \(c_1=\cdots=c_p=0\). A basis is a linearly independent spanning set. When \(V\) is finite-dimensional, a basis is finite and may be written as an ordered list; in general, a basis is a set and need not be finite. The dimension of a finite-dimensional space \(V\), written \(\dim V\), is the number of vectors in a basis, which is the same for every basis.
A list and a set are not interchangeable in every count. A list records order and repetitions; a set does not record either. For example, the list \((\mathbf{u},\mathbf{u})\) has two entries, but the set \(\{\mathbf{u},\mathbf{u}\}\) has only one element. Repetition also makes a nonzero list dependent, since \(1\mathbf{u}-1\mathbf{u}=\mathbf{0}\). When using dimension arguments, be explicit about whether you are counting entries in a list or elements in a set.
Redundant Generators Do Not Determine Dimension
Having many generators does not by itself mean a space has large dimension. A spanning list can include vectors that add nothing to the span. The relevant number for dimension is the size of a basis, not the length of an arbitrary spanning list.
Proof. Let the list be \(\mathbf{v}_1,\dots,\mathbf{v}_p\), and suppose \(\mathbf{v}_k=\sum_{j\ne k}a_j\mathbf{v}_j\). Every vector in the span of the list with \(\mathbf{v}_k\) is a combination of the remaining vectors: in any combination, replace its term \(c_k\mathbf{v}_k\) by \(c_k\sum_{j\ne k}a_j\mathbf{v}_j\), then collect coefficients. Thus the original span is contained in the span after deletion. The remaining vectors are also in the original list, so every combination of them is in the original span. This gives the reverse containment, and the spans are equal. \(\square\)
Worked Example: A Spanning List with a Redundant Vector
Consider the list \(\mathbf{v}_1=(1,0)\), \(\mathbf{v}_2=(0,1)\), \(\mathbf{v}_3=(1,1)\) in \(\mathbb{R}^2\). It is tempting to say that the space has dimension three because there are three generators. Check whether that count is justified.
First, \(\mathbf{v}_3=\mathbf{v}_1+\mathbf{v}_2\), so the third vector is redundant. The first two vectors span \(\mathbb{R}^2\): for any \((x,y)\in\mathbb{R}^2\),
The first two vectors are independent because \(a(1,0)+b(0,1)=(0,0)\) gives \((a,b)=(0,0)\). Thus \(\{\mathbf{v}_1,\mathbf{v}_2\}\) is a basis and \(\dim\mathbb{R}^2=2\), not three.
Check: The dependence in the original list is \(1\mathbf{v}_1+1\mathbf{v}_2-1\mathbf{v}_3=(1,0)+(0,1)-(1,1)=(0,0)\). It confirms that the three-entry list is not a basis.
The standard Spanning Set Reduction Theorem says that a finite spanning list in a finite-dimensional space can be reduced, by deleting redundant vectors, to a basis. The standard Basis Extension Theorem says that an independent list in a finite-dimensional space can be extended by adding vectors to a basis. These results were established earlier; their hypotheses matter here. Each applies to a finite list in a finite-dimensional space, and they explain why arbitrary generator counts are not dimension counts.
Proof. Suppose first that a spanning list has \(p\) entries. By the Spanning Set Reduction Theorem, some subset of its entries is a basis for \(V\). That basis has exactly \(n\) entries, so it cannot have more entries than the original list; hence \(n\le p\). Now suppose an independent list has \(q\) entries. By the Basis Extension Theorem, it can be extended to a basis for \(V\). The resulting basis has \(n\) entries and includes all \(q\) original entries, so \(q\le n\). Both claims also cover the zero-dimensional case: the empty list is a basis of the zero space, and the inequalities then allow zero entries. \(\square\)
Independence Is Not a Spanning Check
A common incomplete argument shows that a candidate list is independent and then calls it a basis. That conclusion omits a separate requirement: the list must span the space in question. The dimension count can expose the omission, but only if the ambient space and the candidate list have both been identified.
Worked Example: An Independent List That Does Not Span
Consider \(\mathbf{u}_1=(1,0,1)\) and \(\mathbf{u}_2=(0,1,1)\) in \(\mathbb{R}^3\). Check independence and then test whether the list spans \(\mathbb{R}^3\).
If \(a\mathbf{u}_1+b\mathbf{u}_2=\mathbf{0}\), then
The first coordinate gives \(a=0\), and the second gives \(b=0\), so the list is independent. To test spanning, ask whether it can produce \(\mathbf{e}_3=(0,0,1)\). The equation \(a\mathbf{u}_1+b\mathbf{u}_2=\mathbf{e}_3\) requires \(a=0\) from the first coordinate and \(b=0\) from the second. Those values give third coordinate \(a+b=0\), not \(1\). Thus \(\mathbf{e}_3\) is not in the span, and the list does not span \(\mathbb{R}^3\).
Check: The list has two independent vectors, while \(\dim\mathbb{R}^3=3\). The theorem on independent-list counts permits at most three entries, so the count alone does not prove failure to span. The explicit missing vector \(\mathbf{e}_3\) does.
This example illustrates why “independent” and “basis” are not synonyms. Independence rules out redundant vectors; spanning rules out missing vectors. A proof that verifies one property has not automatically verified the other. In particular, knowing that a list has at most \(\dim V\) entries does not establish that it spans \(V\).
Use the Dimension of the Space Being Discussed
If \(U\) is a subspace of \(V\), then \(U\) and \(V\) are different spaces and can have different dimensions. A vector in \(U\) may be represented using coordinates from the ambient space \(V\), but the number of ambient coordinates is not generally the dimension of \(U\). The relevant question is how many vectors are in a basis of \(U\).
Proof. Since \(V\) is finite-dimensional, choose a basis of \(U\) and apply the Basis Extension Theorem in \(V\) to extend that independent list to a basis of \(V\). The basis of \(U\) is a sublist of the extended basis. Therefore its number of vectors, \(\dim U\), is no greater than the number in the basis of \(V\), \(\dim V\). \(\square\)
Worked Example: A Plane in a Three-Dimensional Ambient Space
Let \(U=\{(x,y,z)\in\mathbb{R}^3:x+2y-z=0\}\). Determine its dimension without confusing it with the dimension of its ambient space \(\mathbb{R}^3\).
The defining equation gives \(z=x+2y\). Thus each vector in \(U\) has the form
So \(\mathbf{b}_1=(1,0,1)\) and \(\mathbf{b}_2=(0,1,2)\) span \(U\). They are independent: if \(a\mathbf{b}_1+b\mathbf{b}_2=(0,0,0)\), then \((a,b,a+2b)=(0,0,0)\), and the first two coordinates force \(a=0\) and \(b=0\). Hence they form a basis of \(U\), so \(\dim U=2\). The ambient space has dimension three, with standard basis \((1,0,0),(0,1,0),(0,0,1)\). The two dimension counts refer to different spaces.
Check: Each basis vector satisfies the defining equation: \(1+2(0)-1=0\) and \(0+2(1)-2=0\). The displayed representation shows that every vector satisfying the equation is in their span.
A Practical Audit for Dimension Claims
Before accepting a basis or dimension argument, check what the statement actually establishes. First identify the space: is the claim about an ambient space or a subspace? Next identify the object being counted: is it a basis, an independent list, or merely a spanning list? Finally check that the conclusion follows from the property proved. The dimension inequalities give two useful tests: a spanning list cannot have fewer entries than the dimension, and an independent list cannot have more. Neither inequality says that a list with exactly the dimension is automatically a basis.
A faulty argument might say, “The vectors lie in \(\mathbb{R}^3\), so this subspace has dimension three.” The ambient space has dimension three, but that does not show the subspace does. In the plane example, the equation reduces every vector to a combination of two basis vectors, so the subspace has dimension two. Another faulty argument might say, “This list is independent, therefore it is a basis.” The independence check establishes only one property; the example in \(\mathbb{R}^3\) shows that an independent list can miss vectors.
| What has been shown | What follows | What does not follow by itself |
|---|---|---|
| The list spans a finite-dimensional \(V\) | Its length is at least \(\dim V\) | The list is independent |
| The list is independent in finite-dimensional \(V\) | Its length is at most \(\dim V\) | The list spans \(V\) |
| A vector in a spanning list is a combination of the others | Deleting it preserves the span | The original list is a basis |
| \(U\) is a subspace of finite-dimensional \(V\) | \(\dim U\le\dim V\) | \(\dim U=\dim V\) |
Common Mistakes
Counting generators as dimension. The wrong calculation in the first example is “there are three generators, so the dimension is three.” The correction is to remove the redundant vector \(\mathbf{v}_3=\mathbf{v}_1+\mathbf{v}_2\), then use the two-vector basis to conclude that the dimension is two.
Checking only independence. The wrong conclusion in the second example is “the two vectors are independent, so they form a basis of \(\mathbb{R}^3\).” The correction is to test spanning as well: \(\mathbf{e}_3\) cannot be written as their linear combination, so they do not span \(\mathbb{R}^3\).
Using the ambient dimension for a subspace. The wrong calculation for \(U\) is “its vectors have three coordinates, so \(\dim U=3\).” The correction is to solve the equation defining \(U\), express its vectors using two independent generators, and conclude \(\dim U=2\). Coordinates locate vectors in the ambient space; they do not, by themselves, count a basis of the subspace.
Check Your Understanding
For each question, state which property or dimension count justifies your conclusion.
- In \(\mathbb{R}^2\), explain why the list \((1,0),(0,1),(2,1)\) cannot be a basis, and identify a dependence relation.
- Can an independent list of two vectors in \(\mathbb{R}^3\) be declared a basis from independence alone? State what additional check is needed.
- A finite spanning list in a space of dimension five has four entries. What can you conclude?
- For \(U=\{(x,y,z)\in\mathbb{R}^3:x-y+z=0\}\), express a general vector using two generators and determine whether those generators are independent.
- Explain why the number of coordinates used to represent vectors in an ambient space need not equal the dimension of a subspace.
